h3-blackwell-runtime/benchmarks/gb10-post-fc2-block24-targeted-20260826.csv
2026-08-26 16:14:12 +07:00

50 KiB

1IDProcess IDProcess NameHost NameKernel NameContextStreamBlock SizeGrid SizeDeviceCCc2clink__enabled_maskc2clink__presentderived__local_spilling_requestsderived__local_spilling_requests_pctderived__pct_occupancy_per_barrier_countderived__pct_occupancy_per_block_sizederived__pct_occupancy_per_register_countderived__pct_occupancy_per_shared_mem_sizederived_tempMetric0device__attribute_architecturedevice__attribute_async_engine_countdevice__attribute_can_flush_remote_writesdevice__attribute_can_map_host_memorydevice__attribute_can_tex2d_gatherdevice__attribute_can_use_64_bit_stream_mem_opsdevice__attribute_can_use_64_bit_stream_mem_ops_v1device__attribute_can_use_host_pointer_for_registered_memdevice__attribute_can_use_stream_mem_ops_v1device__attribute_can_use_stream_wait_value_nordevice__attribute_can_use_stream_wait_value_nor_v1device__attribute_chipdevice__attribute_clock_ratedevice__attribute_cluster_launchdevice__attribute_compute_capability_majordevice__attribute_compute_capability_minordevice__attribute_compute_modedevice__attribute_compute_preemption_supporteddevice__attribute_concurrent_kernelsdevice__attribute_concurrent_managed_accessdevice__attribute_confidential_computing_modedevice__attribute_cooperative_launchdevice__attribute_cooperative_multi_device_launchdevice__attribute_deferred_mapping_cuda_array_supporteddevice__attribute_device_indexdevice__attribute_direct_managed_mem_access_from_hostdevice__attribute_display_namedevice__attribute_dma_buf_supporteddevice__attribute_ecc_enableddevice__attribute_fb_bus_widthdevice__attribute_fbp_countdevice__attribute_generic_compression_supporteddevice__attribute_global_l1_cache_supporteddevice__attribute_global_memory_bus_widthdevice__attribute_gpu_direct_rdma_flush_writes_optionsdevice__attribute_gpu_direct_rdma_supporteddevice__attribute_gpu_direct_rdma_with_cuda_vmm_supporteddevice__attribute_gpu_direct_rdma_writes_orderingdevice__attribute_gpu_overlapdevice__attribute_gpu_pci_device_iddevice__attribute_gpu_pci_ext_device_iddevice__attribute_gpu_pci_ext_downstream_link_ratedevice__attribute_gpu_pci_ext_downstream_link_widthdevice__attribute_gpu_pci_ext_gendevice__attribute_gpu_pci_ext_gpu_gendevice__attribute_gpu_pci_ext_gpu_link_ratedevice__attribute_gpu_pci_ext_gpu_link_widthdevice__attribute_gpu_pci_revision_iddevice__attribute_gpu_pci_sub_system_iddevice__attribute_handle_type_fabric_supporteddevice__attribute_handle_type_posix_file_descriptor_supporteddevice__attribute_handle_type_win32_handle_supporteddevice__attribute_handle_type_win32_kmt_handle_supporteddevice__attribute_host_native_atomic_supporteddevice__attribute_host_numa_iddevice__attribute_host_register_supporteddevice__attribute_implementationdevice__attribute_integrateddevice__attribute_ipc_event_supporteddevice__attribute_kernel_exec_timeoutdevice__attribute_l2_cache_sizedevice__attribute_l2s_countdevice__attribute_limits_max_cta_per_smdevice__attribute_limits_num_tpcsdevice__attribute_local_l1_cache_supporteddevice__attribute_managed_memorydevice__attribute_max_access_policy_window_sizedevice__attribute_max_block_dim_xdevice__attribute_max_block_dim_ydevice__attribute_max_block_dim_zdevice__attribute_max_blocks_per_multiprocessordevice__attribute_max_gpu_frequency_khzdevice__attribute_max_grid_dim_xdevice__attribute_max_grid_dim_ydevice__attribute_max_grid_dim_zdevice__attribute_max_ipc_per_multiprocessordevice__attribute_max_ipc_per_schedulerdevice__attribute_max_mem_frequency_khzdevice__attribute_max_persisting_l2_cache_sizedevice__attribute_max_pitchdevice__attribute_max_registers_per_blockdevice__attribute_max_registers_per_multiprocessordevice__attribute_max_registers_per_threaddevice__attribute_max_shared_memory_per_blockdevice__attribute_max_shared_memory_per_block_optindevice__attribute_max_shared_memory_per_multiprocessordevice__attribute_max_threads_per_blockdevice__attribute_max_threads_per_multiprocessordevice__attribute_max_warps_per_multiprocessordevice__attribute_max_warps_per_schedulerdevice__attribute_maximum_surface1d_layered_layersdevice__attribute_maximum_surface1d_layered_widthdevice__attribute_maximum_surface1d_widthdevice__attribute_maximum_surface2d_heightdevice__attribute_maximum_surface2d_layered_heightdevice__attribute_maximum_surface2d_layered_layersdevice__attribute_maximum_surface2d_layered_widthdevice__attribute_maximum_surface2d_widthdevice__attribute_maximum_surface3d_depthdevice__attribute_maximum_surface3d_heightdevice__attribute_maximum_surface3d_widthdevice__attribute_maximum_surfacecubemap_layered_layersdevice__attribute_maximum_surfacecubemap_layered_widthdevice__attribute_maximum_surfacecubemap_widthdevice__attribute_maximum_texture1d_layered_layersdevice__attribute_maximum_texture1d_layered_widthdevice__attribute_maximum_texture1d_linear_widthdevice__attribute_maximum_texture1d_mipmapped_widthdevice__attribute_maximum_texture1d_widthdevice__attribute_maximum_texture2d_gather_heightdevice__attribute_maximum_texture2d_gather_widthdevice__attribute_maximum_texture2d_heightdevice__attribute_maximum_texture2d_layered_heightdevice__attribute_maximum_texture2d_layered_layersdevice__attribute_maximum_texture2d_layered_widthdevice__attribute_maximum_texture2d_linear_heightdevice__attribute_maximum_texture2d_linear_pitchdevice__attribute_maximum_texture2d_linear_widthdevice__attribute_maximum_texture2d_mipmapped_heightdevice__attribute_maximum_texture2d_mipmapped_widthdevice__attribute_maximum_texture2d_widthdevice__attribute_maximum_texture3d_depthdevice__attribute_maximum_texture3d_depth_alternatedevice__attribute_maximum_texture3d_heightdevice__attribute_maximum_texture3d_height_alternatedevice__attribute_maximum_texture3d_widthdevice__attribute_maximum_texture3d_width_alternatedevice__attribute_maximum_texturecubemap_layered_layersdevice__attribute_maximum_texturecubemap_layered_widthdevice__attribute_maximum_texturecubemap_widthdevice__attribute_mem_sync_domain_countdevice__attribute_memory_clock_ratedevice__attribute_memory_pools_supporteddevice__attribute_mempool_supported_handle_typesdevice__attribute_mps_enableddevice__attribute_multi_gpu_boarddevice__attribute_multi_gpu_board_group_iddevice__attribute_multicast_supporteddevice__attribute_multiprocessor_countdevice__attribute_num_l2s_per_fbpdevice__attribute_num_schedulers_per_multiprocessordevice__attribute_num_tex_per_multiprocessordevice__attribute_numa_configdevice__attribute_pageable_memory_accessdevice__attribute_pageable_memory_access_uses_host_page_tablesdevice__attribute_pci_bus_iddevice__attribute_pci_device_iddevice__attribute_pci_domain_iddevice__attribute_ram_locationdevice__attribute_ram_typedevice__attribute_reserved_shared_memory_per_blockdevice__attribute_sass_leveldevice__attribute_single_to_double_precision_perf_ratiodevice__attribute_sparse_cuda_array_supporteddevice__attribute_stream_priorities_supporteddevice__attribute_surface_alignmentdevice__attribute_tcc_driverdevice__attribute_tensor_map_access_supporteddevice__attribute_texture_alignmentdevice__attribute_texture_pitch_alignmentdevice__attribute_total_constant_memorydevice__attribute_total_memorydevice__attribute_unified_addressingdevice__attribute_unified_function_pointersdevice__attribute_virtual_address_management_supporteddevice__attribute_warp_sizegpc__cycles_elapsed.avggpc__cycles_elapsed.avg.per_secondgpc__cycles_elapsed.maxgpc__cycles_elapsed.max.per_secondgpc__cycles_elapsed.mingpc__cycles_elapsed.min.per_secondgpc__cycles_elapsed.sumgpc__cycles_elapsed.sum.per_secondgpu__compute_memory_access_throughput.avg.pct_of_peak_sustained_elapsedgpu__compute_memory_access_throughput.max.pct_of_peak_sustained_elapsedgpu__compute_memory_access_throughput.min.pct_of_peak_sustained_elapsedgpu__compute_memory_access_throughput.sum.pct_of_peak_sustained_elapsedgpu__compute_memory_access_throughput_internal_activity.avg.pct_of_peak_sustained_elapsedgpu__compute_memory_access_throughput_internal_activity.max.pct_of_peak_sustained_elapsedgpu__compute_memory_access_throughput_internal_activity.min.pct_of_peak_sustained_elapsedgpu__compute_memory_access_throughput_internal_activity.sum.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput.avg.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput.max.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput.min.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput.sum.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput_internal_activity.avg.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput_internal_activity.max.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput_internal_activity.min.pct_of_peak_sustained_elapsedgpu__compute_memory_request_throughput_internal_activity.sum.pct_of_peak_sustained_elapsedgpu__compute_memory_throughput.avg.pct_of_peak_sustained_elapsedgpu__compute_memory_throughput.max.pct_of_peak_sustained_elapsedgpu__compute_memory_throughput.min.pct_of_peak_sustained_elapsedgpu__compute_memory_throughput.sum.pct_of_peak_sustained_elapsedgpu__time_duration.avggpu__time_duration.maxgpu__time_duration.mingpu__time_duration.sumgr__workids_granted.avggr__workids_granted.maxgr__workids_granted.mingr__workids_granted.sumgr__workids_granted_as_ctas.avggr__workids_granted_as_ctas.maxgr__workids_granted_as_ctas.mingr__workids_granted_as_ctas.sumgr__workids_requested.avggr__workids_requested.maxgr__workids_requested.mingr__workids_requested.sumidc__request_cycles_active.avg.pct