hf-transformers-bot commited on
Commit
465bf98
·
verified ·
1 Parent(s): de9f2a5

Upload benchmark results for run 18043751563

Browse files
2025-09-26/18043751563/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20250926_164926.json ADDED
@@ -0,0 +1,1175 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "Llama-2-7b-hf",
3
+ "benchmark_scenarios": [
4
+ {
5
+ "scenario_name": "eager_eager_attn",
6
+ "metadata": {
7
+ "timestamp": "2025-09-26T16:43:25.483442",
8
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
9
+ "hardware_info": {
10
+ "gpu_name": "NVIDIA A10G",
11
+ "gpu_memory_total_mb": 23028,
12
+ "cpu_count": 16,
13
+ "memory_total_mb": 63607,
14
+ "python_version": "3.10.12",
15
+ "torch_version": "2.8.0+cu126",
16
+ "cuda_version": "12.6"
17
+ },
18
+ "config": {
19
+ "name": "eager",
20
+ "model_id": "meta-llama/Llama-2-7b-hf",
21
+ "variant": "eager",
22
+ "warmup_iterations": 3,
23
+ "measurement_iterations": 5,
24
+ "num_tokens_to_generate": 100,
25
+ "device": "cuda",
26
+ "torch_dtype": "float16",
27
+ "compile_mode": null,
28
+ "compile_options": {},
29
+ "use_cache": true,
30
+ "batch_size": 1,
31
+ "sequence_length": null,
32
+ "attn_implementation": "eager",
33
+ "sdpa_backend": null,
34
+ "custom_params": {}
35
+ }
36
+ },
37
+ "measurements": {
38
+ "latency_seconds": {
39
+ "name": "latency_seconds",
40
+ "measurements": [
41
+ 3.66456005859375,
42
+ 3.661247802734375,
43
+ 3.660901611328125,
44
+ 3.657181640625,
45
+ 3.65410986328125
46
+ ],
47
+ "mean": 3.6596001953124997,
48
+ "median": 3.660901611328125,
49
+ "std": 0.0036055981534505936,
50
+ "min": 3.65410986328125,
51
+ "max": 3.66456005859375,
52
+ "p25": 3.657181640625,
53
+ "p75": 3.661247802734375,
54
+ "p90": 3.66323515625,
55
+ "p95": 3.663897607421875,
56
+ "p99": 3.664427568359375,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.038182689666748044,
63
+ 0.03866620635986328,
64
+ 0.0384640007019043,
65
+ 0.03789846420288086,
66
+ 0.04014198303222656
67
+ ],
68
+ "mean": 0.038670668792724606,
69
+ "median": 0.0384640007019043,
70
+ "std": 0.0007799898410102749,
71
+ "min": 0.03789846420288086,
72
+ "max": 0.04014198303222656,
73
+ "p25": 0.038182689666748044,
74
+ "p75": 0.03866620635986328,
75
+ "p90": 0.03955167236328125,
76
+ "p95": 0.039846827697753905,
77
+ "p99": 0.04008295196533203,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.28840526586275,
84
+ 27.31309252690183,
85
+ 27.315675376405807,
86
+ 27.343460026478294,
87
+ 27.36644593115869
88
+ ],
89
+ "mean": 27.32541582536147,
90
+ "median": 27.315675376405807,
91
+ "std": 0.026927994061448555,
92
+ "min": 27.28840526586275,
93
+ "max": 27.36644593115869,
94
+ "p25": 27.31309252690183,
95
+ "p75": 27.343460026478294,
96
+ "p90": 27.357251569286532,
97
+ "p95": 27.36184875022261,
98
+ "p99": 27.365526494971473,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.0366456005859375,
105
+ 0.03661247802734375,
106
+ 0.03660901611328125,
107
+ 0.03657181640625,
108
+ 0.0365410986328125
109
+ ],
110
+ "mean": 0.036596001953125,
111
+ "median": 0.03660901611328125,
112
+ "std": 3.6055981534506816e-05,
113
+ "min": 0.0365410986328125,
114
+ "max": 0.0366456005859375,
115
+ "p25": 0.03657181640625,
116
+ "p75": 0.03661247802734375,
117
+ "p90": 0.0366323515625,
118
+ "p95": 0.03663897607421875,
119
+ "p99": 0.03664427568359375,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 90.65217391304348,
125
+ "gpu_utilization_max": 94,
126
+ "gpu_utilization_min": 82,
127
+ "gpu_memory_used_mean": 13253,
128
+ "gpu_memory_used_max": 13253,
129
+ "gpu_memory_used_min": 13253,
130
+ "sample_count": 92,
131
+ "gpu_monitoring_status": "success"
132
+ }
133
+ },
134
+ {
135
+ "scenario_name": "compiled_compile_max-autotune_eager_attn",
136
+ "metadata": {
137
+ "timestamp": "2025-09-26T16:44:00.747534",
138
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
