h3-blackwell-runtime/benchmarks/gb10-post-fc2-warmed-step-stats_cuda_gpu_trace.csv

1	Start (ns)	Duration (ns)	CorrId	GrdX	GrdY	GrdZ	BlkX	BlkY	BlkZ	Reg/Trd	StcSMem (MB)	DymSMem (MB)	Bytes (MB)	Throughput (MB/s)	SrcMemKd	DstMemKd	Device	Ctx	GreenCtx	Strm	Name
2	6687024	129408	4655										14.322	110663.498	Device	Device	NVIDIA GB10 (0)	1		7	[CUDA memcpy Device-to-Device]
3	6818800	4288	4667										0.053	12358.158	Device	Device	NVIDIA GB10 (0)	1		7	[CUDA memcpy Device-to-Device]
4	7204912	1888	4680										0.000	4.237	Pageable	Device	NVIDIA GB10 (0)	1		7	[CUDA memcpy Host-to-Device]
5	7275376	2144	4697	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
6	7292688	2816	4708	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3)
7	7305872	3392	4719	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::DivFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
8	7313232	1088	4730	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
9	7320496	1088	4741	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
10	7326704	1056	4752	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3)
11	7332624	2080	4763	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::DivFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
12	7339504	1824	4774	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::DivFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
13	7357872	2752	4788	13	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
14	7370736	5728	4800	26	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl<at::native::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
15	7381104	1120	4811	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
16	7392976	1984	4822	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
17	7417616	38432	4839	3497	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
18	7459632	48832	4854	6993	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 12)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
19	7510000	71520	4869	6993	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
20	7647600	4915936	4896	2336	3	1	256	1	1	212	0.000	0.049					NVIDIA GB10 (0)	1		7	void cutlass::Kernel2<cutlass_80_simt_sgemm_256x128_8x4_tn_align1>(T1::Params)
21	12565488	5035136	4909	195804	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
22	17601648	3808	4927	26	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
23	17607760	44096	4953	56	6	1	128	1	1	130	0.000	0.031					NVIDIA GB10 (0)	1		7	void cutlass::Kernel2<cutlass_80_simt_sgemm_128x64_8x5_tt_align1>(T1::Params)
24	17653744	30368	4966	2174	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
25	17685488	3744	4978	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
26	17691728	1792	4989	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
27	17696080	1344	5000	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3)
28	17700080	2048	5011	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::DivFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
29	17704272	1536	5022	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
30	17708368	1280	5033	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
31	17712368	1376	5044	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3)
32	17716560	2080	5055	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::DivFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
33	17720400	2848	5066	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnOther_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3)
34	17729712	6688	5072										0.000	0.598	Device	Pinned	NVIDIA GB10 (0)	1		7	[CUDA memcpy Device-to-Host]
35	17749968	1024	5083	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnOther_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3)
36	17754544	608	5089										0.000	6.579	Device	Pinned	NVIDIA GB10 (0)	1		7	[CUDA memcpy Device-to-Host]
37	17778576	960	5102										0.000	4.167	Pageable	Device	NVIDIA GB10 (0)	1		7	[CUDA memcpy Host-to-Device]
38	17807664	3766816	5114	1536	3	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::CatArrayBatchedCopy_alignedK_contig<at::native::<unnamed>::OpaqueType<(unsigned int)2>, unsigned int, (int)2, (int)128, (int)1, (int)8>(T1 *, at::native::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::native::<unnamed>::TensorSizeStride<T2, (unsigned int)4>, int, T2)
39	26302704	20288	5127										0.907	44727.718	Pageable	Device	NVIDIA GB10 (0)	1		7	[CUDA memcpy Host-to-Device]
40	26372208	8192	5141	222	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
41	26413968	27680	5153	7090	1	1	128	1	1	21	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
42	26449776	52384	5165	96	3	1	512	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::CatArrayBatchedCopy<at::native::<unnamed>::OpaqueType<(unsigned int)4>, unsigned int, (int)2, (int)64, (int)64>(T1 *, at::native::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::native::<unnamed>::TensorSizeStride<T2, (unsigned int)4>, int, T2)
43	26503152	31872	5176	1773	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::cos_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
44	26536016	39456	5187	1773	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::sin_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
45	26577904	39712	5198	1773	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
46	26618864	873696	5210	1536	4	1	128	1	1	37	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::CatArrayBatchedCopy_alignedK_contig<at::native::<unnamed>::OpaqueType<(unsigned int)4>, unsigned int, (int)3, (int)128, (int)1, (int)16>(T1 *, at::native::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::native::<unnamed>::TensorSizeStride<T2, (unsigned int)4>, int, T2)
47	27495408	217120	5224	7090	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
48	27714544	2240	5236	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
49	27718736	1536	5247	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
50	27722736	3424	5258	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
51	27728848	3168	5272	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
52	27733072	3136	5284	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
53	27737488	4736	5296	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
54	27743504	1088	5309	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
55	27747568	1632	5321	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
56	27751536	1504	5334	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
57	27755760	2016	5345	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
58	27759856	30048	5366	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
59	27792368	3544672	5389	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
60	31338480	2464	5404	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
61	31342576	1984	5422	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
62	31346672	49568	5462	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
63	31398896	3452992	5470	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
64	34853840	2912	5473	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
65	34857968	2466752	5476	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
66	37326064	1920	5487	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
67	37330160	24059776	5516	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
68	61391216	11842304	5533	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
69	73250640	448	5547										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
70	73252304	2395136	5550	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
71	75649264	4001984	5583	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
72	79653872	3999008	5585	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
73	83654864	5297792	5601	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
74	88955088	5731328	5624	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
75	94689520	240315488	5627	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
76	335006992	2176224	5634	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
77	337185008	1952	5637	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
78	337189104	1376	5651	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
79	337193200	1504	5666	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
80	337197296	40608	5689	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
81	337240272	2775488	5701	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
82	340017392	1888	5710	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
83	340021488	7702240	5739	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
84	347725040	5461056	5753	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
85	353188080	3717632	5773	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
86	356907248	5856	5788	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
87	356915440	3776	5806	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
88	356921584	60928	5846	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
89	356984048	4091840	5854	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
90	361077968	1952	5857	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
91	361082096	2437024	5860	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
92	363520432	1920	5871	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
93	363524336	34563520	5900	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
94	398089488	140224	5941	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
95	398231792	11923520	5949	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
96	410157296	2464	5952	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
97	410161392	10756480	5955	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
98	420920560	1920	5966	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
99	420924656	3040	5981	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
100	420929776	16202336	6002	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
101	437133584	5455232	6015	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
102	442590448	1984	6023	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
103	442594544	1440	6034	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
104	442598640	1376	6045	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
105	442602736	3168	6059	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
106	442608848	1344	6071	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
107	442613040	4608	6083	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
108	442618992	1120	6096	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
109	442623216	1632	6108	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
110	442627312	1472	6121	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
111	442631408	1248	6132	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
112	442635504	29472	6153	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
113	442666256	3479584	6176	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
114	446148848	2368	6191	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
115	446152944	2112	6209	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
116	446157040	48416	6249	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
117	446208240	3449888	6257	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
118	449660144	4032	6260	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
119	449666288	2435424	6263	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
120	452104432	1888	6274	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
121	452108528	25770432	6303	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
122	477881584	11819968	6320	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
123	489703312	416	6334										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
124	489704624	2346688	6337	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
125	492053744	3809824	6370	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
126	495865072	3947776	6372	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
127	499815664	5406624	6388	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
128	505224432	5664096	6411	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
129	510891248	240444576	6414	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
130	751338768	2172448	6421	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
131	753512656	1952	6424	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
132	753516784	2912	6438	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
133	753520976	1536	6453	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
134	753525008	46304	6476	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
135	753573072	2749568	6488	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
136	756324592	1920	6497	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
137	756328656	7743936	6526	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
138	764074288	5461696	6540	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
139	769538256	3462944	6560	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
140	773002480	2304	6575	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
141	773006576	1984	6593	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
142	773010672	46624	6633	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
143	773058800	3966080	6641	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
144	777027824	1952	6644	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
145	777031920	2729952	6647	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
146	779763952	1856	6658	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
147	779768048	33135360	6687	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
148	812905712	186080	6728	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
149	813094096	12695904	6736	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
150	825791728	2176	6739	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
151	825795920	10475328	6742	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
152	836272528	2176	6753	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
153	836276464	832	6768	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
154	836278352	15614944	6789	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
155	851894576	5572000	6802	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
156	857469168	2336	6810	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
157	857473264	3104	6821	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
158	857479408	3264	6832	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
159	857485552	11680	6846	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
160	857499856	6144	6858	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
161	857509360	5600	6870	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
162	857516272	2368	6883	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
163	857520368	3360	6895	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
164	857526512	2944	6908	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
165	857530736	8192	6919	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
166	857540848	56608	6940	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
167	857600240	3673856	6963	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
168	861275376	2464	6978	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
169	861279440	2048	6996	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
170	861283536	48960	7036	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
171	861334736	3813056	7044	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
172	865150160	3776	7047	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
173	865156304	2434944	7050	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
174	867593456	1888	7061	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
175	867597584	23937632	7090	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
176	891537648	12480864	7107	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
177	904019856	416	7121										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
178	904021456	2345600	7124	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
179	906369264	3709440	7157	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
180	910080208	3724512	7159	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
181	913807568	5534656	7175	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
182	919343504	5795488	7198	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
183	925141232	239349632	7201	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
184	1164493200	2170784	7208	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
185	1166665936	1920	7211	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
186	1166670064	2912	7225	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
187	1166674288	1504	7240	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
188	1166678288	47104	7263	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
189	1166728400	2757664	7275	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
190	1169489136	1952	7284	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
191	1169493232	7799744	7313	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
192	1177295120	5451936	7327	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
193	1182748880	3475680	7347	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
194	1186227440	2400	7362	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
195	1186231504	2080	7380	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
196	1186235600	46016	7420	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
197	1186282896	3462240	7428	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
198	1189746928	1984	7431	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
199	1189750992	2452064	7434	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
200	1192204528	1888	7445	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
201	1192208624	34645696	7474	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
202	1226856784	139968	7515	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
203	1226999024	12761440	7523	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
204	1239763184	1920	7526	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
205	1239767280	10456096	7529	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
206	1250225392	1952	7540	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
207	1250229456	864	7555	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
