h3-blackwell-runtime/benchmarks/gb10-deployed-modulate-fusion-block24-profile.json

1392 lines
43 KiB
JSON
Raw 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": 0.003516814694739878,
"p50_s": 0.0035167409805580974,
"p90_s": 0.0035479393962305037,
"p95_s": 0.0035552402172470466,
"p99_s": 0.0035610808740602806,
"min_s": 0.0034472449915483594,
"max_s": 0.003562541038263589
},
"modulate_msa": {
"count": 10,
"mean_s": 0.010366741399047896,
"p50_s": 0.01035376702202484,
"p90_s": 0.010706545307766647,
"p95_s": 0.01078284414834343,
"p99_s": 0.010843883220804856,
"min_s": 0.009973047010134906,
"max_s": 0.010859142988920212
},
"linear.attn_qkv_proj.flatten_contiguous": {
"count": 10,
"mean_s": 4.526402335613966e-06,
"p50_s": 4.10401844419539e-06,
"p90_s": 4.755170084536074e-06,
"p95_s": 6.42557861283421e-06,
"p99_s": 7.761905435472728e-06,
"min_s": 3.903987817466259e-06,
"max_s": 8.095987141132355e-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": 1.0039989138022066e-05,
"p50_s": 8.119968697428703e-06,
"p90_s": 1.0654376819729798e-05,
"p95_s": 1.935919281095264e-05,
"p99_s": 2.6323045603930952e-05,
"min_s": 7.39202369004488e-06,
"max_s": 2.8064008802175522e-05
},
"linear.attn_qkv_proj.bias_cast": {
"count": 10,
"mean_s": 3.5263889003545047e-06,
"p50_s": 3.42397834174335e-06,
"p90_s": 3.747199662029743e-06,
"p95_s": 4.049576818943023e-06,
"p99_s": 4.2914785444736485e-06,
"min_s": 3.2800016924738884e-06,
"max_s": 4.351953975856304e-06
},
"linear.attn_qkv_proj.activation_scale": {
"count": 10,
"mean_s": 0.0016631810984108597,
"p50_s": 0.0016464549989905208,
"p90_s": 0.0016899974027182907,
"p95_s": 0.0017555822065332904,
"p99_s": 0.0018080500495852904,
"min_s": 0.0016271350323222578,
"max_s": 0.0018211670103482902
},
"linear.attn_qkv_proj.scale_to_device": {
"count": 10,
"mean_s": 1.1579185957089066e-05,
"p50_s": 9.391980711370707e-06,
"p90_s": 1.7219223082065582e-05,
"p95_s": 1.8961611203849312e-05,
"p99_s": 2.0355521701276303e-05,
"min_s": 9.023991879075766e-06,
"max_s": 2.070399932563305e-05
},
"linear.attn_qkv_proj.activation_quant_pack": {
"count": 10,
"mean_s": 0.0022007547901012003,
"p50_s": 0.002188237995142117,
"p90_s": 0.002310799522092566,
"p95_s": 0.0023156302486313507,
"p99_s": 0.0023194948298623785,
"min_s": 0.002091966976877302,
"max_s": 0.0023204609751701355
},
"linear.attn_qkv_proj.activation_quant_wrap": {
"count": 10,
"mean_s": 6.332807242870331e-06,
"p50_s": 6.1680038925260305e-06,
"p90_s": 6.833561928942799e-06,
"p95_s": 7.344779442064463e-06,
"p99_s": 7.753753452561796e-06,
"min_s": 5.824025720357895e-06,
"max_s": 7.855996955186129e-06
},
"linear.attn_qkv_proj.gemm": {
"count": 10,
"mean_s": 0.02569694590056315,
"p50_s": 0.025759841984836385,
"p90_s": 0.0262955051031895,
"p95_s": 0.026411317544989286,
"p99_s": 0.02650396749842912,
"min_s": 0.024846857995726168,
"max_s": 0.026527129986789078
},
"linear.attn_qkv_proj.slice_reshape": {
"count": 10,
"mean_s": 5.633599357679486e-06,
"p50_s": 5.3679978009313345e-06,
"p90_s": 5.988759221509098e-06,
"p95_s": 6.8743858719244576e-06,
"p99_s": 7.582887192256748e-06,
"min_s": 5.231995601207018e-06,
"max_s": 7.760012522339821e-06
},
"attn_qkv_proj": {
"count": 10,
"mean_s": 0.029643844993552194,
"p50_s": 0.029725486994720995,
"p90_s": 0.030352766008581965,
"p95_s": 0.030365185998380183,
"p99_s": 0.03037512199021876,
"min_s": 0.0287816229974851,
"max_s": 0.030377605988178402
},
"attn_qkv_split_view": {
"count": 10,
"mean_s": 1.3008003588765859e-05,
"p50_s": 1.2336007785052061e-05,
"p90_s": 1.554721384309232e-05,
"p95_s": 1.7541609122417863e-05,
"p99_s": 1.9137125345878303e-05,
"min_s": 9.984010830521584e-06,
"max_s": 1.9536004401743412e-05
},
"attn_qk_rms_rope": {
"count": 10,
"mean_s": 0.012352611508686096,
"p50_s": 0.01237390103051439,
"p90_s": 0.012589196878252551,
"p95_s": 0.012807608451112173,
"p99_s": 0.01298233770939987,
"min_s": 0.011864198022522032,
"max_s": 0.013026020023971796
},
"attention_kernel": {
"count": 10,
"mean_s": 0.27108906718785875,
"p50_s": 0.26181144698057324,
"p90_s": 0.29291597349219956,
"p95_s": 0.312738003252889,
"p99_s": 0.32859562706144063,
"min_s": 0.25998799898661673,
"max_s": 0.3325600330135785
