68 lines
1.3 KiB
JSON
68 lines
1.3 KiB
JSON
|
{
|
||
|
|
"device": "NVIDIA GB10",
|
||
|
|
"cutlass_dsl": "4.6.2",
|
||
|
|
"metadata": {
|
||
|
|
"width": 1344,
|
||
|
|
"height": 768,
|
||
|
|
"frames": 124,
|
||
|
|
"steps": 12,
|
||
|
|
"sampler_step": 1,
|
||
|
|
"seed": 440420,
|
||
|
|
"text_tokens": 100,
|
||
|
|
"hidden_shape": [
|
||
|
|
37810,
|
||
|
|
5376
|
||
|
|
],
|
||
|
|
"segments": [
|
||
|
|
[
|
||
|
|
0,
|
||
|
|
100,
|
||
|
|
1
|
||
|
|
],
|
||
|
|
[
|
||
|
|
100,
|
||
|
|
514,
|
||
|
|
2
|
||
|
|
],
|
||
|
|
[
|
||
|
|
514,
|
||
|
|
37810,
|
||
|
|
0
|
||
|
|
]
|
||
|
|
]
|
||
|
|
},
|
||
|
|
"linear": "attn_out_proj",
|
||
|
|
"mnk": [
|
||
|
|
128,
|
||
|
|
5376,
|
||
|
|
7168
|
||
|
|
],
|
||
|
|
"ring": {
|
||
|
|
"row_capacity": 128,
|
||
|
|
"producer": "vortex_native_quantize_nvfp4",
|
||
|
|
"qdata_bytes": 458752,
|
||
|
|
"sfa_bytes": 57344,
|
||
|
|
"logical_bytes": 516096
|
||
|
|
},
|
||
|
|
"parity": {
|
||
|
|
"tensor_scale_equal": true,
|
||
|
|
"fp4_difference_count": 0,
|
||
|
|
"block_scale_difference_count": 0,
|
||
|
|
"output_equal": true,
|
||
|
|
"max_abs": 0.0,
|
||
|
|
"mean_abs": 0.0
|
||
|
|
},
|
||
|
|
"timing": {
|
||
|
|
"warmup": 10,
|
||
|
|
"iterations": 100,
|
||
|
|
"producer_ms": 0.010279680490493775,
|
||
|
|
"global_scale_ms": 0.01211359977722168,
|
||
|
|
"cute_consumer_ms": 0.03870527982711792,
|
||
|
|
"modeled_ring_ms": 0.06109856009483337,
|
||
|
|
"actual_into_ring_cute_gemm_ms": 0.09652607917785644,
|
||
|
|
"reference_vortex_scale_comfy_pack_gemm_ms": 0.10898783683776855,
|
||
|
|
"modeled_ring_vs_reference": 0.5605998051487184,
|
||
|
|
"actual_ring_vs_reference": 0.8856591889381034
|
||
|
|
}
|
||
|
|
}
|