208	1250231632	15792128	7576	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
209	1266026768	5824160	7589	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
210	1271852272	1952	7597	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
211	1271856368	1408	7608	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
212	1271860432	2848	7619	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
213	1271864560	3200	7633	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
214	1271870704	1376	7645	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
215	1271874800	4416	7657	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
216	1271880912	1120	7670	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
217	1271885040	1600	7682	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
218	1271889136	1472	7695	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
219	1271893200	1280	7706	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
220	1271897392	26016	7727	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
221	1271925968	3545984	7750	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
222	1275473232	2496	7765	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
223	1275477232	2112	7783	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
224	1275481328	47552	7823	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
225	1275530768	4259520	7831	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
226	1279793424	3712	7834	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
227	1279799536	2444704	7837	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
228	1282245872	1824	7848	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
229	1282249936	24882560	7877	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
230	1307135248	12475168	7894	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
231	1319611344	416	7908										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
232	1319612656	2339296	7911	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
233	1321954544	3722080	7944	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
234	1325678832	3736032	7946	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
235	1329416432	5150368	7962	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
236	1334569200	5873152	7985	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
237	1340443888	238904576	7988	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
238	1579351312	2158912	7995	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
239	1581512912	1984	7998	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
240	1581517040	2880	8012	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
241	1581521200	1536	8027	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
242	1581525136	54208	8050	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
243	1581580624	2747456	8062	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
244	1584329936	1920	8071	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
245	1584334064	7804864	8100	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
246	1592141072	5612512	8114	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
247	1597755632	3661824	8134	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
248	1601419504	2400	8149	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
249	1601423568	2112	8167	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
250	1601427696	46240	8207	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
251	1601475824	3859104	8215	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
252	1605337328	1952	8218	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
253	1605341456	2438848	8221	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
254	1607782640	1920	8232	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
255	1607786736	34043744	8261	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
256	1641831824	139840	8302	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
257	1641972944	12366720	8310	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
258	1654341872	1920	8313	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
259	1654345968	10667840	8316	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
260	1665016048	1984	8327	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
261	1665020144	2560	8342	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
262	1665024240	15704512	8363	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
263	1680730352	5454720	8376	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
264	1686186480	1952	8384	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
265	1686189744	1440	8395	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
266	1686193392	1312	8406	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
267	1686197488	3264	8420	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
268	1686203600	1344	8432	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
269	1686207728	4640	8444	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
270	1686213872	1088	8457	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
271	1686217968	1632	8469	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
272	1686222160	1504	8482	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
273	1686226160	1280	8493	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
274	1686230256	27264	8514	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
275	1686258928	3493056	8537	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
276	1689753872	2464	8552	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
277	1689757936	2016	8570	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
278	1689762032	47456	8610	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
279	1689811152	3918208	8618	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
280	1693731056	3392	8621	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
281	1693737200	2845728	8624	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
282	1696585968	1856	8635	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
283	1696590096	25301536	8664	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
284	1721894096	11901664	8681	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
285	1733796816	384	8695										0.000	583.333	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
286	1733798384	2551328	8698	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
287	1736351984	3738560	8731	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
288	1740092624	3951712	8733	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
289	1744046320	5119744	8749	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
290	1749167344	5697824	8772	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
291	1754867920	240193088	8775	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
292	1995062544	2167328	8782	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
293	1997232368	1920	8785	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
294	1997236432	2944	8799	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
295	1997240656	1504	8814	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
296	1997244624	47296	8837	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
297	1997293808	2750016	8849	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
298	2000046288	1952	8858	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
299	2000050416	7806304	8887	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
300	2007859472	5472288	8901	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
301	2013334768	3708736	8921	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
302	2017045712	6624	8936	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
303	2017053936	2144	8954	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
304	2017058000	47936	8994	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
305	2017107216	4063680	9002	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
306	2021173488	1952	9005	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
307	2021177584	2438432	9008	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
308	2023617776	1920	9019	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
309	2023621872	34049568	9048	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
310	2057673968	139872	9089	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
311	2057816272	11888096	9097	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
312	2069705936	1984	9100	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
313	2069710064	10741088	9103	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
314	2080453872	1952	9114	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
315	2080457968	2496	9129	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
316	2080462064	16236448	9150	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
317	2096700624	5467584	9163	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
318	2102170832	1952	9171	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
319	2102174960	1440	9182	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
320	2102179056	1312	9193	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
321	2102181904	3456	9207	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
322	2102187248	1344	9219	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
323	2102191280	4608	9231	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
324	2102197488	1088	9244	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
325	2102201520	1632	9256	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
326	2102205680	1728	9269	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
327	2102209776	1280	9280	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
328	2102213840	26368	9301	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
329	2102242544	3495232	9324	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
330	2105740528	2496	9339	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
331	2105744624	1984	9357	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
332	2105748720	47232	9397	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
333	2105797840	3515456	9405	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
334	2109316304	3360	9408	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
335	2109322480	2434016	9411	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
336	2111758576	1888	9422	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
337	2111762672	25849184	9451	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
338	2137614608	11818912	9468	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
339	2149435280	416	9482										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
340	2149436720	2354816	9485	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
341	2151792912	3750336	9518	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
342	2155544784	4064384	9520	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
343	2159612112	5416480	9536	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
344	2165031152	5667744	9559	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
345	2170700176	242080160	9562	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
346	2412782864	2159488	9569	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
347	2414944496	1952	9572	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
348	2414948592	2944	9586	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
349	2414952560	1504	9601	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
350	2414955312	49536	9624	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
351	2415006928	2752224	9636	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
352	2417760496	1888	9645	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
353	2417764560	7814080	9674	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
354	2425580816	5456192	9688	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
355	2431038672	3483520	9708	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
356	2434524624	6336	9723	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
357	2434532592	36384	9741	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
358	2434571504	54208	9781	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
359	2434628848	4283872	9789	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
360	2438914256	1952	9792	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
361	2438918384	2440160	9795	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
362	2441360624	1888	9806	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
363	2441364720	33763776	9835	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
364	2475131248	139488	9876	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
365	2475272432	12292416	9884	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
366	2487567568	2176	9887	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
367	2487571728	10786304	9890	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
368	2498360560	1920	9901	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
369	2498364656	832	9916	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
370	2498366768	15820672	9937	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
371	2514190576	5490432	9950	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
372	2519683344	2272	9958	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
373	2519687440	1408	9969	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
374	2519691504	3712	9980	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
375	2519697648	8064	9994	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
376	2519707856	1568	10006	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
377	2519711920	8096	10018	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
378	2519722224	6880	10031	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
379	2519730384	1632	10043	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
380	2519734512	14048	10056	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
381	2519750896	3648	10067	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
382	2519757040	93376	10088	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
383	2519853264	3592384	10111	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
384	2523448560	2912	10126	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
385	2523452720	2240	10144	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
386	2523456752	47712	10184	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
387	2523505904	3622240	10192	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
388	2527129808	4000	10195	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
389	2527135984	2437120	10198	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
390	2529576208	1856	10209	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
391	2529580272	25405312	10238	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
392	2554986864	12071296	10255	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
393	2567060336	448	10269										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
394	2567062000	2347552	10272	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
395	2569410832	3711616	10305	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
396	2573123824	4098912	10307	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
397	2577223984	5356480	10323	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
398	2582582512	5691456	10346	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
399	2588277008	239531264	10349	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
400	2827810064	2158528	10356	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
401	2829970640	1984	10359	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
402	2829974768	2720	10373	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
403	2829978864	1504	10388	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
404	2829982928	46080	10411	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
405	2830032080	2749120	10423	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
406	2832783600	1920	10432	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
407	2832787696	7819616	10461	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
408	2840610032	5449728	10475	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
409	2846062832	3479168	10495	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
410	2849543408	2336	10510	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
411	2849547504	1984	10528	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
412	2849551600	46464	10568	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
413	2849600720	3413024	10576	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
414	2853015792	1984	10579	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
415	2853019888	2757152	10582	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
416	2855779536	1856	10593	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
417	2855783632	33791744	10622	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
418	2889576720	141536	10663	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
419	2889721072	12811136	10671	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
420	2902534352	2208	10674	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
421	2902538480	10362432	10677	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
422	2912903408	1952	10688	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
423	2912907472	864	10703	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
424	2912909616	15744352	10724	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
425	2928656624	5465664	10737	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
426	2934124752	1984	10745	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
427	2934128912	1408	10756	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
428	2934132976	3680	10767	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
429	2934139120	3488	10781	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
430	2934145136	1344	10793	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
431	2934147728	4608	10805	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
432	2934154352	1120	10818	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
433	2934158576	1632	10830	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
434	2934162672	1472	10843	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
435	2934166768	1280	10854	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
436	2934170864	25920	10875	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