},
"attn_output_reshape": {
"count": 10,
"mean_s": 4.281604196876287e-06,
"p50_s": 4.128029104322195e-06,
"p90_s": 4.7808163799345495e-06,
"p95_s": 4.9104215577244755e-06,
"p99_s": 5.0141056999564175e-06,
"min_s": 3.936002030968666e-06,
"max_s": 5.040026735514402e-06
},
"linear.attn_out_proj.flatten_contiguous": {
"count": 10,
"mean_s": 4.358391743153334e-06,
"p50_s": 4.128000000491738e-06,
"p90_s": 5.2239862270653244e-06,
"p95_s": 5.2599760238081215e-06,
"p99_s": 5.288767861202359e-06,
"min_s": 3.936002030968666e-06,
"max_s": 5.295965820550919e-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": 9.428808698430658e-06,
"p50_s": 8.216011337935925e-06,
"p90_s": 1.2791971676051616e-05,
"p95_s": 1.2899993453174828e-05,
"p99_s": 1.29864108748734e-05,
"min_s": 7.936032488942146e-06,
"max_s": 1.3008015230298042e-05
},
"linear.attn_out_proj.bias_cast": {
"count": 10,
"mean_s": 3.443198511376977e-06,
"p50_s": 3.4320109989494085e-06,
"p90_s": 3.6207784432917833e-06,
"p95_s": 3.642388037405908e-06,
"p99_s": 3.659675712697208e-06,
"min_s": 3.296008799225092e-06,
"max_s": 3.663997631520033e-06
},
"linear.attn_out_proj.activation_scale": {
"count": 10,
"mean_s": 0.002261684410041198,
"p50_s": 0.0022500380291603506,
"p90_s": 0.0022729580232407898,
"p95_s": 0.0023227820202009752,
"p99_s": 0.0023626412177691234,
"min_s": 0.002238414017483592,
"max_s": 0.0023726060171611607
},
"linear.attn_out_proj.scale_to_device": {
"count": 10,
"mean_s": 1.038879854604602e-05,
"p50_s": 9.832001524046063e-06,
"p90_s": 1.1913618072867393e-05,
"p95_s": 1.224480802193284e-05,
"p99_s": 1.2509759981185198e-05,
"min_s": 9.440002031624317e-06,
"max_s": 1.2575997970998287e-05
},
"linear.attn_out_proj.activation_quant_pack": {
"count": 10,
"mean_s": 0.0029140072059817613,
"p50_s": 0.0028348854975774884,
"p90_s": 0.002980089106131345,
"p95_s": 0.0032665550534147765,
"p99_s": 0.0034957278112415224,
"min_s": 0.002825118019245565,
"max_s": 0.003553021000698209
},
"linear.attn_out_proj.activation_quant_wrap": {
"count": 10,
"mean_s": 6.366416346281767e-06,
"p50_s": 6.287998985499144e-06,
"p90_s": 6.824050797149538e-06,
"p95_s": 6.93204638082534e-06,
"p99_s": 7.018442847765983e-06,
"min_s": 5.904003046452999e-06,
"max_s": 7.0400419645011425e-06
},
"linear.attn_out_proj.gemm": {
"count": 10,
"mean_s": 0.008031879493501037,
"p50_s": 0.007886153471190482,
"p90_s": 0.00864636250771582,
"p95_s": 0.008688777242787181,
"p99_s": 0.008722709030844272,
"min_s": 0.007817146019078791,
"max_s": 0.008731191977858543
},
"linear.attn_out_proj.slice_reshape": {
"count": 10,
"mean_s": 5.478394450619817e-06,
"p50_s": 5.496025551110506e-06,
"p90_s": 5.662406329065561e-06,
"p95_s": 5.727208917960524e-06,
"p99_s": 5.779050989076496e-06,
"min_s": 5.183974280953407e-06,
"max_s": 5.792011506855488e-06
},
"attn_out_proj": {
"count": 10,
"mean_s": 0.013287241506623105,
"p50_s": 0.01306043702061288,
"p90_s": 0.013964672799920664,
"p95_s": 0.014308530397829599,
"p99_s": 0.014583616476156748,
"min_s": 0.012968517025001347,
"max_s": 0.014652387995738536
},
"gate_msa": {
"count": 10,
"mean_s": 0.008519085415173322,
"p50_s": 0.0083449850208126,
"p90_s": 0.008864006400108338,
"p95_s": 0.009121759195113554,
"p99_s": 0.009327961431117728,
"min_s": 0.008315177052281797,
"max_s": 0.009379511990118772
},
"norm2": {
"count": 10,
"mean_s": 0.003516377799678594,
"p50_s": 0.003482725005596876,
"p90_s": 0.003730286500649527,
"p95_s": 0.0037477892503375186,
"p99_s": 0.0037617914500879127,
"min_s": 0.003390222031157464,
"max_s": 0.003765292000025511
},
"modulate_mlp": {
"count": 10,
"mean_s": 0.010381480591604486,
"p50_s": 0.010417310986667871,
"p90_s": 0.010713717294856906,
"p95_s": 0.011112517665605991,
"p99_s": 0.011431557962205262,
"min_s": 0.009898118965793401,
"max_s": 0.011511318036355078
},
"linear.mlp_fc1.flatten_contiguous": {
"count": 10,
"mean_s": 4.295998951420188e-06,
"p50_s": 4.0879822336137295e-06,
"p90_s": 4.825583891943097e-06,
"p95_s": 4.9408088671043505e-06,
"p99_s": 5.032988847233355e-06,
"min_s": 4.00003045797348e-06,
"max_s": 5.056033842265606e-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": 8.068798342719674e-06,
