h3-blackwell-runtime/benchmarks/gb10-post-optimization-block24-profile.json

1402 lines
43 KiB
JSON
Raw Permalink Normal View History

{
"prompt": "A playful orange tabby cat starts in an ordinary cozy living room in a normal house, afternoon light, sofa and rug. The cat crouches, jumps, and does one clean athletic backflip in slow motion. As the backflip completes there is a sharp cinematic cut: the cat lands perfectly on a glowing neon disco dance floor wearing oversized black sunglasses. Mirror ball reflections, colorful lights, joyful party energy, stylish and funny, clear before-and-after transformation.",
"model_path": "/models/minimax_h3_fl2va_pruned_nvfp4.safetensors",
"width": 1344,
"height": 768,
"frames": 124,
"steps": 12,
"sampler_step": 1,
"seed": 440420,
"text_tokens": 100,
"block_index": 24,
"attention": "sage2",
"warmup": 3,
"iterations": 10,
"hidden_shape": [
37810,
5376
],
"output_shape": [
37810,
5376
],
"segments": [
[
0,
100,
1
],
[
100,
514,
2
],
[
514,
37810,
0
]
],
"timings": {
"norm1": {
"count": 10,
"mean_s": 0.003506062604719773,
"p50_s": 0.0034730469924397767,
"p90_s": 0.0035981596854981037,
"p95_s": 0.0036302793218055737,
"p99_s": 0.0036559750308515506,
"min_s": 0.0034213270409964025,
"max_s": 0.0036623989581130445
},
"modulate_msa": {
"count": 10,
"mean_s": 0.01051678290241398,
"p50_s": 0.010468507505720481,
"p90_s": 0.01089892717427574,
"p95_s": 0.010943725591641851,
"p99_s": 0.010979564325534739,
"min_s": 0.010146012005861849,
"max_s": 0.01098852400900796
},
"linear.attn_qkv_proj.flatten_contiguous": {
"count": 10,
"mean_s": 4.558393266052008e-06,
"p50_s": 4.208006430417299e-06,
"p90_s": 4.755187546834348e-06,
"p95_s": 6.281610694713887e-06,
"p99_s": 7.502749213017523e-06,
"min_s": 4.031986463814974e-06,
"max_s": 7.808033842593431e-06
},
"linear.attn_qkv_proj.pre_quant_scale": {
"count": 10,
"mean_s": 0.0,
"p50_s": 0.0,
"p90_s": 0.0,
"p95_s": 0.0,
"p99_s": 0.0,
"min_s": 0.0,
"max_s": 0.0
},
"linear.attn_qkv_proj.packed_weight_wrapper": {
"count": 10,
"mean_s": 9.913597023114561e-06,
"p50_s": 7.968017598614097e-06,
"p90_s": 1.0451191337779158e-05,
"p95_s": 1.924959651660172e-05,
"p99_s": 2.6288320659659805e-05,
"min_s": 7.4239796958863735e-06,
"max_s": 2.804800169542432e-05
},
"linear.attn_qkv_proj.bias_cast": {
"count": 10,
"mean_s": 3.6815996281802655e-06,
"p50_s": 3.5040138754993677e-06,
"p90_s": 3.956805448979139e-06,
"p95_s": 4.554382758215069e-06,
"p99_s": 5.032444605603814e-06,
"min_s": 3.375986125320196e-06,
"max_s": 5.151960067451e-06
},
"linear.attn_qkv_proj.activation_scale": {
"count": 10,
"mean_s": 0.0017137959017418324,
"p50_s": 0.0016890474944375455,
"p90_s": 0.0017734934226609766,
"p95_s": 0.0017952662019524723,
"p99_s": 0.001812684425385669,
"min_s": 0.0016727829934097826,
"max_s": 0.001817038981243968
},
"linear.attn_qkv_proj.scale_to_device": {
"count": 10,
"mean_s": 1.1468783486634494e-05,
"p50_s": 9.303970728069544e-06,
"p90_s": 1.7495953943580386e-05,
"p95_s": 1.7675955314189194e-05,
"p99_s": 1.781995641067624e-05,
"min_s": 8.32002842798829e-06,
"max_s": 1.7855956684798002e-05
},
"linear.attn_qkv_proj.activation_quant_pack": {
"count": 10,
"mean_s": 0.0022020551143214105,
"p50_s": 0.002172503009205684,
"p90_s": 0.002327527030138299,
"p95_s": 0.002348515027551912,
"p99_s": 0.002365305425482802,
"min_s": 0.0021115349954925478,
"max_s": 0.0023695030249655247
},
"linear.attn_qkv_proj.activation_quant_wrap": {
"count": 10,
"mean_s": 6.527994992211461e-06,
"p50_s": 6.503978511318564e-06,
"p90_s": 6.854429375380277e-06,
"p95_s": 7.243192521855234e-06,
"p99_s": 7.554203039035201e-06,
"min_s": 6.096030119806528e-06,
"max_s": 7.631955668330193e-06
},
"linear.attn_qkv_proj.gemm": {
"count": 10,
"mean_s": 0.02555141010088846,
"p50_s": 0.025579269480658695,