437	2934199536	3495488	10898	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
438	2937696496	2528	10913	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
439	2937700560	1920	10931	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
440	2937704656	48864	10971	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
441	2937754864	3453632	10979	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
442	2941210832	3808	10982	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
443	2941217008	2435392	10985	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
444	2943654160	1888	10996	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
445	2943658224	24871360	11025	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
446	2968532176	12473728	11042	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
447	2981008304	416	11056										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
448	2981009616	2342880	11059	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
449	2983354608	3711520	11092	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
450	2987067632	3730304	11094	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
451	2990800112	5451680	11110	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
452	2996253392	5726816	11133	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
453	3001983216	240209920	11136	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
454	3242196240	2180576	11143	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
455	3244379376	1984	11146	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
456	3244383504	2752	11160	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
457	3244387536	1504	11175	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
458	3244391664	45632	11198	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
459	3244438736	2899360	11210	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
460	3247340784	2016	11219	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
461	3247344848	8260608	11248	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
462	3255608560	5606272	11262	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
463	3261217040	3532864	11282	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
464	3264751824	2528	11297	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
465	3264755920	1984	11315	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
466	3264760048	46304	11355	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
467	3264808144	3517760	11363	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
468	3268327664	2144	11366	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
469	3268331760	2417696	11369	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
470	3270751472	1888	11380	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
471	3270755568	34255616	11409	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
472	3305013552	141536	11450	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
473	3305157872	12819968	11458	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
474	3317979344	2464	11461	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
475	3317983472	10346848	11464	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
476	3328333040	2144	11475	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
477	3328337136	864	11490	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
478	3328339280	15735488	11511	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
479	3344077040	5454016	11524	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
480	3349532912	2016	11532	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
481	3349537008	1408	11543	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
482	3349541104	1312	11554	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
483	3349545200	3456	11568	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
484	3349551344	1312	11580	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
485	3349555568	4224	11592	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
486	3349561520	1088	11605	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
487	3349565680	1632	11617	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
488	3349569744	1504	11630	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
489	3349573872	1280	11641	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
490	3349577936	26720	11662	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
491	3349606640	3496640	11685	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
492	3353104752	2400	11700	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
493	3353108720	2112	11718	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
494	3353112816	47680	11758	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
495	3353162960	3506560	11766	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
496	3356672208	3680	11769	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
497	3356678352	2438752	11772	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
498	3359119600	1888	11783	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
499	3359123664	24909184	11812	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
500	3384035568	12198048	11829	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
501	3396235152	416	11843										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
502	3396236784	2509216	11846	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
503	3398747472	3777248	11879	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
504	3402526000	3770368	11881	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
505	3406299344	5082752	11897	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
506	3411384560	5710464	11920	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
507	3417097904	240101408	11923	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
508	3657200912	2166688	11930	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
509	3659370736	1920	11933	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
510	3659374832	2528	11947	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
511	3659378896	1536	11962	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
512	3659382896	45856	11985	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
513	3659431120	2762624	11997	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
514	3662195920	1920	12006	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
515	3662200016	8099264	12035	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
516	3670300976	5523968	12049	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
517	3675826448	3822336	12069	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
518	3679650064	2528	12084	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
519	3679654128	2080	12102	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
520	3679658224	46624	12142	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
521	3679707344	3806144	12150	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
522	3683514768	1920	12153	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
523	3683518704	2444096	12156	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
524	3685965040	1920	12167	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
525	3685969136	34983616	12196	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
526	3720954160	139744	12237	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
527	3721095408	12296512	12245	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
528	3733394640	9696	12248	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
529	3733406960	10774656	12251	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
530	3744184560	7232	12262	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
531	3744194928	3328	12277	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
532	3744200976	15784672	12298	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
533	3759986992	5478976	12311	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
534	3765468400	1952	12319	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
535	3765472496	1440	12330	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
536	3765476592	1280	12341	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
537	3765480656	3456	12355	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
538	3765486800	1312	12367	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
539	3765490800	4448	12379	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
540	3765497040	1312	12392	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
541	3765499728	1632	12404	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
542	3765503216	1472	12417	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
543	3765507312	1280	12428	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
544	3765511408	24768	12449	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
545	3765538032	3497504	12472	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
546	3769037040	2464	12487	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
547	3769041136	2016	12505	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
548	3769045232	48544	12545	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
549	3769095376	3424864	12553	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
550	3772522768	3552	12556	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
551	3772528912	2510688	12559	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
552	3775041776	1888	12570	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
553	3775045872	24845056	12599	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
554	3799893232	11830912	12616	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
555	3811725200	416	12630										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
556	3811726480	2388928	12633	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
557	3814117616	3942304	12666	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
558	3818062032	3950304	12668	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
559	3822014736	5234368	12684	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
560	3827250416	5660672	12707	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
561	3832913136	240325056	12710	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
562	4073240848	2157632	12717	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
563	4075400400	1952	12720	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
564	4075404496	2560	12734	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
565	4075408624	1504	12749	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
566	4075412784	50400	12772	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
567	4075465936	2760608	12784	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
568	4078227856	1952	12793	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
569	4078231792	7826784	12822	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
570	4086061328	5463200	12836	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
571	4091526352	3570432	12856	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
572	4095099184	2336	12871	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
573	4095103184	2208	12889	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
574	4095107312	48384	12929	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
575	4095157488	4305856	12937	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
576	4099465456	1920	12940	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
577	4099469584	2451072	12943	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
578	4101923088	1856	12954	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
579	4101927120	33911744	12983	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
580	4135841008	206816	13024	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
581	4136050928	12091680	13032	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
582	4148144336	2208	13035	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
583	4148148464	10751008	13038	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
584	4158902512	1920	13049	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
585	4158906608	3328	13064	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
586	4158912752	15259232	13085	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
587	4174173648	5466400	13098	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
588	4179642672	1920	13106	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
589	4179646704	1440	13117	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
590	4179650768	1312	13128	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
591	4179654896	3072	13142	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
592	4179661040	1312	13154	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
593	4179665040	4800	13166	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
594	4179671248	1120	13179	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
595	4179675312	1632	13191	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
596	4179679472	1568	13204	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
597	4179683536	1248	13215	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
598	4179687696	28672	13236	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
599	4179718384	3498816	13259	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
600	4183218480	2432	13274	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
601	4183222512	2112	13292	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
602	4183226576	48096	13332	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
603	4183276752	3518400	13340	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
604	4186796464	3456	13343	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
605	4186802448	2435904	13346	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
606	4189240560	1888	13357	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
607	4189244656	26090176	13386	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
608	4215337200	12096896	13403	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
609	4227435376	416	13417										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
610	4227437008	2347104	13420	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
611	4229786864	3730976	13453	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
612	4233519312	4067328	13455	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
613	4237588688	5428096	13471	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
614	4243019024	5745408	13494	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
615	4248766704	242399936	13497	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
616	4491168048	2168800	13504	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
617	4493338896	1920	13507	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
618	4493342960	2880	13521	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
619	4493347184	1504	13536	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
620	4493351184	46592	13559	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
621	4493399248	2748320	13571	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
622	4496149744	1952	13580	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
623	4496153872	7929856	13609	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
624	4504086576	5458016	13623	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
625	4509546736	3492480	13643	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
626	4513041744	2400	13658	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
627	4513045712	1984	13676	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
628	4513049840	47232	13716	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
629	4513100016	4172256	13724	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
630	4517273808	1984	13727	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
631	4517277936	2537024	13730	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
632	4519816432	1888	13741	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
633	4519820560	33533088	13770	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
634	4553355536	157600	13811	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
635	4553515248	12645408	13819	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
636	4566162640	2432	13822	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
637	4566166768	10464160	13825	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
638	4576632208	2400	13836	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
639	4576636144	2208	13851	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
640	4576640240	15692288	13872	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
641	4592334064	5631424	13885	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
642	4597968112	2560	13893	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
643	4597972208	1440	13904	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
644	4597976272	1472	13915	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
645	4597980368	3680	13929	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
646	4597986480	3136	13941	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
647	4597992688	6432	13953	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
648	4598000880	1280	13966	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
649	4598004976	7072	13978	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
650	4598013424	11168	13991	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
651	4598027504	1248	14002	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
652	4598031600	71488	14023	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
653	4598105296	3607424	14046	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
654	4601714000	2464	14061	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
655	4601718000	2080	14079	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
656	4601722096	48672	14119	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
657	4601772304	3869632	14127	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
658	4605645008	4032	14130	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
659	4605651152	2434688	14133	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
660	4608087280	1856	14144	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
661	4608091408	25006880	14173	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
662	4633100528	12395104	14190	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