139
+ "hardware_info": {
140
+ "gpu_name": "NVIDIA A10G",
141
+ "gpu_memory_total_mb": 23028,
142
+ "cpu_count": 16,
143
+ "memory_total_mb": 63607,
144
+ "python_version": "3.10.12",
145
+ "torch_version": "2.8.0+cu126",
146
+ "cuda_version": "12.6"
147
+ },
148
+ "config": {
149
+ "name": "compiled",
150
+ "model_id": "meta-llama/Llama-2-7b-hf",
151
+ "variant": "compiled",
152
+ "warmup_iterations": 3,
153
+ "measurement_iterations": 5,
154
+ "num_tokens_to_generate": 100,
155
+ "device": "cuda",
156
+ "torch_dtype": "float16",
157
+ "compile_mode": "max-autotune",
158
+ "compile_options": {},
159
+ "use_cache": true,
160
+ "batch_size": 1,
161
+ "sequence_length": null,
162
+ "attn_implementation": "eager",
163
+ "sdpa_backend": null,
164
+ "custom_params": {}
165
+ }
166
+ },
167
+ "measurements": {
168
+ "latency_seconds": {
169
+ "name": "latency_seconds",
170
+ "measurements": [
171
+ 7.9243525390625,
172
+ 7.82325439453125,
173
+ 7.84966552734375,
174
+ 7.87963720703125,
175
+ 7.9011279296875
176
+ ],
177
+ "mean": 7.875607519531249,
178
+ "median": 7.87963720703125,
179
+ "std": 0.03593728788018043,
180
+ "min": 7.82325439453125,
181
+ "max": 7.9243525390625,
182
+ "p25": 7.84966552734375,
183
+ "p75": 7.9011279296875,
184
+ "p90": 7.9150626953125,
185
+ "p95": 7.9197076171875,
186
+ "p99": 7.9234235546875,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.038644798278808595,
193
+ 0.03810131072998047,
194
+ 0.038072383880615235,
195
+ 0.03836159896850586,
196
+ 0.03809049606323242
197
+ ],
198
+ "mean": 0.03825411758422852,
199
+ "median": 0.03810131072998047,
200
+ "std": 0.0002224107591601659,
201
+ "min": 0.038072383880615235,
202
+ "max": 0.038644798278808595,
203
+ "p25": 0.03809049606323242,
204
+ "p75": 0.03836159896850586,
205
+ "p90": 0.038531518554687505,
206
+ "p95": 0.03858815841674805,
207
+ "p99": 0.03863347030639649,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.61932751061459,
214
+ 12.782404221688582,
215
+ 12.739396303149112,
216
+ 12.690939617215731,
217
+ 12.656420816104307
218
+ ],
219
+ "mean": 12.697697693754465,
220
+ "median": 12.690939617215731,
221
+ "std": 0.057974662970176255,
222
+ "min": 12.61932751061459,
223
+ "max": 12.782404221688582,
224
+ "p25": 12.656420816104307,
225
+ "p75": 12.739396303149112,
226
+ "p90": 12.765201054272794,
227
+ "p95": 12.773802637980689,
228
+ "p99": 12.780683904947002,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.079243525390625,
235
+ 0.0782325439453125,
236
+ 0.0784966552734375,
237
+ 0.0787963720703125,
238
+ 0.079011279296875
239
+ ],
240
+ "mean": 0.0787560751953125,
241
+ "median": 0.0787963720703125,
242
+ "std": 0.00035937287880180477,
243
+ "min": 0.0782325439453125,
244
+ "max": 0.079243525390625,
245
+ "p25": 0.0784966552734375,
246
+ "p75": 0.079011279296875,
247
+ "p90": 0.079150626953125,
248
+ "p95": 0.079197076171875,
249
+ "p99": 0.079234235546875,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 44.68862275449102,
255
+ "gpu_utilization_max": 91,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13632.221556886228,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 167,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-09-26T16:45:28.368164",
268
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
269
+ "hardware_info": {
270
+ "gpu_name": "NVIDIA A10G",
271
+ "gpu_memory_total_mb": 23028,
272
+ "cpu_count": 16,
273
+ "memory_total_mb": 63607,
274
+ "python_version": "3.10.12",
275
+ "torch_version": "2.8.0+cu126",
276
+ "cuda_version": "12.6"
277
+ },
278
+ "config": {
279
+ "name": "eager",
280
+ "model_id": "meta-llama/Llama-2-7b-hf",
281
+ "variant": "eager",
282
+ "warmup_iterations": 3,
283
+ "measurement_iterations": 5,
284
+ "num_tokens_to_generate": 100,
285
+ "device": "cuda",
286
+ "torch_dtype": "float16",
287
+ "compile_mode": null,
288
+ "compile_options": {},
289