"p90_s": 0.026145932916551828,
"p95_s": 0.026412224461091684,
"p99_s": 0.026625257696723564,
"min_s": 0.024755333026405424,
"max_s": 0.026678516005631536
},
"linear.attn_qkv_proj.slice_reshape": {
"count": 10,
"mean_s": 5.6432094424963e-06,
"p50_s": 5.471985787153244e-06,
"p90_s": 5.9520069044083355e-06,
"p95_s": 6.744018173776566e-06,
"p99_s": 7.377627189271153e-06,
"min_s": 5.2480027079582214e-06,
"max_s": 7.536029443144798e-06
},
"attn_qkv_proj": {
"count": 10,
"mean_s": 0.029553097899770363,
"p50_s": 0.029554627515608445,
"p90_s": 0.03014536619302817,
"p95_s": 0.03036433259549085,
"p99_s": 0.030539505717461,
"min_s": 0.028794804005883634,
"max_s": 0.030583298997953534
},
"attn_qkv_split_view": {
"count": 10,
"mean_s": 1.2614397564902902e-05,
"p50_s": 1.1160009307786822e-05,
"p90_s": 1.3886386295780535e-05,
"p95_s": 1.877520990092306e-05,
"p99_s": 2.2686268785037104e-05,
"min_s": 1.0351999662816525e-05,
"max_s": 2.3664033506065607e-05
},
"attn_qk_rms_rope": {
"count": 10,
"mean_s": 0.012204991799080744,
"p50_s": 0.012200642988318577,
"p90_s": 0.01241076691658236,
"p95_s": 0.012442684968118557,
"p99_s": 0.012468219409347512,
"min_s": 0.011841004015877843,
"max_s": 0.01247460301965475
},
"attention_kernel": {
"count": 10,
"mean_s": 0.26835116609581744,
"p50_s": 0.25847244350006804,
"p90_s": 0.26903588619315993,
"p95_s": 0.31390352409216565,
"p99_s": 0.3497976344113704,
"min_s": 0.25696170097216964,
"max_s": 0.35877116199117154
},
"attn_output_reshape": {
"count": 10,
"mean_s": 4.254397936165333e-06,
"p50_s": 4.191999323666096e-06,
"p90_s": 4.472001455724239e-06,
"p95_s": 4.507991252467037e-06,
"p99_s": 4.5367830898612736e-06,
"min_s": 4.016037564724684e-06,
"max_s": 4.543981049209833e-06
},
"linear.attn_out_proj.flatten_contiguous": {
"count": 10,
"mean_s": 4.1199964471161366e-06,
"p50_s": 4.063971573486924e-06,
"p90_s": 4.222447751089931e-06,
"p95_s": 4.431235720403492e-06,
"p99_s": 4.598266095854342e-06,
"min_s": 3.983965143561363e-06,
"max_s": 4.640023689717054e-06
},
"linear.attn_out_proj.pre_quant_scale": {
"count": 10,
"mean_s": 0.0,
"p50_s": 0.0,
"p90_s": 0.0,
"p95_s": 0.0,
"p99_s": 0.0,
"min_s": 0.0,
"max_s": 0.0
},
"linear.attn_out_proj.packed_weight_wrapper": {
"count": 10,
"mean_s": 9.252689778804779e-06,
"p50_s": 8.911971235647798e-06,
"p90_s": 9.937619324773548e-06,
"p95_s": 1.0808813385665413e-05,
"p99_s": 1.150576863437891e-05,
"min_s": 8.671951945871115e-06,
"max_s": 1.1680007446557283e-05
},
"linear.attn_out_proj.bias_cast": {
"count": 10,
"mean_s": 3.4816039260476828e-06,
"p50_s": 3.4720287658274174e-06,
"p90_s": 3.5679608117789033e-06,
"p95_s": 3.5679870052263142e-06,
"p99_s": 3.5680079599842427e-06,
"min_s": 3.3919932320713997e-06,
"max_s": 3.568013198673725e-06
},
"linear.attn_out_proj.activation_scale": {
"count": 10,
"mean_s": 0.0022514488082379103,
"p50_s": 0.002237695502117276,
"p90_s": 0.0022609557956457136,
"p95_s": 0.0023159494099672883,
"p99_s": 0.0023599443014245478,
"min_s": 0.0022299509728327394,
"max_s": 0.002370943024288863
},
"linear.attn_out_proj.scale_to_device": {
"count": 10,
"mean_s": 1.2071995297446846e-05,
"p50_s": 1.1911994079127908e-05,
"p90_s": 1.2502394383773208e-05,
"p95_s": 1.3395198038779196e-05,
"p99_s": 1.4109440962783993e-05,
"min_s": 1.1135998647660017e-05,
"max_s": 1.428800169378519e-05
},
"linear.attn_out_proj.activation_quant_pack": {
"count": 10,
"mean_s": 0.002843859599670395,
"p50_s": 0.0028387825295794755,
"p90_s": 0.0028701974020805213,
"p95_s": 0.002917386181070469,
"p99_s": 0.002955137204262428,
"min_s": 0.0028025589999742806,
"max_s": 0.0029645749600604177
},
"linear.attn_out_proj.activation_quant_wrap": {
"count": 10,
"mean_s": 6.420805584639311e-06,
"p50_s": 6.4480118453502655e-06,
"p90_s": 6.592011777684093e-06,
"p95_s": 6.664017564617097e-06,