"p50_s": 7.96798849478364e-06,
"p90_s": 8.828815771266818e-06,
"p95_s": 8.88641516212374e-06,
"p99_s": 8.932494674809277e-06,
"min_s": 7.375958375632763e-06,
"max_s": 8.944014552980661e-06
},
"linear.mlp_fc1.bias_cast": {
"count": 10,
"mean_s": 3.4864002373069526e-06,
"p50_s": 3.4399854484945536e-06,
"p90_s": 3.6512385122478006e-06,
"p95_s": 3.7376245018094775e-06,
"p99_s": 3.8067332934588194e-06,
"min_s": 3.328023012727499e-06,
"max_s": 3.824010491371155e-06
},
"linear.mlp_fc1.activation_scale": {
"count": 10,
"mean_s": 0.001673643506364897,
"p50_s": 0.001655454485444352,
"p90_s": 0.0017760470509529114,
"p95_s": 0.0017949110479094088,
"p99_s": 0.0018100022454746066,
"min_s": 0.001621982955839485,
"max_s": 0.0018137750448659062
},
"linear.mlp_fc1.scale_to_device": {
"count": 10,
"mean_s": 1.0702403960749507e-05,
"p50_s": 9.176001185551286e-06,
"p90_s": 1.4611223014071582e-05,
"p95_s": 1.7001610831357533e-05,
"p99_s": 1.8913921085186306e-05,
"min_s": 8.89599323272705e-06,
"max_s": 1.9391998648643494e-05
},
"linear.mlp_fc1.activation_quant_pack": {
"count": 10,
"mean_s": 0.002208998193964362,
"p50_s": 0.002138246491085738,
"p90_s": 0.0023599260370247067,
"p95_s": 0.0025579620152711862,
"p99_s": 0.002716390797868371,
"min_s": 0.0020945259602740407,
"max_s": 0.002755997993517667
},
"linear.mlp_fc1.activation_quant_wrap": {
"count": 10,
"mean_s": 6.103987107053399e-06,
"p50_s": 6.063986802473664e-06,
"p90_s": 6.3119980040937655e-06,
"p95_s": 6.419993587769568e-06,
"p99_s": 6.506390054710209e-06,
"min_s": 5.776004400104284e-06,
"max_s": 6.52798917144537e-06
},
"linear.mlp_fc1.gemm": {
"count": 10,
"mean_s": 0.0395361577917356,
"p50_s": 0.03482407398405485,
"p90_s": 0.04009287208900785,
"p95_s": 0.061921360049745945,
"p99_s": 0.0793841504183365,
"min_s": 0.033949602977372706,
"max_s": 0.08374984801048413
},
"linear.mlp_fc1.slice_reshape": {
"count": 10,
"mean_s": 5.504011642187834e-06,
"p50_s": 5.54401776753366e-06,
"p90_s": 5.731184501200914e-06,
"p95_s": 5.7455908972769976e-06,
"p99_s": 5.757116014137864e-06,
"min_s": 5.13601116836071e-06,
"max_s": 5.759997293353081e-06
},
"mlp_fc1": {
"count": 10,
"mean_s": 0.04350133700063452,
"p50_s": 0.03864074300508946,
"p90_s": 0.044060093007283266,
"p95_s": 0.06623608849185979,
"p99_s": 0.08397688487952111,
"min_s": 0.0377592800068669,
"max_s": 0.0884120839764364
},
"mlp_swiglu": {
"count": 10,
"mean_s": 0.02421614230261184,
"p50_s": 0.024195923004299402,
"p90_s": 0.024524036498041822,
"p95_s": 0.024536607231129895,
"p99_s": 0.02454666381760035,
"min_s": 0.02381745196180418,
"max_s": 0.02454917796421796
},
"linear.mlp_fc2.flatten_contiguous": {
"count": 10,
"mean_s": 4.011200508102774e-06,
"p50_s": 3.951980033889413e-06,
"p90_s": 4.32159285992384e-06,
"p95_s": 4.400801844894886e-06,
"p99_s": 4.464169032871723e-06,
"min_s": 3.7919962778687477e-06,
"max_s": 4.4800108298659325e-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": 7.80478585511446e-06,
"p50_s": 7.807975634932518e-06,
"p90_s": 8.39359126985073e-06,
"p95_s": 8.508790051564574e-06,
"p99_s": 8.60094907693565e-06,
"min_s": 7.039983756840229e-06,
"max_s": 8.623988833278418e-06
},
"linear.mlp_fc2.bias_cast": {
"count": 10,
"mean_s": 3.423896851018071e-06,
"p50_s": 3.4479890018701553e-06,
"p90_s": 3.5391014534980056e-06,
"p95_s": 3.553062560968101e-06,
"p99_s": 3.564231446944177e-06,
"min_s": 3.2160314731299877e-06,
"max_s": 3.567023668438196e-06
},
"linear.mlp_fc2.activation_scale": {
"count": 10,
"mean_s": 0.004466716205934062,
"p50_s": 0.00445973250316456,
"p90_s": 0.004518093593651428,
"p95_s": 0.0045484847767511384,
"p99_s": 0.004572797723230906,
"min_s": 0.004411837027873844,
"max_s": 0.004578875959850848
},
"linear.mlp_fc2.scale_to_device": {
"count": 10,
"mean_s": 9.115197462961077e-06,
"p50_s": 9.271985618397593e-06,
"p90_s": 9.45759820751846e-06,
"p95_s": 9.60881297942251e-06,
"p99_s": 9.729784796945752e-06,
"min_s": 8.351984433829784e-06,
"max_s": 9.760027751326561e-06
},
"linear.mlp_fc2.activation_quant_pack": {
"count": 10,
"mean_s": 0.00589796609710902,
"p50_s": 0.005918154987739399,
"p90_s": 0.0060729933262337,
"p95_s": 0.006076693648356012,
"p99_s": 0.006079653906053864,