663	4645497744	416	14204										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
664	4645499056	2352544	14207	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
665	4647854320	3720192	14240	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
666	4651577584	3872096	14242	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
667	4655452496	5580512	14258	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
668	4661034288	5712640	14281	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
669	4666749168	241404896	14284	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
670	4908156176	2166912	14291	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
671	4910324976	2016	14294	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
672	4910329072	2912	14308	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
673	4910333264	1536	14323	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
674	4910337264	48064	14346	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
675	4910388432	2752704	14358	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
676	4913142992	1920	14367	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
677	4913147120	7930720	14396	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
678	4921079152	5452960	14410	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
679	4926534896	3469120	14430	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
680	4930006352	2464	14445	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
681	4930010320	2016	14463	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
682	4930014448	46528	14503	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
683	4930063600	3409568	14511	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
684	4933475536	1952	14514	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
685	4933479664	2424224	14517	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
686	4935905520	1888	14528	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
687	4935909648	33606624	14557	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
688	4969519376	139712	14598	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
689	4969661680	12773024	14606	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
690	4982436080	2176	14609	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
691	4982440176	10351520	14612	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
692	4992792976	1920	14623	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
693	4992796912	832	14638	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
694	4992799056	15773088	14659	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
695	5008574736	5499680	14672	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
696	5014076656	2496	14680	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
697	5014080752	1440	14691	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
698	5014084848	3712	14702	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
699	5014090992	3584	14716	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
700	5014097008	1344	14728	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
701	5014101392	4544	14740	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
702	5014107408	1088	14753	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
703	5014111472	1600	14765	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
704	5014115536	1920	14778	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
705	5014119664	1248	14789	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
706	5014123728	26144	14810	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
707	5014152496	3804640	14833	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
708	5017958640	22080	14848	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
709	5017983184	32192	14866	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
710	5018018032	47840	14906	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
711	5018068176	3836736	14914	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
712	5021906256	3744	14917	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
713	5021912336	2440928	14920	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
714	5024354576	1856	14931	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
715	5024358640	24941184	14960	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
716	5049301200	12450592	14977	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
717	5061753712	448	14991										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
718	5061755344	2347648	14994	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
719	5064105200	3717760	15027	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
720	5067824464	3725984	15029	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
721	5071551696	5502592	15045	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
722	5077055728	5676000	15068	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
723	5082734864	240707424	15071	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
724	5323444496	2169536	15078	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
725	5325615312	1952	15081	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
726	5325619440	2880	15095	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
727	5325623600	1536	15110	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
728	5325627632	47104	15133	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
729	5325676752	2767680	15145	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
730	5328446704	1920	15154	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
731	5328450800	8481312	15183	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
732	5336933648	5526464	15197	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
733	5342462192	3485760	15217	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
734	5345949936	2400	15232	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
735	5345954032	1984	15250	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
736	5345958128	46688	15290	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
737	5346006224	3422112	15298	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
738	5349430640	1952	15301	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
739	5349434576	2419200	15304	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
740	5351855344	1888	15315	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
741	5351859408	34284960	15344	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
742	5386147088	139712	15385	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
743	5386289392	12727072	15393	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
744	5399018704	1984	15396	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
745	5399022864	10358368	15399	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
746	5409383696	1920	15410	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
747	5409387760	832	15425	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
748	5409389872	15717632	15446	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
749	5425110256	5460384	15459	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
750	5430572240	1952	15467	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
751	5430576368	1440	15478	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
752	5430580464	1312	15489	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
753	5430584560	3392	15503	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
754	5430590672	1376	15515	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
755	5430594800	4256	15527	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
756	5430600976	1088	15540	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
757	5430605040	1600	15552	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
758	5430609104	1504	15565	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
759	5430613328	1248	15576	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
760	5430617296	25312	15597	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
761	5430643952	3500224	15620	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
762	5434146064	2496	15635	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
763	5434150192	2112	15653	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
764	5434154224	49952	15693	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
765	5434206448	4299072	15701	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
766	5438508272	3648	15704	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
767	5438514416	2441600	15707	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
768	5440958704	1888	15718	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
769	5440962800	24888224	15747	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
770	5465852336	12351616	15764	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
771	5478208304	4000	15778										0.000	56.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
772	5478213936	2420480	15781	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
773	5480636656	3707936	15814	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
774	5484346576	3735616	15816	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
775	5488084176	5065568	15832	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
776	5493151984	5827808	15855	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
777	5498981616	241799776	15858	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
778	5740782896	2166496	15865	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
779	5742950864	1984	15868	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
780	5742954704	2912	15882	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
781	5742958928	1536	15897	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
782	5742962896	46112	15920	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
783	5743012048	2754048	15932	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
784	5745768688	1920	15941	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
785	5745772752	7818272	15970	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
786	5753593104	5449504	15984	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
787	5759044816	3466176	16004	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
788	5762512304	2304	16019	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
789	5762516208	1952	16037	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
790	5762520304	47168	16077	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
791	5762569456	3411264	16085	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
792	5765983440	1952	16088	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
793	5765987568	2425440	16091	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
794	5768415504	1888	16102	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
795	5768419568	34226240	16131	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
796	5802648976	140032	16172	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
797	5802791152	11884448	16180	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
798	5814676880	1920	16183	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
799	5814680848	10341376	16186	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
800	5825025264	1984	16197	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
801	5825029360	2560	16212	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
802	5825033456	15733664	16233	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
803	5840768400	5464736	16246	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
804	5846234416	1920	16254	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
805	5846238480	1408	16265	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
806	5846242512	1312	16276	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
807	5846246608	3072	16290	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
808	5846252752	1344	16302	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
809	5846256880	4864	16314	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
810	5846263056	1088	16327	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
811	5846267152	1664	16339	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
812	5846271184	1504	16352	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
813	5846275312	1280	16363	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
814	5846279376	25952	16384	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
815	5846308080	3495520	16407	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
816	5849806032	2368	16422	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
817	5849810128	2112	16440	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
818	5849814224	48096	16480	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
819	5849864432	3438592	16488	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
820	5853306064	3424	16491	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
821	5853312368	2821952	16494	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
822	5856136464	1856	16505	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
823	5856140528	25366272	16534	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
824	5881509168	11847616	16551	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
825	5893358672	416	16565										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
826	5893360240	2496864	16568	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
827	5895859440	3826176	16601	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
828	5899688272	3904192	16603	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
829	5903594704	5218816	16619	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
830	5908815088	5683744	16642	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
831	5914500336	241798560	16645	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
832	6156301584	2171424	16652	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
833	6158475504	1984	16655	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
834	6158479600	2912	16669	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
835	6158483792	1536	16684	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
836	6158487792	48224	16707	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
837	6158538960	2755936	16719	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
838	6161297648	1952	16728	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
839	6161301712	7908928	16757	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
840	6169213200	5461504	16771	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
841	6174677200	3478752	16791	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
842	6178157808	2304	16806	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
843	6178161904	2016	16824	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
844	6178166000	45856	16864	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
845	6178214128	3426912	16872	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
846	6181643472	1984	16875	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
847	6181647600	2425184	16878	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
848	6184075504	1888	16889	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
849	6184079600	34891392	16918	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
850	6218972432	141280	16959	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
851	6219116784	11920992	16967	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
852	6231040240	1952	16970	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
853	6231044336	10766528	16973	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
854	6241813744	2144	16984	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
855	6241817840	832	16999	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
856	6241819952	16326752	17020	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
857	6258148592	5467264	17033	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
858	6263617776	1984	17041	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
859	6263621904	1408	17052	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
860	6263625968	1312	17063	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
861	6263630064	3264	17077	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
862	6263636208	1440	17089	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
863	6263640304	4704	17101	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
864	6263646480	1088	17114	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
865	6263650544	1600	17126	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
866	6263654640	1472	17139	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
867	6263658736	1280	17150	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
868	6263662832	26688	17171	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
869	6263691504	3502624	17194	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
870	6267195632	2400	17209	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
871	6267199696	1984	17227	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
872	6267203568	47584	17267	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
873	6267254000	3522112	17275	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
874	6270778608	3968	17278	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
875	6270784752	2439168	17281	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
876	6273226992	1888	17292	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
877	6273231088	25829856	17321	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
878	6299062512	11811904	17338	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
879	6310876048	416	17352										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
880	6310877328	2378688	17355	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
881	6313257552	3908832	17388	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
882	6317167952	3972352	17390	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
883	6321143472	5239040	17406	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
884	6326385040	5665728	17429	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
885	6332052720	240203808	17432	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
886	6572259600	2177760	17439	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
887	6574439664	1952	17442	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
888	6574443728	2976	17456	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