"p99_s": 6.721622194163501e-06,
"min_s": 6.0800230130553246e-06,
"max_s": 6.736023351550102e-06
},
"linear.attn_out_proj.gemm": {
"count": 10,
"mean_s": 0.007799276692094281,
"p50_s": 0.007773573015583679,
"p90_s": 0.007835476868785917,
"p95_s": 0.00800838442519307,
"p99_s": 0.008146710470318794,
"min_s": 0.007710108999162912,
"max_s": 0.008181291981600225
},
"linear.attn_out_proj.slice_reshape": {
"count": 10,
"mean_s": 5.4480042308568954e-06,
"p50_s": 5.416019121184945e-06,
"p90_s": 5.580816650763155e-06,
"p95_s": 5.710395635105669e-06,
"p99_s": 5.814058822579682e-06,
"min_s": 5.296024028211832e-06,
"max_s": 5.839974619448185e-06
},
"attn_out_proj": {
"count": 10,
"mean_s": 0.012975754507351667,
"p50_s": 0.012935250502778217,
"p90_s": 0.013021398818818853,
"p95_s": 0.01331132840423379,
"p99_s": 0.013543272072565742,
"min_s": 0.012829803046770394,
"max_s": 0.01360125798964873
},
"gate_msa": {
"count": 10,
"mean_s": 0.008599226904334501,
"p50_s": 0.008538733003661036,
"p90_s": 0.0088796345226001,
"p95_s": 0.008909831769415177,
"p99_s": 0.008933989566867239,
"min_s": 0.00840773998061195,
"max_s": 0.008940029016230255
},
"norm2": {
"count": 10,
"mean_s": 0.0036033073149155824,
"p50_s": 0.0035515105118975043,
"p90_s": 0.0037464877241291106,
"p95_s": 0.003796427365159616,
"p99_s": 0.0038363790779840203,
"min_s": 0.0035186070017516613,
"max_s": 0.003846367006190121
},
"modulate_mlp": {
"count": 10,
"mean_s": 0.010574622795684264,
"p50_s": 0.010596931009786204,
"p90_s": 0.01087350157322362,
"p95_s": 0.010878188785864041,
"p99_s": 0.010881938555976375,
"min_s": 0.010094380006194115,
"max_s": 0.01088287599850446
},
"linear.mlp_fc1.flatten_contiguous": {
"count": 10,
"mean_s": 4.360004095360637e-06,
"p50_s": 4.327972419559956e-06,
"p90_s": 4.619208630174398e-06,
"p95_s": 4.669604822993279e-06,
"p99_s": 4.709921777248382e-06,
"min_s": 4.128029104322195e-06,
"max_s": 4.720001015812159e-06
},
"linear.mlp_fc1.pre_quant_scale": {
"count": 10,
"mean_s": 0.0,
"p50_s": 0.0,
"p90_s": 0.0,
"p95_s": 0.0,
"p99_s": 0.0,
"min_s": 0.0,
"max_s": 0.0
},
"linear.mlp_fc1.packed_weight_wrapper": {
"count": 10,
"mean_s": 8.18080734461546e-06,
"p50_s": 8.128001354634762e-06,
"p90_s": 8.4448023699224e-06,
"p95_s": 8.718419121578335e-06,
"p99_s": 8.937312522903086e-06,
"min_s": 7.760012522339821e-06,
"max_s": 8.992035873234272e-06
},
"linear.mlp_fc1.bias_cast": {
"count": 10,
"mean_s": 3.4687924198806287e-06,
"p50_s": 3.455992555245757e-06,
"p90_s": 3.5727629438042644e-06,
"p95_s": 3.5943987313658e-06,
"p99_s": 3.6117073614150284e-06,
"min_s": 3.311957698315382e-06,
"max_s": 3.6160345189273357e-06
},
"linear.mlp_fc1.activation_scale": {
"count": 10,
"mean_s": 0.0017569320043548942,
"p50_s": 0.0017358949990011752,
"p90_s": 0.0018160303065087645,
"p95_s": 0.0018329426558921114,
"p99_s": 0.001846472535398789,
"min_s": 0.0017267030198127031,
"max_s": 0.001849855005275458
},
"linear.mlp_fc1.scale_to_device": {
"count": 10,
"mean_s": 1.0102399392053485e-05,
"p50_s": 9.367999155074358e-06,
"p90_s": 1.0531197767704722e-05,
"p95_s": 1.4289590762928117e-05,
"p99_s": 1.729630515910685e-05,
"min_s": 8.5439532995224e-06,
"max_s": 1.804798375815153e-05
},
"linear.mlp_fc1.activation_quant_pack": {
"count": 10,
"mean_s": 0.00217321349773556,
"p50_s": 0.0021457590046338737,
"p90_s": 0.0022131046222057193,
"p95_s": 0.00231779980531428,
"p99_s": 0.0024015559518011285,
"min_s": 0.0021100639714859426,
"max_s": 0.002422494988422841
},
"linear.mlp_fc1.activation_quant_wrap": {
"count": 10,
"mean_s": 6.355205550789833e-06,
"p50_s": 6.3680054154247046e-06,
"p90_s": 6.510427920147777e-06,
"p95_s": 6.5752043155953285e-06,
"p99_s": 6.62702543195337e-06,
"min_s": 6.128044333308935e-06,
"max_s": 6.639980711042881e-06
},
"linear.mlp_fc1.gemm": {