+ "use_cache": true,
290
+ "batch_size": 1,
291
+ "sequence_length": null,
292
+ "attn_implementation": "sdpa",
293
+ "sdpa_backend": null,
294
+ "custom_params": {}
295
+ }
296
+ },
297
+ "measurements": {
298
+ "latency_seconds": {
299
+ "name": "latency_seconds",
300
+ "measurements": [
301
+ 3.413284912109375,
302
+ 3.41301416015625,
303
+ 3.412535400390625,
304
+ 3.409957275390625,
305
+ 3.410779052734375
306
+ ],
307
+ "mean": 3.4119141601562504,
308
+ "median": 3.412535400390625,
309
+ "std": 0.0013109360923351419,
310
+ "min": 3.409957275390625,
311
+ "max": 3.413284912109375,
312
+ "p25": 3.410779052734375,
313
+ "p75": 3.41301416015625,
314
+ "p90": 3.413176611328125,
315
+ "p95": 3.4132307617187503,
316
+ "p99": 3.41327408203125,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.03522812652587891,
323
+ 0.034587390899658205,
324
+ 0.03485878372192383,
325
+ 0.03461833572387695,
326
+ 0.034761249542236326
327
+ ],
328
+ "mean": 0.034810777282714846,
329
+ "median": 0.034761249542236326,
330
+ "std": 0.00023059628607379455,
331
+ "min": 0.034587390899658205,
332
+ "max": 0.03522812652587891,
333
+ "p25": 0.03461833572387695,
334
+ "p75": 0.03485878372192383,
335
+ "p90": 0.03508038940429688,
336
+ "p95": 0.03515425796508789,
337
+ "p99": 0.035213352813720705,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.297290608594704,
344
+ 29.299614741540346,
345
+ 29.303725314777168,
346
+ 29.32588062662591,
347
+ 29.318814984462676
348
+ ],
349
+ "mean": 29.30906525520016,
350
+ "median": 29.303725314777168,
351
+ "std": 0.011263067181874776,
352
+ "min": 29.297290608594704,
353
+ "max": 29.32588062662591,
354
+ "p25": 29.299614741540346,
355
+ "p75": 29.318814984462676,
356
+ "p90": 29.323054369760616,
357
+ "p95": 29.324467498193265,
358
+ "p99": 29.32559800093938,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.03413284912109375,
365
+ 0.0341301416015625,
366
+ 0.034125354003906254,
367
+ 0.03409957275390625,
368
+ 0.03410779052734375
369
+ ],
370
+ "mean": 0.0341191416015625,
371
+ "median": 0.034125354003906254,
372
+ "std": 1.3109360923350178e-05,
373
+ "min": 0.03409957275390625,
374
+ "max": 0.03413284912109375,
375
+ "p25": 0.03410779052734375,
376
+ "p75": 0.0341301416015625,
377
+ "p90": 0.03413176611328125,
378
+ "p95": 0.0341323076171875,
379
+ "p99": 0.0341327408203125,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.9186046511628,
385
+ "gpu_utilization_max": 98,
386
+ "gpu_utilization_min": 95,
387
+ "gpu_memory_used_mean": 13849,
388
+ "gpu_memory_used_max": 13849,
389
+ "gpu_memory_used_min": 13849,
390
+ "sample_count": 86,
391
+ "gpu_monitoring_status": "success"
392
+ }
393
+ },
394
+ {
395
+ "scenario_name": "eager_sdpa_math",
396
+ "metadata": {
397
+ "timestamp": "2025-09-26T16:46:00.850819",
398
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
399
+ "hardware_info": {
400
+ "gpu_name": "NVIDIA A10G",
401
+ "gpu_memory_total_mb": 23028,
402
+ "cpu_count": 16,
403
+ "memory_total_mb": 63607,
404
+ "python_version": "3.10.12",
405
+ "torch_version": "2.8.0+cu126",
406
+ "cuda_version": "12.6"
407
+ },
408
+ "config": {
409
+ "name": "eager",
410
+ "model_id": "meta-llama/Llama-2-7b-hf",
411
+ "variant": "eager",
412
+ "warmup_iterations": 3,
413
+ "measurement_iterations": 5,
414
+ "num_tokens_to_generate": 100,
415
+ "device": "cuda",
416
+ "torch_dtype": "float16",
417
+ "compile_mode": null,
418
+ "compile_options": {},
419
+ "use_cache": true,
420
+ "batch_size": 1,
421
+ "sequence_length": null,
422
+ "attn_implementation": "sdpa",
423
+ "sdpa_backend": "math",
424
+ "custom_params": {}
425
+ }
426
+ },
427
+ "measurements": {
428
+ "latency_seconds": {
429
+ "name": "latency_seconds",
430
+ "measurements": [
431
+ 3.532656005859375,
432
+ 3.5324873046875,
433
+ 3.53662353515625,
434
+ 3.52774560546875,
435
+ 3.52828173828125
436
+ ],
437
+ "mean": 3.531558837890625,