"min_s": 0.005693643004633486,
"max_s": 0.006080393970478326
},
"linear.mlp_fc2.activation_quant_wrap": {
"count": 10,
"mean_s": 6.0623977333307265e-06,
"p50_s": 6.007991032674909e-06,
"p90_s": 6.356823723763227e-06,
"p95_s": 6.378407124429941e-06,
"p99_s": 6.395673844963312e-06,
"min_s": 5.792011506855488e-06,
"max_s": 6.399990525096655e-06
},
"linear.mlp_fc2.gemm": {
"count": 10,
"mean_s": 0.05543897160096094,
"p50_s": 0.055413160007447004,
"p90_s": 0.05576852477970533,
"p95_s": 0.05588342241535429,
"p99_s": 0.05597534052387346,
"min_s": 0.05484625697135925,
"max_s": 0.05599832005100325
},
"linear.mlp_fc2.slice_reshape": {
"count": 10,
"mean_s": 5.4095988161861895e-06,
"p50_s": 5.376001354306936e-06,
"p90_s": 5.587184568867088e-06,
"p95_s": 5.601617158390582e-06,
"p99_s": 5.613163230009377e-06,
"min_s": 5.279958713799715e-06,
"max_s": 5.616049747914076e-06
},
"mlp_fc2": {
"count": 10,
"mean_s": 0.06587910819798708,
"p50_s": 0.06583923898870125,
"p90_s": 0.06626416540239007,
"p95_s": 0.06628907018457539,
"p99_s": 0.06630899401032367,
"min_s": 0.0654114960343577,
"max_s": 0.06631397496676072
},
"gate_mlp": {
"count": 10,
"mean_s": 0.009955293708480894,
"p50_s": 0.008543304516933858,
"p90_s": 0.010343933285912495,
"p95_s": 0.016277333159814576,
"p99_s": 0.021024053058936263,
"min_s": 0.008402952051255852,
"max_s": 0.02221073303371668
},
"block_total": {
"count": 10,
"mean_s": 0.5063105094013736,
"p50_s": 0.49055024900007993,
"p90_s": 0.5626000013318845,
"p95_s": 0.5668660971510691,
"p99_s": 0.5702789738064167,
"min_s": 0.4873858370119706,
"max_s": 0.5711321929702535
}
},
"module_forward": {
"count": 10,
"mean_s": 0.468727556290105,
"p50_s": 0.46669237900641747,
"p90_s": 0.47172567221568895,
"p95_s": 0.4803373321017716,
"p99_s": 0.4872266600106377,
"min_s": 0.4628840069635771,
"max_s": 0.48894899198785424
},
"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": 9093805568
},
"profiler_top_events": [
{
"key": "aten::rms_norm",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1399.279,
"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": 1339.6789999999999,
"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": 1062.143,
"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::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 790.6550000000007,
"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": "aten::contiguous",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 676.1750000000002,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[3, 5376], []]"
},
{
"key": "aten::clone",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 665.4230000000007,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[3, 5376], []]"
},
{
"key": "aten::zeros",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 565.4390000000003,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], []]"
},
{
"key": "aten::zero_",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 515.6469999999999,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[37888, 336]]"
},
{
"key": "comfy_kitchen::rms_rope_split_half_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 240.76800000000003,
"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::empty",
"count": 42,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 161.05600000000118,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 8009559552,
"input_shapes": "[[], [], [], [], [], []]"
},
{
"key": "aten::copy_",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 144.70400000000018,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[3, 5376], [3, 5376], []]"
},
{
"key": "aten::linear",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 126.6869999999999,
"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": "comfy_kitchen::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 106.64000000000033,
"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": "aten::linear",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 99.10400000000027,
"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": "aten::linear",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 95.64800000000014,