"count": 10,
"mean_s": 0.034308884787606075,
"p50_s": 0.03437204146757722,
"p90_s": 0.034552122588502245,
"p95_s": 0.034628701777546664,
"p99_s": 0.0346899651287822,
"min_s": 0.033615602005738765,
"max_s": 0.03470528096659109
},
"linear.mlp_fc1.slice_reshape": {
"count": 10,
"mean_s": 5.462393164634704e-06,
"p50_s": 5.399982910603285e-06,
"p90_s": 5.71361742913723e-06,
"p95_s": 5.86480600759387e-06,
"p99_s": 5.985756870359183e-06,
"min_s": 5.231995601207018e-06,
"max_s": 6.01599458605051e-06
},
"mlp_fc1": {
"count": 10,
"mean_s": 0.038320799911161885,
"p50_s": 0.03837749600643292,
"p90_s": 0.03857543897465803,
"p95_s": 0.03861309501517099,
"p99_s": 0.03864321984758135,
"min_s": 0.03753324900753796,
"max_s": 0.03865075105568394
},
"mlp_swiglu": {
"count": 10,
"mean_s": 0.023907500400673597,
"p50_s": 0.023899421998066828,
"p90_s": 0.02409050110145472,
"p95_s": 0.024095325541566126,
"p99_s": 0.02409918509365525,
"min_s": 0.023705605999566615,
"max_s": 0.024100149981677532
},
"linear.mlp_fc2.flatten_contiguous": {
"count": 10,
"mean_s": 4.129589069634676e-06,
"p50_s": 4.135974450036883e-06,
"p90_s": 4.268810153007507e-06,
"p95_s": 4.470394924283027e-06,
"p99_s": 4.631662741303444e-06,
"min_s": 3.871973603963852e-06,
"max_s": 4.671979695558548e-06
},
"linear.mlp_fc2.pre_quant_scale": {
"count": 10,
"mean_s": 0.0,
"p50_s": 0.0,
"p90_s": 0.0,
"p95_s": 0.0,
"p99_s": 0.0,
"min_s": 0.0,
"max_s": 0.0
},
"linear.mlp_fc2.packed_weight_wrapper": {
"count": 10,
"mean_s": 7.93120707385242e-06,
"p50_s": 7.888011168688536e-06,
"p90_s": 8.115218952298164e-06,
"p95_s": 8.417596109211444e-06,
"p99_s": 8.65949783474207e-06,
"min_s": 7.424037903547287e-06,
"max_s": 8.719973266124725e-06
},
"linear.mlp_fc2.bias_cast": {
"count": 10,
"mean_s": 3.532791743054986e-06,
"p50_s": 3.503984771668911e-06,
"p90_s": 3.6463839933276177e-06,
"p95_s": 3.7111865822225807e-06,
"p99_s": 3.7630286533385513e-06,
"min_s": 3.375986125320196e-06,
"max_s": 3.775989171117544e-06
},
"linear.mlp_fc2.activation_scale": {
"count": 10,
"mean_s": 0.004426897293888033,
"p50_s": 0.004421310004545376,
"p90_s": 0.004461953247664497,
"p95_s": 0.004538503623916767,
"p99_s": 0.004599743924918585,
"min_s": 0.004358205944299698,
"max_s": 0.004615054000169039
},
"linear.mlp_fc2.scale_to_device": {
"count": 10,
"mean_s": 1.4316797023639084e-05,
"p50_s": 9.480019798502326e-06,
"p90_s": 2.5710387853905554e-05,
"p95_s": 3.182317886967211e-05,
"p99_s": 3.671341168228537e-05,
"min_s": 8.736038580536842e-06,
"max_s": 3.793596988543868e-05
},
"linear.mlp_fc2.activation_quant_pack": {
"count": 10,
"mean_s": 0.005852458404842764,
"p50_s": 0.00577670949860476,
"p90_s": 0.006046092335600406,
"p95_s": 0.006052780675236135,
"p99_s": 0.006058131346944719,
"min_s": 0.005697708984371275,
"max_s": 0.0060594690148718655
},
"linear.mlp_fc2.activation_quant_wrap": {
"count": 10,
"mean_s": 6.2527891714125875e-06,
"p50_s": 6.199959898367524e-06,
"p90_s": 6.398366531357169e-06,
"p95_s": 6.679186481051146e-06,
"p99_s": 6.903842440806329e-06,
"min_s": 6.064015906304121e-06,
"max_s": 6.960006430745125e-06
},
"linear.mlp_fc2.gemm": {
"count": 10,
"mean_s": 0.05624732690048404,
"p50_s": 0.0563016329833772,
"p90_s": 0.05672116960049607,
"p95_s": 0.056825684782234015,
"p99_s": 0.056909296927624385,
"min_s": 0.05543639202369377,
"max_s": 0.05693019996397197
},
"linear.mlp_fc2.slice_reshape": {
"count": 10,
"mean_s": 5.435204366222024e-06,
"p50_s": 5.359994247555733e-06,
"p90_s": 5.553592927753925e-06,
"p95_s": 5.8487930800765744e-06,
"p99_s": 6.084953201934695e-06,
"min_s": 5.199981387704611e-06,
"max_s": 6.143993232399225e-06
},
"mlp_fc2": {
"count": 10,
"mean_s": 0.06660801698453725,
"p50_s": 0.06663533198297955,
"p90_s": 0.06701652241172269,
"p95_s": 0.06725638319621793,
"p99_s": 0.06744827182381413,