438
+ "median": 3.5324873046875,
439
+ "std": 0.0032556753489862062,
440
+ "min": 3.52774560546875,
441
+ "max": 3.53662353515625,
442
+ "p25": 3.52828173828125,
443
+ "p75": 3.532656005859375,
444
+ "p90": 3.5350365234375,
445
+ "p95": 3.535830029296875,
446
+ "p99": 3.536464833984375,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.03931683349609375,
453
+ 0.03776348876953125,
454
+ 0.03846031951904297,
455
+ 0.03916543960571289,
456
+ 0.03825449752807617
457
+ ],
458
+ "mean": 0.03859211578369141,
459
+ "median": 0.03846031951904297,
460
+ "std": 0.0005782551405846638,
461
+ "min": 0.03776348876953125,
462
+ "max": 0.03931683349609375,
463
+ "p25": 0.03825449752807617,
464
+ "p75": 0.03916543960571289,
465
+ "p90": 0.03925627593994141,
466
+ "p95": 0.03928655471801758,
467
+ "p99": 0.03931077774047852,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.307313204041616,
474
+ 28.30866507780598,
475
+ 28.27555689938085,
476
+ 28.346715206725477,
477
+ 28.342407839775717
478
+ ],
479
+ "mean": 28.31613164554593,
480
+ "median": 28.30866507780598,
481
+ "std": 0.026098341588559403,
482
+ "min": 28.27555689938085,
483
+ "max": 28.346715206725477,
484
+ "p25": 28.307313204041616,
485
+ "p75": 28.342407839775717,
486
+ "p90": 28.344992259945574,
487
+ "p95": 28.345853733335524,
488
+ "p99": 28.346542912047486,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.03532656005859375,
495
+ 0.035324873046875,
496
+ 0.0353662353515625,
497
+ 0.0352774560546875,
498
+ 0.0352828173828125
499
+ ],
500
+ "mean": 0.03531558837890625,
501
+ "median": 0.035324873046875,
502
+ "std": 3.2556753489861066e-05,
503
+ "min": 0.0352774560546875,
504
+ "max": 0.0353662353515625,
505
+ "p25": 0.0352828173828125,
506
+ "p75": 0.03532656005859375,
507
+ "p90": 0.035350365234375,
508
+ "p95": 0.03535830029296875,
509
+ "p99": 0.03536464833984375,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 96.14606741573034,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 88,
517
+ "gpu_memory_used_mean": 13849,
518
+ "gpu_memory_used_max": 13849,
519
+ "gpu_memory_used_min": 13849,
520
+ "sample_count": 89,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-09-26T16:46:34.287615",
528
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
529
+ "hardware_info": {
530
+ "gpu_name": "NVIDIA A10G",
531
+ "gpu_memory_total_mb": 23028,
532
+ "cpu_count": 16,
533
+ "memory_total_mb": 63607,
534
+ "python_version": "3.10.12",
535
+ "torch_version": "2.8.0+cu126",
536
+ "cuda_version": "12.6"
537
+ },
538
+ "config": {
539
+ "name": "eager",
540
+ "model_id": "meta-llama/Llama-2-7b-hf",
541
+ "variant": "eager",
542
+ "warmup_iterations": 3,
543
+ "measurement_iterations": 5,
544
+ "num_tokens_to_generate": 100,
545
+ "device": "cuda",
546
+ "torch_dtype": "float16",
547
+ "compile_mode": null,
548
+ "compile_options": {},
549
+ "use_cache": true,
550
+ "batch_size": 1,
551
+ "sequence_length": null,
552
+ "attn_implementation": "sdpa",
553
+ "sdpa_backend": "flash_attention",
554
+ "custom_params": {}
555
+ }
556
+ },
557
+ "measurements": {
558
+ "latency_seconds": {
559
+ "name": "latency_seconds",
560
+ "measurements": [
561
+ 3.404427001953125,
562
+ 3.406873291015625,
563
+ 3.41010693359375,
564
+ 3.40758984375,
565
+ 3.4124208984375
566
+ ],
567
+ "mean": 3.4082835937500002,
568
+ "median": 3.40758984375,
569
+ "std": 0.002749014420326911,
570
+ "min": 3.404427001953125,
571
+ "max": 3.4124208984375,
572
+ "p25": 3.406873291015625,
573
+ "p75": 3.41010693359375,
574
+ "p90": 3.4114953125,
575
+ "p95": 3.41195810546875,
576
+ "p99": 3.41232833984375,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.035428768157958986,
583
+ 0.034643455505371096,
584
+ 0.03465542221069336,
585
+ 0.03479132843017578,
586
+ 0.03462128067016602
587
+ ],
588
+ "mean": 0.034828050994873044,
589
+ "median": 0.03465542221069336,
590