889	6574447984	1504	17471	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
890	6574451920	48864	17494	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
891	6574502192	2758368	17506	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
892	6577262832	1920	17515	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
893	6577266928	7814496	17544	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
894	6585084176	5459264	17558	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
895	6590546160	3478464	17578	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
896	6594025904	2400	17593	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
897	6594029808	2112	17611	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
898	6594033904	47328	17651	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
899	6594083024	3426688	17659	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
900	6597512432	1952	17662	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
901	6597516528	2426080	17665	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
902	6599944464	1856	17676	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
903	6599948528	33846624	17705	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
904	6633797872	172064	17746	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
905	6633972944	12456320	17754	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
906	6646431952	2208	17757	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
907	6646436080	10598688	17760	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
908	6657036656	1952	17771	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
909	6657040656	864	17786	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
910	6657042832	15399488	17807	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
911	6672444656	5615488	17820	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
912	6678061456	2016	17828	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
913	6678065456	1664	17839	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
914	6678069744	1312	17850	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
915	6678073584	3648	17864	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
916	6678079696	1344	17876	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
917	6678083824	11104	17888	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
918	6678096208	1856	17901	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
919	6678100208	2080	17913	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
920	6678104304	4032	17926	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
921	6678110448	1504	17937	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
922	6678114544	62208	17958	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
923	6678178992	3617312	17981	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
924	6681798928	2560	17996	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
925	6681803024	2080	18014	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
926	6681807056	48160	18054	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
927	6681857264	3838528	18062	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
928	6685698288	4000	18065	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
929	6685704432	2440192	18068	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
930	6688146672	1920	18079	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
931	6688150768	26015712	18108	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
932	6714169584	12030144	18125	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
933	6726202224	544	18139										0.000	411.765	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
934	6726203984	2354880	18142	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
935	6728560880	3711456	18175	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
936	6732274928	3963872	18177	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
937	6736241872	5437760	18193	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
938	6741681360	5725024	18216	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
939	6747408624	239633120	18219	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
940	6987044112	2163136	18226	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
941	6989208784	1952	18229	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
942	6989212944	2688	18243	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
943	6989217040	1504	18258	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
944	6989221104	46848	18281	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
945	6989270352	2756896	18293	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
946	6992028880	1952	18302	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
947	6992033040	7841664	18331	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
948	6999876048	5460608	18345	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
949	7005338864	3489408	18365	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
950	7008830704	2496	18380	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
951	7008834768	1984	18398	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
952	7008838864	46144	18438	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
953	7008886992	3410336	18446	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
954	7012299984	1952	18449	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
955	7012304112	2667616	18452	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
956	7014974448	3712	18463	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
957	7014980848	34079488	18492	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
958	7049062704	139456	18533	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
959	7049203952	12768992	18541	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
960	7061975248	1984	18544	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
961	7061979376	10367616	18547	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
962	7072348496	1920	18558	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
963	7072352496	864	18573	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
964	7072354640	15705376	18594	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
965	7088061680	5511744	18607	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
966	7093574960	2272	18615	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
967	7093578992	1408	18626	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
968	7093583088	3680	18637	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
969	7093589200	3616	18651	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
970	7093595248	1408	18663	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
971	7093599472	4448	18675	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
972	7093605616	1088	18688	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
973	7093609744	1664	18700	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
974	7093613808	1472	18713	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
975	7093617936	1280	18724	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
976	7093622000	26848	18745	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
977	7093650640	3741248	18768	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
978	7097394416	13184	18783	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
979	7097409136	13472	18801	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
980	7097425136	97696	18841	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
981	7097525488	3970560	18849	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
982	7101498576	3552	18852	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
983	7101504752	2437920	18855	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
984	7103944944	1888	18866	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
985	7103949040	24860704	18895	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
986	7128811728	12400832	18912	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
987	7141214064	416	18926										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
988	7141215376	2338976	18929	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
989	7143557328	3707360	18962	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
990	7147267280	3739648	18964	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
991	7151008976	5512160	18980	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
992	7156522416	5684192	19003	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
993	7162209552	242227488	19006	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
994	7404438896	2180384	19013	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
995	7406620880	1952	19016	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
996	7406624976	2560	19030	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
997	7406629072	1536	19045	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
998	7406633200	45024	19068	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
999	7406680304	2751680	19080	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1000	7409433840	1856	19089	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1001	7409437936	7852832	19118	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1002	7417292080	5460768	19132	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1003	7422755056	3467584	19152	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1004	7426225392	2368	19167	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1005	7426229488	2112	19185	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1006	7426233616	45408	19225	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1007	7426281712	3408320	19233	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1008	7429691600	1952	19236	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1009	7429695728	2422752	19239	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1010	7432121584	1856	19250	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1011	7432125648	34878368	19279	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1012	7467005072	142560	19320	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1013	7467150576	11880896	19328	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1014	7479033072	1952	19331	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1015	7479037168	10359680	19334	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1016	7489399088	6624	19345	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1017	7489407216	1568	19360	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1018	7489411312	15720800	19381	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1019	7505134864	5521792	19394	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1020	7510659312	3904	19402	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1021	7510665424	1440	19413	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1022	7510669552	3008	19424	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1023	7510675696	3520	19438	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1024	7510681776	2048	19450	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1025	7510685936	6688	19462	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1026	7510694128	2848	19475	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1027	7510698288	1600	19487	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1028	7510702320	2304	19500	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1029	7510706416	2240	19511	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1030	7510710512	29504	19532	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1031	7510741296	3520864	19555	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1032	7514265104	5600	19570	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1033	7514271984	1952	19588	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1034	7514275952	52032	19628	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1035	7514330416	4260224	19636	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1036	7518592208	1984	19639	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1037	7518596336	2447712	19642	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1038	7521046800	1856	19653	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1039	7521050608	25226208	19682	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1040	7546278096	12453792	19699	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1041	7558733712	416	19713										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1042	7558735472	2351712	19716	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1043	7561090288	3714304	19749	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1044	7564807376	3786336	19751	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1045	7568596176	5154144	19767	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1046	7573753072	5813440	19790	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1047	7579568368	240922976	19793	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1048	7820493072	2156896	19800	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1049	7822651664	1952	19803	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1050	7822655728	3360	19817	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1051	7822661840	1568	19832	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1052	7822665872	51264	19855	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1053	7822719184	2771712	19867	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1054	7825492144	1952	19876	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1055	7825496304	7849312	19905	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1056	7833347344	5451136	19919	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1057	7838800240	3479136	19939	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1058	7842280656	2368	19954	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1059	7842284752	2144	19972	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1060	7842288880	47424	20012	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1061	7842339056	3417600	20020	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1062	7845758160	2016	20023	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1063	7845762288	2423840	20026	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1064	7848189168	1888	20037	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1065	7848193008	34270688	20066	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1066	7882465552	140768	20107	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1067	7882607856	11865984	20115	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1068	7894475120	2464	20118	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1069	7894479088	10348864	20121	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1070	7904830704	1952	20132	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1071	7904834800	2656	20147	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1072	7904838896	15733856	20168	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1073	7920575728	5463552	20181	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1074	7926041840	1920	20189	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1075	7926045904	1440	20200	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1076	7926050000	1312	20211	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1077	7926054096	3520	20225	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1078	7926060272	1344	20237	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1079	7926064368	4992	20249	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1080	7926070640	1120	20262	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1081	7926074640	1632	20274	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1082	7926078704	1504	20287	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1083	7926082800	1216	20298	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1084	7926086896	25632	20319	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1085	7926115568	3505824	20342	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1086	7929623792	2432	20357	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1087	7929627888	1920	20375	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1088	7929631984	47648	20415	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1089	7929681136	3448928	20423	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1090	7933133040	4032	20426	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1091	7933139248	2809952	20429	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1092	7935952112	1856	20440	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1093	7935956272	25293632	20469	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1094	7961252080	11798880	20486	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1095	7973053296	448	20500										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1096	7973054672	2382432	20503	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1097	7975439600	3919904	20536	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1098	7979362544	3922944	20538	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1099	7983287504	5215584	20554	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1100	7988505808	5690784	20577	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1101	7994200656	241148000	20580	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1102	8235351312	2184512	20587	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1103	8237538512	1984	20590	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1104	8237542640	3040	20604	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1105	8237548752	1568	20619	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1106	8237552784	52064	20642	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1107	8237606128	2756640	20654	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1108	8240365808	1888	20663	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1109	8240369904	7952640	20692	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1110	8248324368	5449216	20706	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1111	8253775088	3472800	20726	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1112	8257249520	2400	20741	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1113	8257253616	1984	20759	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1114	8257257712	45696	20799	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1115	