"min_s": 0.0657997639500536,
"max_s": 0.06749624398071319
},
"gate_mlp": {
"count": 10,
"mean_s": 0.010363909183070064,
"p50_s": 0.008649876486742869,
"p90_s": 0.010803180868970225,
"p95_s": 0.017989072922500765,
"p99_s": 0.023737786565325224,
"min_s": 0.008425659965723753,
"max_s": 0.025174964976031333
},
"block_total": {
"count": 10,
"mean_s": 0.499170052097179,
"p50_s": 0.4875695880036801,
"p90_s": 0.5122190653986763,
"p95_s": 0.5512197931850096,
"p99_s": 0.5824203754140763,
"min_s": 0.4862319889944047,
"max_s": 0.5902205209713429
}
},
"module_forward": {
"count": 10,
"mean_s": 0.4679248433036264,
"p50_s": 0.46821730851661414,
"p90_s": 0.46862109302892346,
"p95_s": 0.4686407850269461,
"p99_s": 0.46865653862536416,
"min_s": 0.4657743179704994,
"max_s": 0.4686604770249687
},
"module_forward_checksum": -23765568.0,
"fused_elementwise": true,
"profiler_summary": {
"profiled_iterations": 1,
"runtime_kernel_launches": 0,
"runtime_kernel_launches_per_block": 0.0,
"positive_self_device_allocated_bytes": 9093546496
},
"profiler_top_events": [
{
"key": "aten::rms_norm",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1284.5120000000002,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": -303104,
"input_shapes": "[[37810, 5376], [], [5376], []]"
},
{
"key": "aten::_fused_rms_norm",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1275.904,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 5376], [], [5376], []]"
},
{
"key": "aten::linear",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 253.10399999999981,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 5376], [21504, 5376], []]"
},
{
"key": "comfy_kitchen::quantize_nvfp4",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 204.17599999999993,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 5376], [], [], [], []]"
},
{
"key": "comfy_kitchen::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 188.06399999999985,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": -512,
"input_shapes": "[[37824, 2688], [21504, 2688], [], [], [37888, 336], [21504, 336], [], [], []]"
},
{
"key": "comfy_kitchen::rms_rope_split_half_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 176.30299999999988,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 37810, 56, 128], [1, 37810, 56, 128], [1, 37810, 1, 48, 2, 2], [128], [128], [], []]"
},
{
"key": "aten::_to_copy",
"count": 8,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 134.94399999999905,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], [], [], []]"
},
{
"key": "aten::linear",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 108.04799999999977,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 14336], [5376, 14336], []]"
},
{
"key": "aten::linear",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 102.22400000000016,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 7168], [5376, 7168], []]"
},
{
"key": "aten::to",
"count": 12,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 101.503999999999,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], []]"
},
{
"key": "aten::empty",
"count": 34,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 100.47999999999882,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 8009298432,
"input_shapes": "[[], [], [], [], [], []]"
},
{
"key": "aten::linear",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 96.33600000000024,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 5376], [28672, 5376], []]"
},
{
"key": "comfy_kitchen::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 92.97600000000011,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": -512,
"input_shapes": "[[37824, 7168], [5376, 7168], [], [], [37888, 896], [5376, 896], [], [], []]"
},
{