+ "std": 0.00030621555017184757,
591
+ "min": 0.03462128067016602,
592
+ "max": 0.035428768157958986,
593
+ "p25": 0.034643455505371096,
594
+ "p75": 0.03479132843017578,
595
+ "p90": 0.0351737922668457,
596
+ "p95": 0.03530128021240234,
597
+ "p99": 0.03540327056884766,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.37351863988561,
604
+ 29.35242712539771,
605
+ 29.324593611677372,
606
+ 29.346254856174696,
607
+ 29.304708585564168
608
+ ],
609
+ "mean": 29.340300563739913,
610
+ "median": 29.346254856174696,
611
+ "std": 0.023661973427498824,
612
+ "min": 29.304708585564168,
613
+ "max": 29.37351863988561,
614
+ "p25": 29.324593611677372,
615
+ "p75": 29.35242712539771,
616
+ "p90": 29.36508203409045,
617
+ "p95": 29.36930033698803,
618
+ "p99": 29.372674979306094,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.034044270019531246,
625
+ 0.034068732910156246,
626
+ 0.034101069335937505,
627
+ 0.0340758984375,
628
+ 0.034124208984375
629
+ ],
630
+ "mean": 0.0340828359375,
631
+ "median": 0.0340758984375,
632
+ "std": 2.749014420327089e-05,
633
+ "min": 0.034044270019531246,
634
+ "max": 0.034124208984375,
635
+ "p25": 0.034068732910156246,
636
+ "p75": 0.034101069335937505,
637
+ "p90": 0.034114953125,
638
+ "p95": 0.0341195810546875,
639
+ "p99": 0.0341232833984375,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.93023255813954,
645
+ "gpu_utilization_max": 98,
646
+ "gpu_utilization_min": 94,
647
+ "gpu_memory_used_mean": 13849,
648
+ "gpu_memory_used_max": 13849,
649
+ "gpu_memory_used_min": 13849,
650
+ "sample_count": 86,
651
+ "gpu_monitoring_status": "success"
652
+ }
653
+ },
654
+ {
655
+ "scenario_name": "eager_sdpa_efficient_attention",
656
+ "metadata": {
657
+ "timestamp": "2025-09-26T16:47:06.718077",
658
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
659
+ "hardware_info": {
660
+ "gpu_name": "NVIDIA A10G",
661
+ "gpu_memory_total_mb": 23028,
662
+ "cpu_count": 16,
663
+ "memory_total_mb": 63607,
664
+ "python_version": "3.10.12",
665
+ "torch_version": "2.8.0+cu126",
666
+ "cuda_version": "12.6"
667
+ },
668
+ "config": {
669
+ "name": "eager",
670
+ "model_id": "meta-llama/Llama-2-7b-hf",
671
+ "variant": "eager",
672
+ "warmup_iterations": 3,
673
+ "measurement_iterations": 5,
674
+ "num_tokens_to_generate": 100,
675
+ "device": "cuda",
676
+ "torch_dtype": "float16",
677
+ "compile_mode": null,
678
+ "compile_options": {},
679
+ "use_cache": true,
680
+ "batch_size": 1,
681
+ "sequence_length": null,
682
+ "attn_implementation": "sdpa",
683
+ "sdpa_backend": "efficient_attention",
684
+ "custom_params": {}
685
+ }
686
+ },
687
+ "measurements": {
688
+ "latency_seconds": {
689
+ "name": "latency_seconds",
690
+ "measurements": [
691
+ 3.408900634765625,
692
+ 3.412423828125,
693
+ 3.413165283203125,
694
+ 3.4110166015625,
695
+ 3.41103564453125
696
+ ],
697
+ "mean": 3.4113083984375003,
698
+ "median": 3.41103564453125,
699
+ "std": 0.0014593882574423344,
700
+ "min": 3.408900634765625,
701
+ "max": 3.413165283203125,
702
+ "p25": 3.4110166015625,
703
+ "p75": 3.412423828125,
704
+ "p90": 3.4128687011718752,
705
+ "p95": 3.4130169921875,
706
+ "p99": 3.4131356250000002,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03523984146118164,
713
+ 0.0347523193359375,
714
+ 0.03471257781982422,
715
+ 0.03472208023071289,
716
+ 0.034785087585449216
717
+ ],
718
+ "mean": 0.034842381286621094,
719
+ "median": 0.0347523193359375,
720
+ "std": 0.00020034450207666088,
721
+ "min": 0.03471257781982422,
722
+ "max": 0.03523984146118164,
723
+ "p25": 0.03472208023071289,
724
+ "p75": 0.034785087585449216,
725
+ "p90": 0.03505793991088867,
726
+ "p95": 0.035148890686035154,
727
+ "p99": 0.035221651306152346,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.334970629578173,
734
+ 29.304683426427214,
735
+ 29.29831745685454,
736
+ 29.316773173778323,
737
+ 29.31660950548119
738
+ ],