"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::mul",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 84.79999999999927,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 2048,
"input_shapes": "[[], []]"
},
{
"key": "comfy_kitchen::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 84.40000000000055,
"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": "comfy_kitchen::scaled_mm_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 82.54399999999987,
"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::quantize_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 76.83200000000033,
"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::_to_copy",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 53.56800000000112,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], [], [], []]"
},
{
"key": "comfy_kitchen::quantize_nvfp4",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 49.42399999999998,
"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::fill_",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 45.039999999999964,
"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::mean",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 44.86399999999958,
"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": "aten::to",
"count": 10,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 38.496000000001914,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], []]"
},
{
"key": "aten::empty_like",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 32.57599999999957,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[3, 5376], [], [], [], [], []]"
},
{
"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": 31.98400000000038,
"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": "aten::copy_",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 31.167999999999665,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], []]"
},
{
"key": "aten::split",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 29.360000000000582,
"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::silu",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 24.271999999999935,
"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::to",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 22.640000000000327,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], [], [], [], [], []]"
},
{
"key": "aten::narrow",
"count": 3,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 18.144000000000233,
"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::mul_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 16.751999999999498,
"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::reshape",
"count": 6,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 11.904000000000451,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], []]"
},
{
"key": "aten::slice",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 11.407999999999447,
"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::empty_strided",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 11.264000000001033,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 2048,
"input_shapes": "[[], [], [], [], [], []]"
},
{
"key": "aten::chunk",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 10.527999999999338,
"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": 9.88799999999992,
"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::view",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 8.368000000001302,
"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::fill_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 7.423999999999978,
"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::split",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 7.216000000000349,
"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": 7.200000000000273,
"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": 3,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 7.024000000000342,
"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::zero_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 6.22400000000016,
"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::to",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 5.631999999999721,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[5376], [], [], [], []]"