8257305840	3431232	20807	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1116	8260739280	1952	20810	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1117	8260743472	2425536	20813	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1118	8263171312	1888	20824	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1119	8263175408	34385728	20853	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1120	8297562448	140160	20894	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1121	8297705712	11916032	20902	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1122	8309624048	2432	20905	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1123	8309628144	10800032	20908	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1124	8320430320	1920	20919	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1125	8320434416	832	20934	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1126	8320436624	16112576	20955	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1127	8336551152	5465952	20968	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1128	8342019312	1952	20976	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1129	8342023440	1440	20987	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1130	8342027536	1312	20998	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1131	8342031600	3680	21012	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1132	8342037776	1312	21024	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1133	8342041808	4896	21036	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1134	8342047984	1088	21049	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1135	8342052080	1632	21061	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1136	8342056176	1472	21074	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1137	8342060272	1248	21085	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1138	8342064368	26112	21106	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1139	8342093040	3493632	21129	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1140	8345587952	2560	21144	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1141	8345592016	1984	21162	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1142	8345596144	47840	21202	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1143	8345646320	3444576	21210	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1144	8349092208	4096	21213	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1145	8349098224	2448192	21216	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1146	8351548656	1888	21227	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1147	8351552784	25738272	21256	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1148	8377295632	11891808	21273	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1149	8389188464	544	21287										0.000	411.765	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1150	8389190192	2346944	21290	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1151	8391538928	3761664	21323	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1152	8395303152	4026528	21325	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1153	8399331536	5417472	21341	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1154	8404751600	5663296	21364	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1155	8410416368	241593728	21367	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1156	8652011792	2164960	21374	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1157	8654178544	1920	21377	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1158	8654182640	2976	21391	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1159	8654186896	1536	21406	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1160	8654190832	48416	21429	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1161	8654242032	2752160	21441	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1162	8656995568	1888	21450	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1163	8656999664	7942144	21479	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1164	8664943920	5458208	21493	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1165	8670404848	3475616	21513	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1166	8673883376	4320	21528	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1167	8673889680	19776	21546	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1168	8673912016	52288	21586	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1169	8673967312	4322208	21594	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1170	8678291696	1952	21597	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1171	8678295760	2444640	21600	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1172	8680743152	1920	21611	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1173	8680747248	34577664	21640	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1174	8715326704	141088	21681	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1175	8715469072	12382976	21689	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1176	8727853392	1984	21692	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1177	8727857360	10443584	21695	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1178	8738303216	1952	21706	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1179	8738307312	2592	21721	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1180	8738311408	15308832	21742	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1181	8753622256	5479104	21755	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1182	8759102704	1952	21763	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1183	8759106800	1408	21774	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1184	8759110896	3648	21785	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1185	8759117040	3072	21799	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1186	8759123184	1376	21811	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1187	8759127184	4640	21823	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1188	8759133456	1152	21836	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1189	8759137264	1600	21848	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1190	8759141584	1472	21861	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1191	8759145712	1248	21872	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1192	8759149776	25216	21893	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1193	8759176432	3694624	21916	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1194	8762874096	2400	21931	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1195	8762878256	2176	21949	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1196	8762882288	50528	21989	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1197	8762934480	3638240	21997	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1198	8766574800	1952	22000	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1199	8766578928	2442912	22003	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1200	8769024368	1920	22014	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1201	8769028368	25552160	22043	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1202	8794582288	12077248	22060	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1203	8806660976	448	22074										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1204	8806662096	2352160	22077	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1205	8809015568	3721376	22110	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1206	8812738768	4098432	22112	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1207	8816839888	5346304	22128	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1208	8822189264	5725184	22151	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1209	8827916528	241606016	22154	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1210	9069525264	2174688	22161	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1211	9071702320	1952	22164	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1212	9071706384	2720	22178	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1213	9071710416	1536	22193	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1214	9071714512	47232	22216	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1215	9071763664	2758720	22228	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1216	9074524400	1952	22237	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1217	9074528496	7921504	22266	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1218	9082451344	5461920	22280	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1219	9087915248	3486688	22300	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1220	9091405040	2400	22315	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1221	9091409136	2080	22333	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1222	9091413200	46048	22373	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1223	9091462352	3874528	22381	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1224	9095339248	1952	22384	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1225	9095343344	2624832	22387	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1226	9097970928	1856	22398	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1227	9097975024	33697120	22427	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1228	9131674896	139840	22468	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1229	9131817200	11919712	22476	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1230	9143739600	1952	22479	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1231	9143743760	10381664	22482	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1232	9154128944	1952	22493	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1233	9154133104	1824	22508	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1234	9154136304	15711680	22529	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1235	9169850608	5507136	22542	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1236	9175359728	1920	22550	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1237	9175363824	1568	22561	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1238	9175367920	3168	22572	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1239	9175374096	3392	22586	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1240	9175380144	1344	22598	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1241	9175384336	4928	22610	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1242	9175390736	1120	22623	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1243	9175394448	1632	22635	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1244	9175398640	1696	22648	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1245	9175402768	7616	22659	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1246	9175412976	26336	22680	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1247	9175441520	3844544	22703	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1248	9179287792	2336	22718	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1249	9179291888	2048	22736	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1250	9179295984	48768	22776	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1251	9179347184	3828800	22784	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1252	9183177968	4032	22787	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1253	9183184112	2423808	22790	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1254	9185611024	1888	22801	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1255	9185615056	24620416	22830	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1256	9210238192	12466528	22847	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1257	9222706064	480	22861										0.000	466.667	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1258	9222707760	2350880	22864	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1259	9225061616	3725056	22897	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1260	9228787952	3735968	22899	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1261	9232525520	5525984	22915	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1262	9238053104	5802432	22938	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1263	9243858128	242206816	22941	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1264	9486066960	2168960	22948	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1265	9488237840	2016	22951	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1266	9488241904	2944	22965	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1267	9488246128	1536	22980	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1268	9488250096	45504	23003	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1269	9488297168	2763104	23015	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1270	9491063024	1920	23024	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1271	9491067120	7944544	23053	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1272	9499014416	5450496	23067	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1273	9504466192	3475360	23087	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1274	9507942800	2368	23102	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1275	9507946736	2048	23120	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1276	9507950832	47552	23160	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1277	9507999984	3421152	23168	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1278	9511424272	1952	23171	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1279	9511428336	2483008	23174	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1280	9513912656	2368	23185	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1281	9513916848	34376960	23214	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1282	9548296464	139616	23255	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1283	9548438768	12796832	23263	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1284	9561237712	2240	23266	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1285	9561241840	10357696	23269	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1286	9571601648	1920	23280	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1287	9571605744	832	23295	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1288	9571607856	15753632	23316	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1289	9587363056	5458112	23329	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1290	9592823024	1984	23337	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1291	9592827216	1408	23348	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1292	9592831216	1312	23359	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1293	9592835312	3296	23373	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1294	9592841456	1344	23385	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1295	9592845456	4448	23397	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1296	9592851664	1088	23410	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1297	9592855696	1632	23422	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1298	9592859888	1472	23435	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1299	9592863984	1280	23446	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1300	9592868080	25184	23467	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1301	9592894704	3628416	23490	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1302	9596533904	17728	23505	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1303	9596554480	11680	23523	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1304	9596568848	69088	23563	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1305	9596639472	4118272	23571	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1306	9600760016	3840	23574	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1307	9600766224	2419840	23577	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1308	9603187952	1888	23588	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1309	9603192048	24930496	23617	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1310	9628124400	12466368	23634	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1311	9640592240	544	23648										0.000	411.765	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1312	9640593680	2350944	23651	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1313	9642946832	3729024	23684	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1314	9646678224	3748576	23686	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1315	9650428112	5354496	23702	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1316	9655784720	5783808	23725	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1317	9661571312	241193152	23728	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1318	9902766384	2168640	23735	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1319	9904937200	1952	23738	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1320	9904941296	2944	23752	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1321	9904945520	1568	23767	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1322	9904949456	47232	23790	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1323	9904998640	2753088	23802	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1324	9907754224	1888	23811	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1325	9907758320	7844192	23840	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1326	9915605264	5462816	23854	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1327	9921070352	3466848	23874	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1328	9924539632	2304	23889	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1329	9924543728	2048	23907	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1330	9924547792	47456	23947	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1331	9924596944	3416192	23955	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1332	9928016112	1952	23958	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1333	9928020208	2424832	23961	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1334	9930448112	1856	23972	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1335	9930452208	34503808	24001	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1336	9964957968	140736	24042	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1337	9965101264	12747232	24050	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1338	9977850096	2208	24053	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1339	9977854192	10374624	24056	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1340	9988231408	1920	24067	