739
+ "mean": 29.314270838423887,
740
+ "median": 29.31660950548119,
741
+ "std": 0.012543055863330294,
742
+ "min": 29.29831745685454,
743
+ "max": 29.334970629578173,
744
+ "p25": 29.304683426427214,
745
+ "p75": 29.316773173778323,
746
+ "p90": 29.327691647258234,
747
+ "p95": 29.331331138418204,
748
+ "p99": 29.33424273134618,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.03408900634765625,
755
+ 0.034124238281250004,
756
+ 0.034131652832031255,
757
+ 0.034110166015625,
758
+ 0.034110356445312504
759
+ ],
760
+ "mean": 0.034113083984375,
761
+ "median": 0.034110356445312504,
762
+ "std": 1.4593882574424714e-05,
763
+ "min": 0.03408900634765625,
764
+ "max": 0.034131652832031255,
765
+ "p25": 0.034110166015625,
766
+ "p75": 0.034124238281250004,
767
+ "p90": 0.034128687011718756,
768
+ "p95": 0.034130169921875,
769
+ "p99": 0.03413135625,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.83720930232558,
775
+ "gpu_utilization_max": 98,
776
+ "gpu_utilization_min": 95,
777
+ "gpu_memory_used_mean": 13849,
778
+ "gpu_memory_used_max": 13849,
779
+ "gpu_memory_used_min": 13849,
780
+ "sample_count": 86,
781
+ "gpu_monitoring_status": "success"
782
+ }
783
+ },
784
+ {
785
+ "scenario_name": "compiled_compile_max-autotune_sdpa_default",
786
+ "metadata": {
787
+ "timestamp": "2025-09-26T16:47:39.141216",
788
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
789
+ "hardware_info": {
790
+ "gpu_name": "NVIDIA A10G",
791
+ "gpu_memory_total_mb": 23028,
792
+ "cpu_count": 16,
793
+ "memory_total_mb": 63607,
794
+ "python_version": "3.10.12",
795
+ "torch_version": "2.8.0+cu126",
796
+ "cuda_version": "12.6"
797
+ },
798
+ "config": {
799
+ "name": "compiled",
800
+ "model_id": "meta-llama/Llama-2-7b-hf",
801
+ "variant": "compiled",
802
+ "warmup_iterations": 3,
803
+ "measurement_iterations": 5,
804
+ "num_tokens_to_generate": 100,
805
+ "device": "cuda",
806
+ "torch_dtype": "float16",
807
+ "compile_mode": "max-autotune",
808
+ "compile_options": {},
809
+ "use_cache": true,
810
+ "batch_size": 1,
811
+ "sequence_length": null,
812
+ "attn_implementation": "sdpa",
813
+ "sdpa_backend": null,
814
+ "custom_params": {}
815
+ }
816
+ },
817
+ "measurements": {
818
+ "latency_seconds": {
819
+ "name": "latency_seconds",
820
+ "measurements": [
821
+ 3.704238037109375,
822
+ 3.698341796875,
823
+ 3.69627490234375,
824
+ 3.6978251953125,
825
+ 3.695230712890625
826
+ ],
827
+ "mean": 3.6983821289062506,
828
+ "median": 3.6978251953125,
829
+ "std": 0.0031297057137456142,
830
+ "min": 3.695230712890625,
831
+ "max": 3.704238037109375,
832
+ "p25": 3.69627490234375,
833
+ "p75": 3.698341796875,
834
+ "p90": 3.7018795410156247,
835
+ "p95": 3.7030587890624997,
836
+ "p99": 3.7040021875,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03787689590454101,
843
+ 0.03779420852661133,
844
+ 0.037839424133300784,
845
+ 0.037473377227783204,
846
+ 0.03735923385620117
847
+ ],
848
+ "mean": 0.037668627929687504,
849
+ "median": 0.03779420852661133,
850
+ "std": 0.00021079132247064732,
851
+ "min": 0.03735923385620117,
852
+ "max": 0.03787689590454101,
853
+ "p25": 0.037473377227783204,
854
+ "p75": 0.037839424133300784,
855
+ "p90": 0.037861907196044924,
856
+ "p95": 0.03786940155029297,
857
+ "p99": 0.0378753970336914,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 26.996105271365234,
864
+ 27.039144971537603,
865
+ 27.054264804977457,
866
+ 27.04292245257123,
867
+ 27.061909734392245
868
+ ],
869
+ "mean": 27.038869446968754,
870
+ "median": 27.04292245257123,
871
+ "std": 0.02286109960620271,
872
+ "min": 26.996105271365234,
873
+ "max": 27.061909734392245,
874
+ "p25": 27.039144971537603,
875
+ "p75": 27.054264804977457,
876
+ "p90": 27.05885176262633,
877
+ "p95": 27.06038074850929,
878
+ "p99": 27.061603937215654,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.037042380371093746,
885