},
{
"key": "aten::fill_",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 4.895999999999731,
"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": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 4.895999999998821,
"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::squeeze",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 3.9840000000003783,
"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::slice",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 3.6640000000006694,
"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": 3,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 3.4719999999997526,
"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": 3.3599999999996726,
"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::view",
"count": 6,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.8159999999998035,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[], []]"
},
{
"key": "aten::slice",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.639999999999418,
"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": 2.6239999999997963,
"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::reshape",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.4160000000001673,
"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::alias",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 2.1920000000000073,
"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::slice",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.8239999999996144,
"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::reshape",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.7760000000007494,
"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": 1.5039999999999054,
"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::as_strided",
"count": 4,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 1.4720000000006621,
"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": 1.2960000000002765,
"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": 0.9600000000000364,
"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::alias",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.8800000000001091,
"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": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.8159999999998035,
"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": 0.815999999998894,
"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::as_strided",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.7520000000004075,
"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": 0.7200000000002547,
"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::alias",
"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, 5376]]"
},
{
"key": "aten::view",
"count": 2,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.6239999999997963,
"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::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.47999999999956344,
"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::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.4639999999999418,
"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::as_strided",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.44799999999941065,
"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": 0.41600000000016735,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[5376, 448], []]"
},
{
"key": "aten::as_strided",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.3999999999996362,
"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": 0.35199999999986176,
"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::alias",
"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": "[[37810, 28672]]"
},
{
"key": "aten::view",
"count": 1,
"device_type": "DeviceType.CPU",
"cpu_time_total_us": 0.25600000000031287,
"device_time_total_us": 0.0,
"self_device_time_total_us": 0.0,
"self_device_memory_usage_bytes": 0,
"input_shapes": "[[5376, 896], []]"
},
{
"key": "[memory]",
"count": 37,
"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": -9093500416,
"input_shapes": "[]"
}
]
}