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1341	9988235504	2528	24082	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1342	9988239600	15701312	24103	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1343	10003943664	5469120	24116	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1344	10009414896	1920	24124	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1345	10009418992	1408	24135	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1346	10009423088	1376	24146	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1347	10009427184	3136	24160	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1348	10009433328	1344	24172	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1349	10009437424	6720	24184	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1350	10009445616	1088	24197	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1351	10009449712	1600	24209	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1352	10009453808	1472	24222	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1353	10009457904	1312	24233	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1354	10009462000	27584	24254	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1355	10009492688	3491904	24277	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1356	10012987664	2528	24292	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1357	10012991824	2016	24310	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1358	10012995824	48352	24350	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1359	10013046000	4232896	24358	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1360	10017281232	4096	24361	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1361	10017287376	2498976	24364	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1362	10019789040	1952	24375	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1363	10019793104	24911776	24404	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1364	10044707056	12388768	24421	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1365	10057100048	416	24435										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1366	10057102288	2363744	24438	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1367	10059469072	3784768	24471	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1368	10063255792	3747296	24473	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1369	10067004368	5073216	24489	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1370	10072080656	5698400	24512	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1371	10077780336	242541280	24515	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1372	10320322960	2168832	24522	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1373	10322493680	1920	24525	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1374	10322497776	2944	24539	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1375	10322502032	1536	24554	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1376	10322505968	49696	24577	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1377	10322557232	2751232	24589	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1378	10325310704	1952	24598	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1379	10325314800	7992640	24627	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1380	10333310288	5463872	24641	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1381	10338775408	3477760	24661	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1382	10342254832	2368	24676	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1383	10342258928	2048	24694	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1384	10342263024	46048	24734	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1385	10342312176	3447488	24742	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1386	10345762032	1952	24745	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1387	10345766128	2423296	24748	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1388	10348191984	1920	24759	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1389	10348196080	34563424	24788	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1390	10382762288	140576	24829	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1391	10382905648	12654624	24837	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1392	10395562192	2688	24840	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1393	10395566320	10480992	24843	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1394	10406050032	1920	24854	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1395	10406054096	1088	24869	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1396	10406056656	15701088	24890	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1397	10421760240	5465056	24903	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1398	10427228432	1920	24911	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1399	10427232496	1408	24922	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1400	10427236592	1280	24933	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1401	10427240688	3232	24947	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1402	10427246864	1344	24959	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1403	10427250928	4832	24971	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1404	10427257040	1120	24984	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1405	10427261232	1632	24996	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1406	10427265264	1472	25009	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1407	10427269360	1280	25020	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1408	10427273456	25920	25041	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1409	10427302128	3508896	25064	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1410	10430812144	2464	25079	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1411	10430816496	2016	25097	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1412	10430820592	48512	25137	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1413	10430871792	3516544	25145	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1414	10434391280	4384	25148	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1415	10434397424	2736736	25151	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1416	10437136624	1888	25162	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1417	10437140720	25314144	25191	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1418	10462456176	12035712	25208	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1419	10474496016	448	25222										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1420	10474497360	2454048	25225	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1421	10476953840	3817408	25258	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1422	10480773360	3902560	25260	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1423	10484678928	5091520	25276	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1424	10489772272	5706880	25299	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1425	10495482096	242566784	25302	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1426	10738051344	2171840	25309	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1427	10740226256	1984	25312	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1428	10740230384	3104	25326	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1429	10740236496	1536	25341	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1430	10740240528	47072	25364	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1431	10740289776	2751712	25376	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1432	10743043312	1888	25385	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1433	10743047408	7940800	25414	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1434	10750989584	5517184	25428	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1435	10756508912	3821792	25448	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1436	10760332496	2496	25463	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1437	10760336720	2048	25481	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1438	10760340720	46816	25521	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1439	10760388816	3812448	25529	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1440	10764204272	1920	25532	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1441	10764208336	2426976	25535	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1442	10766637296	1888	25546	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1443	10766641392	34306016	25575	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1444	10800949488	140000	25616	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1445	10801091824	12269024	25624	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1446	10813362416	5312	25627	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1447	10813370608	10774496	25630	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1448	10824147184	4960	25641	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1449	10824153424	7264	25656	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1450	10824163568	15771104	25677	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1451	10839935984	5466784	25690	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1452	10845405424	1920	25698	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1453	10845409616	1440	25709	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1454	10845413648	3168	25720	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1455	10845419760	3168	25734	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1456	10845425808	1312	25746	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1457	10845430000	4544	25758	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1458	10845436144	1088	25771	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1459	10845440272	1632	25783	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1460	10845444336	1472	25796	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1461	10845448400	1312	25807	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1462	10845452528	26560	25828	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1463	10845481200	3503456	25851	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1464	10848987408	2368	25866	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1465	10848991472	2016	25884	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1466	10848995568	48320	25924	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1467	10849045712	3440576	25932	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1468	10852487568	4032	25935	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1469	10852493584	2674176	25938	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1470	10855170288	1920	25949	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1471	10855174416	25492928	25978	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1472	10880668912	11806688	25995	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1473	10892477296	448	26009										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1474	10892478928	2395648	26012	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1475	10894875888	3943904	26045	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1476	10898821328	3916544	26047	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1477	10902740176	5236320	26063	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1478	10907977936	5662944	26086	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1479	10913642736	242114368	26089	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1480	11155759408	2167616	26096	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1481	11157928336	1920	26099	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1482	11157932400	2912	26113	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1483	11157938448	1504	26128	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1484	11157942384	47264	26151	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1485	11157991632	2756704	26163	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1486	11160751312	1984	26172	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1487	11160755472	7890016	26201	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1488	11168647472	5505440	26215	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1489	11174154512	3673952	26235	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1490	11177830640	2592	26250	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1491	11177834704	2048	26268	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1492	11177841872	55936	26308	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1493	11177899216	4036064	26316	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1494	11181937872	1984	26319	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1495	11181942000	2423648	26322	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1496	11184367856	1888	26333	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1497	11184371952	34814112	26362	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1498	11219188016	140544	26403	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1499	11219331312	11914592	26411	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1500	11231248592	1952	26414	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1501	11231252720	10789568	26417	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1502	11242043632	2112	26428	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1503	11242047696	960	26443	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1504	11242049968	16312800	26464	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1505	11258365168	5469024	26477	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1506	11263835472	2144	26485	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1507	11263840496	1408	26496	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1508	11263844592	1280	26507	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1509	11263848720	3296	26521	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1510	11263854864	1440	26533	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1511	11263858928	4320	26545	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1512	11263865072	1120	26558	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1513	11263869168	1600	26570	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1514	11263873264	1504	26583	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1515	11263877360	1248	26594	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1516	11263881424	25120	26615	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1517	11263908080	3509536	26638	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1518	11267419344	2432	26653	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1519	11267423472	1984	26671	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1520	11267427568	48000	26711	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1521	11267476944	3441024	26719	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1522	11270920432	3776	26722	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1523	11270926576	2436416	26725	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1524	11273364720	1856	26736	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1525	11273368816	25923936	26765	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1526	11299295472	11797920	26782	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1527	11311094640	448	26796										0.000	500.000	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1528	11311095984	2385056	26799	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1529	11313484016	3934176	26832	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1530	11317421296	3955424	26834	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1531	11321379536	5190528	26850	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1532	11326571728	5673920	26873	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1533	11332247792	242693856	26876	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1534	11574944016	2174560	26883	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1535	11577120976	1984	26886	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1536	11577125104	2752	26900	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1537	11577129168	1536	26915	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1538	11577133264	45664	26938	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1539	11577180368	2752608	26950	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1540	11579935952	1920	26959	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1541	11579940048	7985536	26988	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1542	11587928368	5481344	27002	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1543	11593411216	3706112	27022	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1544	11597119728	2560	27037	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1545	11597123856	2336	27055	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1546	11597127920	47072	27095	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1547	11597178096	4002432	27103	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1548	11601182960	1952	27106	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1549	11601186800	2423616	27109	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1550	11603611888	1888	27120	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1551	11603615984	34343840	27149	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1552	11637962000	140544	27190	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1553	11638105328	11934080	27198	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1554	11650041072	2144	27201	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1555	11650045168	10771616	27204	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1556	11660819696	2016	27215	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1557	11660823792	1056	27230	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1558	11660827888	16033728	27251	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1559	11676863728	5465600	27264	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1560	11682330896	1952	27272	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1561	11682334992	1440	27283	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1562	11682337936	1312	27294	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1563	11682341072	3296	27308	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1564	