+ 0.03698341796875,
886
+ 0.0369627490234375,
887
+ 0.036978251953125,
888
+ 0.03695230712890625
889
+ ],
890
+ "mean": 0.03698382128906249,
891
+ "median": 0.036978251953125,
892
+ "std": 3.129705713745548e-05,
893
+ "min": 0.03695230712890625,
894
+ "max": 0.037042380371093746,
895
+ "p25": 0.0369627490234375,
896
+ "p75": 0.03698341796875,
897
+ "p90": 0.03701879541015625,
898
+ "p95": 0.037030587890624994,
899
+ "p99": 0.037040021874999995,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.52688172043011,
905
+ "gpu_utilization_max": 93,
906
+ "gpu_utilization_min": 87,
907
+ "gpu_memory_used_mean": 13845,
908
+ "gpu_memory_used_max": 13845,
909
+ "gpu_memory_used_min": 13845,
910
+ "sample_count": 93,
911
+ "gpu_monitoring_status": "success"
912
+ }
913
+ },
914
+ {
915
+ "scenario_name": "compiled_compile_max-autotune_sdpa_math",
916
+ "metadata": {
917
+ "timestamp": "2025-09-26T16:48:13.866576",
918
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
919
+ "hardware_info": {
920
+ "gpu_name": "NVIDIA A10G",
921
+ "gpu_memory_total_mb": 23028,
922
+ "cpu_count": 16,
923
+ "memory_total_mb": 63607,
924
+ "python_version": "3.10.12",
925
+ "torch_version": "2.8.0+cu126",
926
+ "cuda_version": "12.6"
927
+ },
928
+ "config": {
929
+ "name": "compiled",
930
+ "model_id": "meta-llama/Llama-2-7b-hf",
931
+ "variant": "compiled",
932
+ "warmup_iterations": 3,
933
+ "measurement_iterations": 5,
934
+ "num_tokens_to_generate": 100,
935
+ "device": "cuda",
936
+ "torch_dtype": "float16",
937
+ "compile_mode": "max-autotune",
938
+ "compile_options": {},
939
+ "use_cache": true,
940
+ "batch_size": 1,
941
+ "sequence_length": null,
942
+ "attn_implementation": "sdpa",
943
+ "sdpa_backend": "math",
944
+ "custom_params": {}
945
+ }
946
+ },
947
+ "measurements": {
948
+ "latency_seconds": {
949
+ "name": "latency_seconds",
950
+ "measurements": [
951
+ 3.875684814453125,
952
+ 3.94908349609375,
953
+ 3.96232568359375,
954
+ 3.92354052734375,
955
+ 3.88472802734375
956
+ ],
957
+ "mean": 3.9190725097656247,
958
+ "median": 3.92354052734375,
959
+ "std": 0.03421547926020662,
960
+ "min": 3.875684814453125,
961
+ "max": 3.96232568359375,
962
+ "p25": 3.88472802734375,
963
+ "p75": 3.94908349609375,
964
+ "p90": 3.95702880859375,
965
+ "p95": 3.95967724609375,
966
+ "p99": 3.96179599609375,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.04063577651977539,
973
+ 0.04220953750610352,
974
+ 0.041759265899658206,
975
+ 0.04347548675537109,
976
+ 0.04051670455932617
977
+ ],
978
+ "mean": 0.04171935424804687,
979
+ "median": 0.041759265899658206,
980
+ "std": 0.001090539400175915,
981
+ "min": 0.04051670455932617,
982
+ "max": 0.04347548675537109,
983
+ "p25": 0.04063577651977539,
984
+ "p75": 0.04220953750610352,
985
+ "p90": 0.04296910705566406,
986
+ "p95": 0.04322229690551758,
987
+ "p99": 0.04342484878540039,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.80189173977255,
994
+ 25.322331143141277,
995
+ 25.237703304919147,
996
+ 25.487184165190804,
997
+ 25.741827818092258
998
+ ],
999
+ "mean": 25.518187634223207,
1000
+ "median": 25.487184165190804,
1001
+ "std": 0.2229329004533217,
1002
+ "min": 25.237703304919147,
1003
+ "max": 25.80189173977255,
1004
+ "p25": 25.322331143141277,
1005
+ "p75": 25.741827818092258,
1006
+ "p90": 25.777866171100435,
1007
+ "p95": 25.789878955436492,
1008
+ "p99": 25.79948918290534,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.03875684814453125,
1015
+ 0.0394908349609375,
1016
+ 0.0396232568359375,
1017
+ 0.0392354052734375,
1018
+ 0.0388472802734375
1019
+ ],
1020
+ "mean": 0.039190725097656254,
1021
+ "median": 0.0392354052734375,
1022
+ "std": 0.00034215479260206645,
1023
+ "min": 0.03875684814453125,
1024
+ "max": 0.0396232568359375,
1025
+ "p25": 0.0388472802734375,
1026
+ "p75": 0.0394908349609375,
1027
+ "p90": 0.0395702880859375,
1028
+ "p95": 0.0395967724609375,
1029
+ "p99": 0.039617959960937504,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 88.06060606060606,
1035
+ "gpu_utilization_max": 93,
1036
+ "gpu_utilization_min": 82,
1037
+ "gpu_memory_used_mean": 13845,
1038
+ "gpu_memory_used_max": 13845,
1039
+ "gpu_memory_used_min": 13845,
1040
+ "sample_count": 99,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-09-26T16:48:55.045297",
1048
+ "commit_id": "071eb5334f5a9ac2c7a13515219be8a272388ec6",
1049
+ "hardware_info": {
1050
+ "gpu_name": "NVIDIA A10G",
1051
+ "gpu_memory_total_mb": 23028,
1052
+ "cpu_count": 16,
1053
+ "memory_total_mb": 63607,
1054
+ "python_version": "3.10.12",
1055
+ "torch_version": "2.8.0+cu126",
1056
+ "cuda_version": "12.6"
1057
+ },
1058
+ "config": {
1059
+ "name": "compiled",
1060
+ "model_id": "meta-llama/Llama-2-7b-hf",
1061
+ "variant": "compiled",
1062
+ "warmup_iterations": 3,
1063
+ "measurement_iterations": 5,
1064
+ "num_tokens_to_generate": 100,
1065
+ "device": "cuda",
1066
+ "torch_dtype": "float16",
1067
+ "compile_mode": "max-autotune",
1068
+ "compile_options": {},
1069
+ "use_cache": true,
1070
+ "batch_size": 1,
1071
+ "sequence_length": null,
1072
+ "attn_implementation": "sdpa",
1073
+ "sdpa_backend": "efficient_attention",
1074
+ "custom_params": {}
1075
+ }
1076
+ },
1077
+ "measurements": {
1078
+ "latency_seconds": {
1079
+ "name": "latency_seconds",
1080
+ "measurements": [
1081
+ 3.69835791015625,
1082
+ 3.701083740234375,
1083
+ 3.68698193359375,
1084
+ 3.696963134765625,
1085
+ 3.694296630859375
1086
+ ],
1087
+ "mean": 3.695536669921874,
1088
+ "median": 3.696963134765625,
1089
+ "std": 0.004805942113313335,
1090
+ "min": 3.68698193359375,
1091
+ "max": 3.701083740234375,
1092
+ "p25": 3.694296630859375,
1093
+ "p75": 3.69835791015625,
1094
+ "p90": 3.6999934082031247,
1095
+ "p95": 3.7005385742187498,
1096
+ "p99": 3.70097470703125,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.03811468887329102,
1103
+ 0.03744339370727539,
1104
+ 0.03744377517700195,
1105
+ 0.037391456604003906,
1106
+ 0.037659870147705075
1107
+ ],
1108
+ "mean": 0.03761063690185547,
1109
+ "median": 0.03744377517700195,
1110
+ "std": 0.0002684568593743473,
1111
+ "min": 0.037391456604003906,
1112
+ "max": 0.03811468887329102,
1113
+ "p25": 0.03744339370727539,
1114
+ "p75": 0.037659870147705075,
1115
+ "p90": 0.03793276138305664,
1116
+ "p95": 0.03802372512817383,
1117
+ "p99": 0.038096496124267576,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.039027165376524,
1124
+ 27.019113054077344,
1125
+ 27.122454571544015,
1126
+ 27.049228340855407,
1127
+ 27.068752185375484
1128
+ ],
1129
+ "mean": 27.059715063445754,
1130
+ "median": 27.049228340855407,
1131
+ "std": 0.03522612408751987,
1132
+ "min": 27.019113054077344,
1133
+ "max": 27.122454571544015,
1134
+ "p25": 27.039027165376524,
1135
+ "p75": 27.068752185375484,
1136
+ "p90": 27.1009736170766,
1137
+ "p95": 27.111714094310308,
1138
+ "p99": 27.120306476097273,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.0369835791015625,
1145
+ 0.037010837402343746,
1146
+ 0.0368698193359375,
1147
+ 0.036969631347656254,
1148
+ 0.03694296630859375
1149
+ ],
1150
+ "mean": 0.03695536669921875,
1151
+ "median": 0.036969631347656254,
1152
+ "std": 4.8059421133133996e-05,
1153
+ "min": 0.0368698193359375,
1154
+ "max": 0.037010837402343746,
1155
+ "p25": 0.03694296630859375,
1156
+ "p75": 0.0369835791015625,
1157
+ "p90": 0.03699993408203125,
1158
+ "p95": 0.037005385742187494,
1159
+ "p99": 0.037009747070312494,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.35483870967742,
1165
+ "gpu_utilization_max": 93,
1166
+ "gpu_utilization_min": 86,
1167
+ "gpu_memory_used_mean": 13845,
1168
+ "gpu_memory_used_max": 13845,
1169
+ "gpu_memory_used_min": 13845,
1170
+ "sample_count": 93,
1171
+ "gpu_monitoring_status": "success"
1172
+ }
1173
+ }
1174
+ ]
1175
+ }