_of_peak_sustained_elapsedidc__request_cycles_active.max.pct_of_peak_sustained_elapsedidc__request_cycles_active.min.pct_of_peak_sustained_elapsedidc__request_cycles_active.sum.pct_of_peak_sustained_elapsedinst_executedl1tex__data_bank_reads.avg.pct_of_peak_sustained_elapsedl1tex__data_bank_reads.max.pct_of_peak_sustained_elapsedl1tex__data_bank_reads.min.pct_of_peak_sustained_elapsedl1tex__data_bank_reads.sum.pct_of_peak_sustained_elapsedl1tex__data_bank_writes.avg.pct_of_peak_sustained_elapsedl1tex__data_bank_writes.max.pct_of_peak_sustained_elapsedl1tex__data_bank_writes.min.pct_of_peak_sustained_elapsedl1tex__data_bank_writes.sum.pct_of_peak_sustained_elapsedl1tex__data_pipe_lsu_wavefronts.avg.pct_of_peak_sustained_elapsedl1tex__data_pipe_lsu_wavefronts.max.pct_of_peak_sustained_elapsedl1tex__data_pipe_lsu_wavefronts.min.pct_of_peak_sustained_elapsedl1tex__data_pipe_lsu_wavefronts.sum.pct_of_peak_sustained_elapsedl1tex__data_pipe_tex_wavefronts.avg.pct_of_peak_sustained_elapsedl1tex__data_pipe_tex_wavefronts.max.pct_of_peak_sustained_elapsedl1tex__data_pipe_tex_wavefronts.min.pct_of_peak_sustained_elapsedl1tex__data_pipe_tex_wavefronts.sum.pct_of_peak_sustained_elapsedl1tex__f_wavefronts.avg.pct_of_peak_sustained_elapsedl1tex__f_wavefronts.max.pct_of_peak_sustained_elapsedl1tex__f_wavefronts.min.pct_of_peak_sustained_elapsedl1tex__f_wavefronts.sum.pct_of_peak_sustained_elapsedl1tex__lsu_writeback_active.avg.pct_of_peak_sustained_elapsedl1tex__lsu_writeback_active.max.pct_of_peak_sustained_elapsedl1tex__lsu_writeback_active.min.pct_of_peak_sustained_elapsedl1tex__lsu_writeback_active.sum.pct_of_peak_sustained_elapsedl1tex__lsuin_requests.avg.pct_of_peak_sustained_elapsedl1tex__lsuin_requests.max.pct_of_peak_sustained_elapsedl1tex__lsuin_requests.min.pct_of_peak_sustained_elapsedl1tex__lsuin_requests.sum.pct_of_peak_sustained_elapsedl1tex__m_l1tex2xbar_req_cycles_active.avg.pct_of_peak_sustained_elapsedl1tex__m_l1tex2xbar_req_cycles_active.max.pct_of_peak_sustained_elapsedl1tex__m_l1tex2xbar_req_cycles_active.min.pct_of_peak_sustained_elapsedl1tex__m_l1tex2xbar_req_cycles_active.sum.pct_of_peak_sustained_elapsedl1tex__m_xbar2l1tex_read_sectors.avg.pct_of_peak_sustained_elapsedl1tex__m_xbar2l1tex_read_sectors.max.pct_of_peak_sustained_elapsedl1tex__m_xbar2l1tex_read_sectors.min.pct_of_peak_sustained_elapsedl1tex__m_xbar2l1tex_read_sectors.sum.pct_of_peak_sustained_elapsedl1tex__t_sector_hit_rate.pctl1tex__tex_writeback_active.avg.pct_of_peak_sustained_elapsedl1tex__tex_writeback_active.max.pct_of_peak_sustained_elapsedl1tex__tex_writeback_active.min.pct_of_peak_sustained_elapsedl1tex__tex_writeback_active.sum.pct_of_peak_sustained_elapsedl1tex__texin_sm2tex_req_cycles_active.avg.pct_of_peak_sustained_elapsedl1tex__texin_sm2tex_req_cycles_active.max.pct_of_peak_sustained_elapsedl1tex__texin_sm2tex_req_cycles_active.min.pct_of_peak_sustained_elapsedl1tex__texin_sm2tex_req_cycles_active.sum.pct_of_peak_sustained_elapsedl1tex__throughput.avg.pct_of_peak_sustained_activel1tex__throughput.max.pct_of_peak_sustained_activel1tex__throughput.min.pct_of_peak_sustained_activel1tex__throughput.sum.pct_of_peak_sustained_activelaunch__barrier_countlaunch__block_dim_xlaunch__block_dim_ylaunch__block_dim_zlaunch__block_sizelaunch__cluster_dim_xlaunch__cluster_dim_ylaunch__cluster_dim_zlaunch__cluster_max_activelaunch__cluster_max_potential_sizelaunch__cluster_scheduling_policylaunch__cluster_sizelaunch__context_idlaunch__device_idlaunch__func_cache_configlaunch__function_pcslaunch__grid_dim_xlaunch__grid_dim_ylaunch__grid_dim_zlaunch__grid_sizelaunch__kernel_namelaunch__occupancy_cluster_gpu_pctlaunch__occupancy_cluster_pctlaunch__occupancy_limit_barrierslaunch__occupancy_limit_blockslaunch__occupancy_limit_registerslaunch__occupancy_limit_shared_memlaunch__occupancy_limit_warpslaunch__occupancy_per_barrier_countlaunch__occupancy_per_block_sizelaunch__occupancy_per_cluster_sizelaunch__occupancy_per_register_countlaunch__occupancy_per_shared_mem_sizelaunch__persisting_l2_cache_sizelaunch__preferred_cluster_dim_xlaunch__preferred_cluster_dim_ylaunch__preferred_cluster_dim_zlaunch__preferred_cluster_sizelaunch__registers_per_threadlaunch__registers_per_thread_allocatedlaunch__shared_mem_config_sizelaunch__shared_mem_per_blocklaunch__shared_mem_per_block_allocatedlaunch__shared_mem_per_block_driverlaunch__shared_mem_per_block_dynamiclaunch__shared_mem_per_block_staticlaunch__sm_countlaunch__stack_sizelaunch__stream_idlaunch__thread_countlaunch__tpc_countlaunch__tpc_enabledlaunch__uses_cdplaunch__uses_green_contextlaunch__uses_mpslaunch__uses_nvlink_centric_schedulinglaunch__uses_vgpulaunch__waves_per_multiprocessorlts__average_gcomp_input_sector_success_rate.pctlts__average_gcomp_output_sector_compression_achieved_rate.ratiolts__d_atomic_input_cycles_active.avg.pct_of_peak_sustained_elapsedlts__d_atomic_input_cycles_active.max.pct_of_peak_sustained_elapsedlts__d_atomic_input_cycles_active.min.pct_of_peak_sustained_elapsedlts__d_atomic_input_cycles_active.sum.pct_of_peak_sustained_elapsedlts__d_decomp_input_sectors.avg.pct_of_peak_sustained_elapsedlts__d_decomp_input_sectors.max.pct_of_peak_sustained_elapsedlts__d_decomp_input_sectors.min.pct_of_peak_sustained_elapsedlts__d_decomp_input_sectors.sum.pct_of_peak_sustained_elapsedlts__d_sectors.avg.pct_of_peak_sustained_elapsedlts__d_sectors.max.pct_of_peak_sustained_elapsedlts__d_sectors.min.pct_of_peak_sustained_elapsedlts__d_sectors.sum.pct_of_peak_sustained_elapsedlts__gcomp_input_sectors.avglts__gcomp_input_sectors.maxlts__gcomp_input_sectors.minlts__gcomp_input_sectors.sumlts__lts2xbar_cycles_active.avg.pct_of_peak_sustained_elapsedlts__lts2xbar_cycles_active.max.pct_of_peak_sustained_elapsedlts__lts2xbar_cycles_active.min.pct_of_peak_sustained_elapsedlts__lts2xbar_cycles_active.sum.pct_of_peak_sustained_elapsedlts__t_sector_hit_rate.pctlts__t_sectors.avg.pct_of_peak_sustained_elapsedlts__t_sectors.max.pct_of_peak_sustained_elapsedlts__t_sectors.min.pct_of_peak_sustained_elapsedlts__t_sectors.sum.pct_of_peak_sustained_elapsedlts__t_tag_requests.avg.pct_of_peak_sustained_elapsedlts__t_tag_requests.max.pct_of_peak_sustained_elapsedlts__t_tag_requests.min.pct_of_peak_sustained_elapsedlts__t_tag_requests.sum.pct_of_peak_sustained_elapsedlts__throughput.avg.pct_of_peak_sustained_elapsedlts__throughput.max.pct_of_peak_sustained_elapsedlts__throughput.min.pct_of_peak_sustained_elapsedlts__throughput.sum.pct_of_peak_sustained_elapsedlts__xbar2lts_cycles_active.avg.pct_of_peak_sustained_elapsedlts__xbar2lts_cycles_active.max.pct_of_peak_sustained_elapsedlts__xbar2lts_cycles_active.min.pct_of_peak_sustained_elapsedlts__xbar2lts_cycles_active.sum.pct_of_peak_sustained_elapsednuma__cpu_affinitynuma__dev_display_name_allnuma__id_cpunuma__id_memorynvlink__bandwidthnvlink__count_logicalnvlink__count_physicalnvlink__destination_portsnvlink__dev0Idnvlink__dev0typenvlink__dev1Idnvlink__dev1typenvlink__dev_display_name_allnvlink__enabled_masknvlink__is_direct_linknvlink__is_nvswitch_connectednvlink__max_countnvlink__peer_accessnvlink__peer_atomicnvlink__source_portsnvlink__system_accessnvlink__system_atomicprofiler__perfworks_session_reuseprofiler__replayer_bytes_mem_accessible.avgprofiler__replayer_bytes_mem_accessible.maxprofiler__replayer_bytes_mem_accessible.minprofiler__replayer_bytes_mem_accessible.sumprofiler__replayer_bytes_mem_backed_up.avgprofiler__replayer_bytes_mem_backed_up.maxprofiler__replayer_bytes_mem_backed_up.minprofiler__replayer_bytes_mem_backed_up.sumprofiler__replayer_passesprofiler__replayer_passes_type_warmupsass__inst_executed_per_opcodesass__inst_executed_per_opcode_categorysass__inst_executed_per_opcode_with_modifier_allsass__inst_executed_per_opcode_with_modifier_selectivesass__inst_executed_register_spillingsass__inst_executed_register_spilling_mem_localsass__inst_executed_register_spilling_mem_sharedsass__inst_executed_register_spilling_op_readsass__inst_executed_register_spilling_op_writesass__thread_inst_executed_per_opcode_categorysass__thread_inst_executed_true_per_opcodesass__thread_inst_executed_true_per_opcode_with_modifier_allsass__thread_inst_executed_true_per_opcode_with_modifier_selectivesm__cycles_active.avgsm__cycles_active.maxsm__cycles_active.minsm__cycles_active.sumsm__inst_executed.avg.pct_of_peak_sustained_elapsedsm__inst_executed.max.pct_of_peak_sustained_elapsedsm__inst_executed.min.pct_of_peak_sustained_elapsedsm__inst_executed.sum.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_adu.avg.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_adu.max.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_adu.min.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_adu.sum.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_cbu_pred_on_any.avg.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_cbu_pred_on_any.max.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_cbu_pred_on_any.min.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_cbu_pred_on_any.sum.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_ipa.avg.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_ipa.max.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_ipa.min.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_ipa.sum.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_lsu.max.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_lsu.min.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_lsu.sum.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_tex.avg.