"key": "comfy_kitchen::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 85.71199999999999,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": -512,
"input_shapes": "[[37824, 3584], [5376, 3584], [], [], [37888, 448], [5376, 448], [], [], []]"
},
{
"key": "comfy_kitchen::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 83.24800000000005,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": -512,
"input_shapes": "[[37824, 2688], [28672, 2688], [], [], [37888, 336], [28672, 336], [], [], []]"
},
{
"key": "aten::copy_",
"count": 8,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 76.6719999999998,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], []]"
},
{
"key": "aten::zeros",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 69.5039999999999,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], []]"
},
{
"key": "aten::mean",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 64.27199999999993,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 14336,
"input_shapes": "[[1, 37810, 56, 128], [], [], []]"
},
{
"key": "comfy_kitchen::quantize_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 57.96799999999985,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 7168], [], [], [], []]"
},
{
"key": "sageattention_sm89::qk_int8_sv_f8_accum_f16_fuse_v_scale_attn_inst_buf",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 51.983999999999924,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 37810, 56, 128], [1, 37810, 56, 128], [1, 128, 56, 37824], [1, 37810, 56, 128], [1, 56, 1184], [1, 56, 591], [1, 56, 128], [], [], [], [], []]"
},
{
"key": "comfy_kitchen::quantize_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 49.69599999999991,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 14336], [], [], [], []]"
},
{
"key": "aten::mul",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 47.5359999999996,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 2048,
"input_shapes": "[[], []]"
},
{
"key": "aten::to",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 46.91199999999981,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], [], []]"
},
{
"key": "aten::silu",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 33.74400000000014,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 1084227584,
"input_shapes": "[[37810, 14336]]"
},
{
"key": "aten::mul_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 28.192000000000007,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 14336], [37810, 14336]]"
},
{
"key": "aten::zero_",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 27.455999999999904,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 336]]"
},
{
"key": "aten::empty_strided",
"count": 8,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 24.495999999999867,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 4096,
"input_shapes": "[[], [], [], [], [], []]"
},
{
"key": "aten::slice",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 20.159999999999854,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37824, 21504], [], [], [], []]"
},
{
"key": "aten::split",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 19.759999999999764,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 21504], [], []]"
},
{
"key": "aten::fill_",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 15.248000000000047,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 336], []]"
},
{
"key": "aten::chunk",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 11.583999999999833,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672], [], []]"
},
{
"key": "aten::zero_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 11.376000000000204,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 896]]"
},
{
"key": "aten::narrow",