11682347248	1312	27320	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1565	11682351248	4576	27332	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1566	11682357488	1088	27345	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1567	11682361584	1632	27357	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1568	11682365680	1472	27370	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1569	11682369776	1280	27381	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1570	11682373872	26848	27402	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1571	11682402544	3503616	27425	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1572	11685908720	2496	27440	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1573	11685912784	1984	27458	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1574	11685916912	47392	27498	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1575	11685966064	3441888	27506	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1576	11689409744	3264	27509	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1577	11689415920	2435552	27512	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1578	11691853072	1856	27523	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1579	11691857136	26544096	27552	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1580	11718404336	11803168	27569	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1581	11730208624	416	27583										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1582	11730210192	2344192	27586	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1583	11732556080	3879392	27619	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1584	11736438000	3889536	27621	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1585	11740330192	5382944	27637	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1586	11745714416	5676192	27660	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1587	11751392528	241273408	27663	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1588	11992667440	2164000	27670	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1589	11994834128	1952	27673	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1590	11994838256	2848	27687	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1591	11994843376	1536	27702	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1592	11994847344	47232	27725	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1593	11994896592	2758016	27737	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1594	11997657328	2080	27746	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1595	11997661392	7871520	27775	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1596	12005536048	5454944	27789	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1597	12010993904	3469600	27809	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1598	12014465264	2304	27824	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1599	12014469360	2080	27842	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1600	12014473456	46176	27882	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1601	12014522608	3422656	27890	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1602	12017946864	1952	27893	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1603	12017950960	2425152	27896	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1604	12020378864	1888	27907	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1605	12020382992	34162048	27936	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1606	12054547696	139200	27977	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1607	12054688272	12332832	27985	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1608	12067024112	1952	27988	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1609	12067028208	10806528	27991	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1610	12077837552	1920	28002	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1611	12077841616	3072	28017	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1612	12077847792	15536672	28038	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1613	12093387024	5629920	28051	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1614	12099018992	1920	28059	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1615	12099023152	1408	28070	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1616	12099027184	1472	28081	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1617	12099031280	3360	28095	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1618	12099036400	1312	28107	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1619	12099040496	4448	28119	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1620	12099046640	1184	28132	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1621	12099050704	1824	28144	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1622	12099054832	1632	28157	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1623	12099058928	1280	28168	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1624	12099063024	25696	28189	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1625	12099091696	3543520	28212	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1626	12102636784	2464	28227	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1627	12102640880	2304	28245	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1628	12102644944	63616	28285	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1629	12102710480	3828128	28293	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1630	12106541296	3296	28296	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1631	12106547440	2443872	28299	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1632	12108993776	1888	28310	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1633	12108997872	24936000	28339	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1634	12133935344	11814080	28356	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1635	12145751952	416	28370										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1636	12145753296	2352896	28373	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1637	12148107472	3725216	28406	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1638	12151834832	3740704	28408	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1639	12155578576	5090944	28424	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1640	12160671984	5663328	28447	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1641	12166337776	241822912	28450	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1642	12408162608	2157568	28457	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1643	12410323184	1952	28460	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1644	12410327280	2976	28474	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1645	12410331568	1504	28489	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1646	12410335216	50112	28512	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1647	12410386704	2750304	28524	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1648	12413139184	1920	28533	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1649	12413143280	7919840	28562	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1650	12421064976	5455904	28576	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1651	12426523888	3468640	28596	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1652	12429995248	2368	28611	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1653	12429999344	1984	28629	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1654	12430003408	48128	28669	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1655	12430054640	3424576	28677	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1656	12433481968	1952	28680	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1657	12433486096	2426240	28683	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1658	12435913968	1856	28694	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1659	12435918064	33735648	28723	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1660	12469656848	140768	28764	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1661	12469799152	12803680	28772	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1662	12482604336	2176	28775	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1663	12482608368	10379136	28778	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1664	12492989680	1952	28789	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1665	12492993776	864	28804	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1666	12492995920	15888288	28825	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1667	12508887280	5504480	28838	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1668	12514393328	2112	28846	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1669	12514397424	1408	28857	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1670	12514401520	1280	28868	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1671	12514405616	3744	28882	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1672	12514411728	1344	28894	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1673	12514414544	4896	28906	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1674	12514421904	1088	28919	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::CUDAFunctorOnSelf_add<long>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1675	12514426096	1632	28931	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)
1676	12514430192	1760	28944	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::CUDAFunctor_add<float>, std::array<char *, (unsigned long)3>, (int)4, TrivialOffsetCalculator<(int)2, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)2>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1677	12514434288	1248	28955	1	1	1	128	1	1	23	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::<unnamed>::lerp_tensor_kernel(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float, float, float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1678	12514438352	34208	28976	3024	1	1	2	32	1	164	0.000	0.000					NVIDIA GB10 (0)	1		7	std::enable_if<!T7, void>::type internal::gemvx::kernel<int, int, float, float, float, float, (bool)0, (bool)1, (bool)1, (bool)0, (int)6, (bool)0, cublasGemvParamsEx<int, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<const float>, cublasGemvTensorStridedBatched<float>, float>>(T13)
1679	12514475248	3824608	28999	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1680	12518302960	2528	29014	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1681	12518307056	2016	29032	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1682	12518311152	47136	29072	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1683	12518360272	3841696	29080	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1684	12522203344	3968	29083	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1685	12522209552	2437024	29086	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1686	12524648688	1888	29097	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1687	12524652816	24963488	29126	296	168	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1688	12549617904	12417312	29143	529340	1	1	128	1	1	117	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::rope_kernel<__nv_bfloat16, __nv_bfloat16, __nv_bfloat16, (bool)1, (bool)1, (bool)1, (bool)1, (bool)0>(const T1 *, const T1 *, const T2 *, const T3 *, const T3 *, T1 *, T1 *, long, long, long, int, int, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, long, float)
1689	12562036624	416	29157										0.000	538.462	Device		NVIDIA GB10 (0)	1		7	[CUDA memset]
1690	12562038256	2346336	29160	56	37	1	32	4	1	60	0.000	0.002					NVIDIA GB10 (0)	1		7	void at::native::reduce_kernel<(int)128, (int)4, at::native::ReduceOp<c10::BFloat16, at::native::MeanOps<c10::BFloat16, float, float, c10::BFloat16>, unsigned int, c10::BFloat16, (int)4, (int)4>>(T3)
1691	12564387056	3717536	29193	1184	56	1	512	1	1	27	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)32, (unsigned int)1, (bool)0, (bool)0, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1692	12568106224	3741536	29195	591	56	1	1024	1	1	25	0.000	0.000					NVIDIA GB10 (0)	1		7	void QuantInt8Kernel<(unsigned int)128, (unsigned int)64, (unsigned int)1, (bool)0, (bool)1, __nv_bfloat16>(T6 *, T6 *, signed char *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1693	12571849040	5525152	29211	591	56	1	1024	1	1	24	0.033	0.000					NVIDIA GB10 (0)	1		7	void TransposePadPermuteKernel<(unsigned int)128, (unsigned int)64, (bool)1, __nv_bfloat16>(T4 *, T4 *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1694	12577376496	5689024	29234	56	1	128	256	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void MeanScaleKernel<(unsigned int)64, (bool)0, __nv_bfloat16>(T3 *, signed char *, float *, float *, float, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int)
1695	12583066864	242627456	29237	296	56	1	32	4	1	255	0.000	0.033					NVIDIA GB10 (0)	1		7	void qk_int_sv_f8_attn_kernel<(unsigned int)128, (unsigned int)64, (unsigned int)32, (unsigned int)64, (unsigned int)128, (DataType)1, (QuantGranularity)2, (QuantGranularity)2, float, (bool)1, __nv_bfloat16, (ComputeUnit)1, (MaskMode)0, (bool)0, (bool)1, (bool)0, (bool)1>(signed char *, signed char *, signed char *, T11 *, float *, float *, float *, float *, float *, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, float)
1696	12825695632	2167232	29244	256	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_bf16_vec_kernel(const uint4 *, const unsigned short *, float *, long, long)
1697	12827865328	1984	29247	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_warp_kernel(const float *, float *, long, float)
1698	12827869456	2976	29261	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1699	12827873744	1504	29276	1	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1700	12827877616	47264	29299	8288	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1701	12827926736	2752640	29311	529536	1	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	void comfy::<unnamed>::quantize_nvfp4_kernel<__nv_bfloat16, __nv_fp4x2_e2m1, __nv_fp8_e4m3, (bool)1, (bool)1>(const T1 *, const float *, T2 *, T3 *, unsigned long, unsigned long, unsigned long, unsigned long, float)
1702	12830681328	1952	29320	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1703	12830685424	7922240	29349	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1704	12838610192	5462688	29363	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1705	12844074192	3470944	29383	37810	1	1	32	4	1	38	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::<unnamed>::vectorized_layer_norm_kernel<c10::BFloat16, float, (bool)1>(int, T2, const T1 *, const T1 *, const T1 *, T2 *, T2 *, T1 *)
1706	12847546608	2432	29398	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1707	12847550704	2112	29416	63	1	1	128	1	1	18	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3)
1708	12847554800	47168	29456	6216	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1709	12847604976	3418752	29464	256	1	1	128	1	1	60	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::partial_absmax_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, float *, long, long)
1710	12851025136	1952	29467	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1711	12851029232	2467008	29470	37888	1	1	256	1	1	39	0.000	0.000					NVIDIA GB10 (0)	1		7	void <unnamed>::quantize_nvfp4_modulated_bf16_kernel<float>(const unsigned short *, const T1 *, const T1 *, const int *, const float *, unsigned char *, unsigned char *, long, long, long, long, long, long)
1712	12853498096	2080	29481	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1713	12853502192	34443328	29510	296	224	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1714	12887948560	140224	29551	16576	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<unsigned char>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1715	12888091888	11968896	29559	256	1	1	128	1	1	28	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::partial_absmax_swiglu_bf16_kernel(const unsigned short *, float *, long, long)
1716	12900063440	2208	29562	1	1	1	128	1	1	24	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::final_scale_bf16_compat_kernel(const float *, float *, long, float)
1717	12900067536	10367680	29565	37888	1	1	256	1	1	45	0.000	0.000					NVIDIA GB10 (0)	1		7	<unnamed>::quantize_nvfp4_swiglu_bf16_kernel(const unsigned short *, const float *, unsigned char *, unsigned char *, long, long, long, long, long)
1718	12910436592	1952	29576	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3)
1719	12910440688	832	29591	1	1	1	128	1	1	16	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::FillFunctor<float>, std::array<char *, (unsigned long)1>>(int, T2, T3)
1720	12910442800	15742528	29612	296	42	1	384	1	1	168	0.000	0.088					NVIDIA GB10 (0)	1		7	cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu
1721	12926187760	5472992	29625	37810	6	1	128	1	1	32	0.000	0.000					NVIDIA GB10 (0)	1		7	_gate_add_kernel
1722	12931662032	1984	29633	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1723	12931666160	1440	29644	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::AUnaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)2>>(int, T2, T3)
1724	12931670256	3712	29655	1	1	1	128	1	1	36	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)4, at::native::floor_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1725	12931676400	3456	29669	1	1	1	128	1	1	33	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int>, TrivialOffsetCalculator<(int)1, unsigned int>, at::native::memory::LoadWithCast<(int)1>, at::native::memory::StoreWithCast<(int)1>>(int, T1, T2, T4, T5, T6, T7)
1726	12931682448	1344	29681	1	1	1	128	1	1	40	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::vectorized_elementwise_kernel<(int)2, at::native::<unnamed>::launch_clamp_scalar(at::TensorIteratorBase &, c10::Scalar, c10::Scalar, at::native::detail::ClampLimits)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(long) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3)
1727	12931686640	4352	29693	1	1	1	128	1	1	48	0.000	0.000					NVIDIA GB10 (0)	1		7	void at::native::index_elementwise_kernel<(int)128, (int)4, void at::native::gpu_index_kernel<void at::native::index_kernel_impl<at::native::OpaqueType<(int)4>>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>)::[lambda(char *, const char *, long) (instance 1)]>(at::TensorIteratorBase &, c10::ArrayRef<long>, c10::ArrayRef<long>, const T1 &, bool)::[lambda(int) (instance 1)]>(long, T3)

The file is too large to be shown. View raw