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_tex.max.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_tex.min.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_tex.sum.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_uniform.avg.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_uniform.max.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_uniform.min.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_uniform.sum.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_xu.avg.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_xu.max.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_xu.min.pct_of_peak_sustained_elapsedsm__inst_executed_pipe_xu.sum.pct_of_peak_sustained_elapsedsm__instruction_throughput_internal_activity.avg.pct_of_peak_sustained_elapsedsm__instruction_throughput_internal_activity.max.pct_of_peak_sustained_elapsedsm__instruction_throughput_internal_activity.min.pct_of_peak_sustained_elapsedsm__instruction_throughput_internal_activity.sum.pct_of_peak_sustained_elapsedsm__issue_active.avg.pct_of_peak_sustained_elapsedsm__issue_active.max.pct_of_peak_sustained_elapsedsm__issue_active.min.pct_of_peak_sustained_elapsedsm__issue_active.sum.pct_of_peak_sustained_elapsedsm__maximum_warps_avg_per_active_cyclesm__maximum_warps_per_active_cycle_pctsm__memory_throughput.avg.pct_of_peak_sustained_elapsedsm__memory_throughput.max.pct_of_peak_sustained_elapsedsm__memory_throughput.min.pct_of_peak_sustained_elapsedsm__memory_throughput.sum.pct_of_peak_sustained_elapsedsm__memory_throughput_internal_activity.avg.pct_of_peak_sustained_elapsedsm__memory_throughput_internal_activity.max.pct_of_peak_sustained_elapsedsm__memory_throughput_internal_activity.min.pct_of_peak_sustained_elapsedsm__memory_throughput_internal_activity.sum.pct_of_peak_sustained_elapsedsm__mio2rf_writeback_active.avg.pct_of_peak_sustained_elapsedsm__mio2rf_writeback_active.max.pct_of_peak_sustained_elapsedsm__mio2rf_writeback_active.min.pct_of_peak_sustained_elapsedsm__mio2rf_writeback_active.sum.pct_of_peak_sustained_elapsedsm__mio_inst_issued.avg.pct_of_peak_sustained_elapsedsm__mio_inst_issued.max.pct_of_peak_sustained_elapsedsm__mio_inst_issued.min.pct_of_peak_sustained_elapsedsm__mio_inst_issued.sum.pct_of_peak_sustained_elapsedsm__mio_pq_read_cycles_active.avg.pct_of_peak_sustained_elapsedsm__mio_pq_read_cycles_active.max.pct_of_peak_sustained_elapsedsm__mio_pq_read_cycles_active.min.pct_of_peak_sustained_elapsedsm__mio_pq_read_cycles_active.sum.pct_of_peak_sustained_elapsedsm__mio_pq_write_cycles_active.avg.pct_of_peak_sustained_elapsedsm__mio_pq_write_cycles_active.max.pct_of_peak_sustained_elapsedsm__mio_pq_write_cycles_active.min.pct_of_peak_sustained_elapsedsm__mio_pq_write_cycles_active.sum.pct_of_peak_sustained_elapsedsm__pipe_aluheavy_cycles_active.avg.pct_of_peak_sustained_elapsedsm__pipe_aluheavy_cycles_active.max.pct_of_peak_sustained_elapsedsm__pipe_aluheavy_cycles_active.min.pct_of_peak_sustained_elapsedsm__pipe_aluheavy_cycles_active.sum.pct_of_peak_sustained_elapsedsm__pipe_fma_cycles_active.avg.pct_of_peak_sustained_elapsedsm__pipe_fma_cycles_active.max.pct_of_peak_sustained_elapsedsm__pipe_fma_cycles_active.min.pct_of_peak_sustained_elapsedsm__pipe_fma_cycles_active.sum.pct_of_peak_sustained_elapsedsm__pipe_fmaheavy_cycles_active.avg.pct_of_peak_sustained_elapsedsm__pipe_fmaheavy_cycles_active.max.pct_of_peak_sustained_elapsedsm__pipe_fmaheavy_cycles_active.min.pct_of_peak_sustained_elapsedsm__pipe_fmaheavy_cycles_active.sum.pct_of_peak_sustained_elapsedsm__pipe_fp64_cycles_active.avg.pct_of_peak_sustained_elapsedsm__pipe_fp64_cycles_active.max.pct_of_peak_sustained_elapsedsm__pipe_fp64_cycles_active.min.pct_of_peak_sustained_elapsedsm__pipe_fp64_cycles_active.sum.pct_of_peak_sustained_elapsedsm__pipe_tensor_cycles_active.avg.pct_of_peak_sustained_elapsedsm__pipe_tensor_cycles_active.max.pct_of_peak_sustained_elapsedsm__pipe_tensor_cycles_active.min.pct_of_peak_sustained_elapsedsm__pipe_tensor_cycles_active.sum.pct_of_peak_sustained_elapsedsm__pipe_tma_cycles_active.avg.pct_of_peak_sustained_elapsedsm__pipe_tma_cycles_active.max.pct_of_peak_sustained_elapsedsm__pipe_tma_cycles_active.min.pct_of_peak_sustained_elapsedsm__pipe_tma_cycles_active.sum.pct_of_peak_sustained_elapsedsm__throughput.avg.pct_of_peak_sustained_elapsedsm__throughput.max.pct_of_peak_sustained_elapsedsm__throughput.min.pct_of_peak_sustained_elapsedsm__throughput.sum.pct_of_peak_sustained_elapsedsm__warps_active.avg.pct_of_peak_sustained_activesm__warps_active.avg.per_cycle_activesm__warps_active.max.pct_of_peak_sustained_activesm__warps_active.max.per_cycle_activesm__warps_active.min.pct_of_peak_sustained_activesm__warps_active.min.per_cycle_activesm__warps_active.sum.pct_of_peak_sustained_activesm__warps_active