"count": 3,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 10.959999999999582,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 21504], [], [], []]"
},
{
"key": "aten::reshape",
"count": 8,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 10.032000000001062,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], []]"
},
{
"key": "aten::fill_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 9.83999999999969,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 896], []]"
},
{
"key": "aten::split",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 8.079999999999927,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672], [], []]"
},
{
"key": "aten::as_strided",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 8.016000000000076,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37824, 21504], [], [], []]"
},
{
"key": "aten::view",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 7.6159999999999854,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 336], []]"
},
{
"key": "aten::squeeze",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 7.472000000000207,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 1, 56, 128], []]"
},
{
"key": "aten::zero_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 7.2159999999998945,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 448]]"
},
{
"key": "aten::reshape",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 5.664000000000442,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 5376], []]"
},
{
"key": "aten::narrow",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 5.6479999999996835,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672], [], [], []]"
},
{
"key": "aten::fill_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 5.536000000000058,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 448], []]"
},
{
"key": "aten::slice",
"count": 3,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 5.023999999999887,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 21504], [], [], [], []]"
},
{
"key": "aten::to",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 4.480000000000047,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[5376], [], [], [], []]"
},
{
"key": "aten::slice",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 3.680000000000291,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37824, 5376], [], [], [], []]"
},
{
"key": "aten::view",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 3.4880000000000564,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810], []]"
},
{
"key": "aten::slice",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 3.4400000000005093,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672], [], [], [], []]"
},
{
"key": "aten::view",
"count": 8,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.7360000000005584,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], []]"
},
{
"key": "aten::view",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.512000000000171,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 7168], []]"
},
{
"key": "aten::slice",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.2719999999999345,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37824, 28672], [], [], [], []]"
},
{
"key": "aten::alias",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.2080000000000837,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 21504]]"
},
{
"key": "aten::as_strided",
"count": 3,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.9360000000001492,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 21504], [], [], []]"
},
{
"key": "aten::view",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.7760000000000673,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 5376], []]"
},
{
"key": "aten::reshape",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.6320000000000618,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 37810, 56, 128], []]"
},
{
"key": "aten::as_strided",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.5199999999999818,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 37810, 56, 128], [], [], []]"
},
{
"key": "aten::reshape",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.1840000000001965,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 7168], []]"
},
{
"key": "aten::reshape",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.1520000000000437,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672], []]"
},
{
"key": "aten::reshape",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.1520000000000437,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 14336], []]"
},
{
"key": "aten::reshape",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.1359999999999673,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 21504], []]"
},
{
"key": "aten::as_strided",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.83199999999988,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37824, 5376], [], [], []]"
},
{
"key": "aten::as_strided",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.7039999999997235,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672], [], [], []]"
},
{
"key": "aten::view",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.7039999999997235,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 896], []]"
},
{
"key": "aten::alias",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.6720000000000255,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 5376]]"
},
{
"key": "aten::view",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.6399999999998727,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 448], []]"
},
{
"key": "aten::alias",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.5920000000000982,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 37810, 56, 128]]"
},
{
"key": "aten::as_strided",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.4800000000000182,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37824, 28672], [], [], []]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.44800000000032014,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 37810, 56, 128], []]"
},
{
"key": "aten::alias",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.41600000000016735,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672]]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.38400000000001455,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 21504], []]"
},
{
"key": "aten::as_strided",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.36799999999993815,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[1, 1, 56, 128], [], [], []]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.3520000000003165,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 14336], []]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.35199999999986176,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37810, 28672], []]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.3200000000001637,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[21504, 336], []]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.3040000000000873,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[28672, 336], []]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.30399999999963256,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[5376, 896], []]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.2719999999999345,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[5376, 448], []]"
},
{
"key": "[memory]",
"count": 33,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.0,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": -9093241344,
"input_shapes": "[]"
}
]
}