.sum.per_cycle_activesmsp__average_warp_latency_per_inst_issued.ratiosmsp__average_warps_active_per_inst_executed.ratiosmsp__average_warps_issue_stalled_barrier_per_issue_active.ratiosmsp__average_warps_issue_stalled_branch_resolving_per_issue_active.ratiosmsp__average_warps_issue_stalled_dispatch_stall_per_issue_active.ratiosmsp__average_warps_issue_stalled_drain_per_issue_active.ratiosmsp__average_warps_issue_stalled_lg_throttle_per_issue_active.ratiosmsp__average_warps_issue_stalled_long_scoreboard_per_issue_active.ratiosmsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratiosmsp__average_warps_issue_stalled_membar_per_issue_active.ratiosmsp__average_warps_issue_stalled_mio_throttle_per_issue_active.ratiosmsp__average_warps_issue_stalled_misc_per_issue_active.ratiosmsp__average_warps_issue_stalled_no_instruction_per_issue_active.ratiosmsp__average_warps_issue_stalled_not_selected_per_issue_active.ratiosmsp__average_warps_issue_stalled_selected_per_issue_active.ratiosmsp__average_warps_issue_stalled_short_scoreboard_per_issue_active.ratiosmsp__average_warps_issue_stalled_sleeping_per_issue_active.ratiosmsp__average_warps_issue_stalled_tex_throttle_per_issue_active.ratiosmsp__average_warps_issue_stalled_wait_per_issue_active.ratiosmsp__issue_active.avg.pct_of_peak_sustained_activesmsp__issue_active.avg.per_cycle_activesmsp__issue_active.max.pct_of_peak_sustained_activesmsp__issue_active.max.per_cycle_activesmsp__issue_active.min.pct_of_peak_sustained_activesmsp__issue_active.min.per_cycle_activesmsp__issue_active.sum.pct_of_peak_sustained_activesmsp__issue_active.sum.per_cycle_activesmsp__issue_inst0.avg.pct_of_peak_sustained_activesmsp__issue_inst0.max.pct_of_peak_sustained_activesmsp__issue_inst0.min.pct_of_peak_sustained_activesmsp__issue_inst0.sum.pct_of_peak_sustained_activesmsp__maximum_warps_avg_per_active_cyclesmsp__thread_inst_executed_per_inst_executed.ratiosmsp__thread_inst_executed_pred_on_per_inst_executed.ratiosmsp__warps_active.avg.peak_sustainedsmsp__warps_active.avg.per_cycle_activesmsp__warps_active.max.peak_sustainedsmsp__warps_active.max.per_cycle_activesmsp__warps_active.min.peak_sustainedsmsp__warps_active.min.per_cycle_activesmsp__warps_active.sum.peak_sustainedsmsp__warps_active.sum.per_cycle_activesmsp__warps_eligible.avg.per_cycle_activesmsp__warps_eligible.max.per_cycle_activesmsp__warps_eligible.min.per_cycle_activesmsp__warps_eligible.sum.per_cycle_active
2%%%/register%/bytecyclehzcyclehzcyclehzcyclehz%%%%%%%%%%%%%%%%%%%%nsnsnsnsblockblockblockblock%%%%inst%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%blockblockblockclusterblock%%blockblockblockblockblockbyteregister/threadregister/threadbytebyte/blockbyte/blockbyte/blockbyte/blockbyte/blockSMthread%%%%%%%%%%%%%sectorsectorsectorsector%%%%%%%%%%%%%%%%%%%%%bytebytebytebytebytebytebytebytepasspassinstbytebyteinstinstcyclecyclecyclecycle%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%warp%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%warp%warp%warp%warpcyclecycleinstinstinstinstinstinstinstinstinstinstinstinstinstinstinstinstinst%%%%%%%%warpwarpwarpwarpwarpwarpwarpwarpwarpwarpwarpwarpwarp
3060python3.12127.0.0.1cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu17(384, 1, 1)(296, 168, 1)012.1010no data62531642254975300004321011101010443241800011210111No-CC11100NVIDIA GB10002564112561001001772935902117943200016042500161614318010010144311125165824242424111342177281024102464242418000214748364765535655354185330001887436821474836476553665536255491521013761024001024153648122048327683276865536327682048327681310721638416384163842046327683276820483276826843545632768131072327683276865536327682048327686500020971201310723276832768131072163843276816384819216384819220463276832768485330001100004864101110151010241264115120151232655361306617692161113256409993.252138683834.25564224072139154479.34563733862137295935.952256399738554735336.9977.1077.1377.0677.1038.7338.7738.6838.7373.9874.1873.7773.98000077.1077.1377.0677.10263760322637603226376032263760324968049680496804968049680496804968049680497284972849728497280.400.410.400.40192040449516.9017.0616.7516.905.795.855.745.7941.2741.6740.9141.270000000037.4037.6537.1837.4026.4126.6226.2126.419.009.068.949.0044.4344.8244.1344.439000000.110.110.110.1143.2343.6042.9343.2383841138400008PolicySpread010CachePreferNone129616814972800324114300157020282388471859200001681681024008908889088102488064048102471909555224all000001036000000000042.7442.7642.7242.74000073.9874.1873.7773.9889.1177.1077.1377.0677.1038.5538.5738.5338.5577.1077.1377.0677.1034.0834.1234.0534.081111000000001000000000074145902827414590282741459028212604803479474145902827414590282741459028212604803479417019204044951920404495192040449519204044950no datano data00000057979728.965802719957917994278302699017.9218.0517.7917.924.564.584.554.560.660.660.650.66000026.4126.6426.1826.4100003.203.223.183.200000000018.0318.1917.9018.03122526.4126.6426.1826.410.120.120.120.129.409.499.339.4010.4510.5310.3710.453.863.893.823.863.863.893.823.866.916.986.856.910.850.860.850.852.772.802.752.77000078.9979.6078.3878.990.550.560.550.5578.9979.6078.3878.9920.8310.0020.8510.0120.819.9920.83480.0013.6713.750.230.160.040.0000.613.8200.070.020.080.311.000.083.9303.7818.290.1820.050.2017.480.1718.2935.1281.7180.0182.4281.71331.2425.64122.50123.00122.002304480.000.240.260.2345.95
4160python3.12127.0.0.1void qk_int_sv_f8_attn_kernel<128, 64, 32, 64, 128, 1, 2, 2, float, 1, __nv_bfloat16, 1, 0, 0, 1, 0, 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)17(32, 4, 1)(296, 56, 1)012.1011458688no data30420256332400152004321011101010443241800011210111No-CC11100NVIDIA GB1000256411256100100177293590211794320001604250016161431801001014431112516582424242411134217728102410246424241800021474836476553565535418533000188743682147483647655366553625549152101376102400102415364812204832768327686553632768204832768131072163841638416384204632768327682048327682684354563276813107232768327686553632768204832768650002097120131072327683276813107216384327681638481921638481922046327683276848533000110000486410111015101024126411512015123265536130661769216111325535217702137182697.805535999512137484559.615532885322136282151.0822140870808548730791.2131.5931.6031.5831.5923.6723.6723.6623.6731.4631.4731.4631.46000031.5931.6031.5831.592589960002589960002589960002589960000000000000000.000.000.000.003832238440811.8211.8411.7811.822.382.392.372.3826.6626.7126.5526.660000000025.1025.1525.0125.1019.5519.5919.4719.557.177.187.147.1718.9218.9518.8418.920.8000000.000.000.000.0026.8926.9426.7826.891324112800008PolicySpread010CachePreferNone129656116576002424231215210102732120047185920000255256102400337923379210243276804810247212172824all00000172.67000000000015.9815.9915.9815.98000031.4631.4731.4631.4698.8431.5931.6031.5831.5923.6523.6623.6523.6531.5931.6031.5831.5923.8423.8523.8323.8411110000000010000000000741459028274145902827414590282126048034794741459028274145902827414590282126048034794170383223844083832238440838322384408383223844081458688no datano data000000548783193.985497108805462699182634159331136.0636.1335.9236.061.181.191.181.180.000.000.000.00000019.5519.5919.4719.5500000.220.220.220.2220.0620.1019.9820.06000036.0636.1335.9236.06816.6719.5519.5919.4719.550.000.000.000.006.286.296.256.286.916.926.886.911.221.221.211.221.221.221.211.2212.9412.9712.8912.9415.0015.0314.9415.0019.2319.2719.1619.23000075.5175.6675.2275.51000075.5175.6675.2275.5116.657.9916.698.0116.597.9616.65383.645.485.480.190.000.160.000.000.141.2400.120.080.030.2610.25002.0136.490.3636.570.3736.350.3636.4970.0763.5163.6363.2163.5123231.80122.00122.00121.992304383.820.460.460.4688.19
5260python3.12127.0.0.1cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu17(384, 1, 1)(296, 42, 1)012.1010no data62531642254975300004321011101010443241800011210111No-CC11100NVIDIA GB10002564112561001001772935902117943200016042500161614318010010144311125165824242424111342177281024102464242418000214748364765535655354185330001887436821474836476553665536255491521013761024001024153648122048327683276865536327682048327681310721638416384163842046327683276820483276826843545632768131072327683276865536327682048327686500020971201310723276832768131072163843276816384819216384819220463276832768485330001100004864101110151010241264115120151232655361306617692161113218139065.502138750195.73181425562139161755.38181288462137545229.70725562628555000782.9179.1079.1679.0579.1039.7139.7839.6539.7176.6776.9076.4976.67000079.1079.1679.0579.1084811528481152848115284811521238412384123841238412384123841238412384124321243212432124320.340.340.330.3462283074717.4617.6617.3217.465.946.015.895.9442.4142.9042.0842.410000000038.7539.0538.4538.7527.1027.3026.8927.1011.0811.1610.9911.0846.0646.4145.7046.069000000.080.090.080.0842.6743.0042.3442.6783841138400008PolicySpread010CachePreferNone1296421124320032411430015702028238847185920000168168102400890888908810248806404810247477388824all00000259000000000043.5343.5743.5043.53000076.6776.9076.4976.6789.8879.1079.1679.0579.1039.5539.5939.5339.5579.1079.1679.0579.1041.2141.2741.1941.21111100000000100000000007414590282741459028274145902821260480347947414590282741459028274145902821260480347941706228307476228307476228307476228307470no datano data00000019579361.17196157581953821593980933618.0918.2217.9518.094.554.574.524.550.610.620.600.61000027.1027.3026.8927.1000003.343.373.323.340000000018.1918.3918.0618.19122527.1027.3026.8927.100.090.090.090.099.739.849.669.7310.6710.7510.5910.673.883.923.843.883.883.923.843.886.796.876.746.790.800.810.790.802.522.542.502.52000081.8882.5181.2581.880.550.550.540.5581.8882.5181.2581.8820.8310.0020.8710.0220.799.9820.83480.0014.2814.360.220.160.040.0000.603.9000.060.020.070.311.000.074.0803.8517.510.1819.180.1916.760.1717.5133.6282.4981.0683.0682.49331.2325.41122.50123.01122.002304480.000.230.250.2243.96
6360python3.12127.0.0.1cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu17(384, 1, 1)(296, 224, 1)012.1010no data62531642254975300004321011101010443241800011210111No-CC11100NVIDIA GB100025641125610010017729359021179432000160425001616143180100101443111251658242424241113421772810241024642424180002147483647655356553541853300018874368214748364765536655362554915210137610240010241536481220483276832768655363276820483276813107216384163841638420463276832768204832768268435456327681310723276832768655363276820483276865000209712013107232768327681310721638432768163848192163848192204632768327684853300011000048641011101510102412641151201512326553613066176921611132773039522137085387.93773229412137610343.16772472202135517018.843092158088548341551.7174.9975.0274.9674.9937.6637.6937.6137.6671.9572.1371.7771.95000074.9975.0274.9674.99361726083617260836172608361726086625666256662566625666256662566625666256663046630466304663040.390.400.390.39256059208316.4416.5616.3416.445.635.675.605.6340.1540.4339.9240.150000000036.3936.7236.0936.3925.7025.9225.5125.708.758.838.698.7543.2343.5942.8743.239000000.110.110.100.1143.3043.6642.9443.3083841138400008PolicySpread010CachePreferNone129622416630400324114300157020282388471859200001681681024008908889088102488064048102472546073624all000001381.33000000000041.5741.5941.5541.57000071.9572.1371.7771.9589.1174.9975.0274.9674.9937.5037.5237.4837.5074.9975.0274.9674.9933.1433.1733.1233.141111000000001000000000074145902827414590282741459028212604803479474145902827414590282741459028212604803479417025605920832560592083256059208325605920830no datano data00000077183115.257722426077101423370478953217.4417.5817.3117.444.444.464.434.440.640.640.630.64000025.7025.9025.4525.7000003.123.143.093.120000000017.5417.6617.3917.54122525.7025.9025.4525.700.120.120.120.129.159.229.069.1510.1710.2410.0810.173.763.783.733.763.763.783.733.766.726.776.686.720.830.840.830.832.692.712.672.69000076.8577.3976.2276.850.540.540.530.5476.8577.3976.2276.8520.8310.0020.8410.0120.819.9920.83480.0014.3814.460.240.160.040.0000.613.8200.070.020.080.311.000.084.4103.7817.390.1719.030.1916.600.1717.3933.3882.6181.0283.2982.61331.2525.64122.50123.00122.002304480.000.230.250.2243.68
7460python3.12127.0.0.1cutlass3x_sm120_bstensorop_s16864gemm_block_scaled_ue4m3xe2m1_ue4m3xe2m1_f32_bf16_bf16_128x128x256_1x1x1_0_tnn_align32_o_vs16_bias_bf16_relu17(384, 1, 1)(296, 42, 1)012.1010no data62531642254975300004321011101010443241800011210111No-CC11100NVIDIA GB10002564112561001001772935902117943200016042500161614318010010144311125165824242424111342177281024102464242418000214748364765535655354185330001887436821474836476553665536255491521013761024001024153648122048327683276865536327682048327681310721638416384163842046327683276820483276826843545632768131072327683276865536327682048327686500020971201310723276832768131072163843276816384819216384819220463276832768485330001100004864101110151010241264115120151232655361306617692161113234588907.252135252700.54345972162135765618.77345643472133736539.891383556298541010802.1681.5881.6181.5681.5840.9541.0040.9040.9580.3080.4580.1580.30000081.5881.6181.5681.58161989761619897616198976161989761238412384123841238412384123841238412384124321243212432124320.220.220.220.22119293765418.2118.3518.0718.216.136.186.096.1343.9744.3143.6343.970000000040.6240.9340.3040.6227.9628.2927.6427.9612.1812.2812.0712.1848.3148.6847.9348.319000000.040.040.040.0446.0946.4545.7446.0983841138400008PolicySpread010CachePreferNone1296421124320032411430015702028238847185920000168168102400890888908810248806404810247477388824all00000259000000000044.4144.4244.4044.41000080.3080.4580.1580.3091.0781.5881.6181.5681.5840.8040.8240.7940.8081.5881.6181.5681.5840.5840.6240.5640.581111000000001000000000074145902827414590282741459028212604803479474145902827414590282741459028212604803479417011929376541192937654119293765411929376540no datano data00000036250542.503631725436167629174002604018.1818.3817.9818.184.454.474.424.450.550.550.540.55000027.9628.1827.7527.9600003.543.573.513.540000000018.2918.4918.1618.29122527.9628.1827.7527.960.050.050.050.0510.1810.3010.1010.1810.9211.0110.8410.923.903.953.863.903.903.953.863.906.536.586.486.530.700.710.700.702.102.122.082.10000085.8886.5485.2185.880.540.550.530.5485.8886.5485.2185.8820.8310.0020.8710.0220.799.9820.83480.0014.3414.430.170.160.040.0000.604.0600.050.010.070.311.000.053.8103.9817.440.1719.040.1916.720.1717.4433.4882.5681.1583.0582.56331.2225.05122.50123.01122.002304480.000.230.250.2243.74