hf-transformers-bot commited on
Commit
6473907
·
verified ·
1 Parent(s): c27d1ec

Upload benchmark results for run 18287931791

Browse files
2025-10-06/18287931791/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20251006_164915.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": "2025-10-06T16:43:12.763782",
8
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 3.66782861328125,
42
+ 3.66227490234375,
43
+ 3.662161376953125,
44
+ 3.65299658203125,
45
+ 3.649034912109375
46
+ ],
47
+ "mean": 3.65885927734375,
48
+ "median": 3.662161376953125,
49
+ "std": 0.0068397177093238015,
50
+ "min": 3.649034912109375,
51
+ "max": 3.66782861328125,
52
+ "p25": 3.65299658203125,
53
+ "p75": 3.66227490234375,
54
+ "p90": 3.6656071289062497,
55
+ "p95": 3.66671787109375,
56
+ "p99": 3.66760646484375,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.038629119873046874,
63
+ 0.037889503479003904,
64
+ 0.03714307022094727,
65
+ 0.03750191879272461,
66
+ 0.03713206481933594
67
+ ],
68
+ "mean": 0.037659135437011715,
69
+ "median": 0.03750191879272461,
70
+ "std": 0.000559067626931023,
71
+ "min": 0.03713206481933594,
72
+ "max": 0.038629119873046874,
73
+ "p25": 0.03714307022094727,
74
+ "p75": 0.037889503479003904,
75
+ "p90": 0.03833327331542968,
76
+ "p95": 0.03848119659423828,
77
+ "p99": 0.03859953521728515,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.264087432520387,
84
+ 27.305432461119423,
85
+ 27.306278917506038,
86
+ 27.374786084358984,
87
+ 27.404506234826243
88
+ ],
89
+ "mean": 27.331018226066213,
90
+ "median": 27.306278917506038,
91
+ "std": 0.05111263783385129,
92
+ "min": 27.264087432520387,
93
+ "max": 27.404506234826243,
94
+ "p25": 27.305432461119423,
95
+ "p75": 27.374786084358984,
96
+ "p90": 27.39261817463934,
97
+ "p95": 27.39856220473279,
98
+ "p99": 27.403317428807554,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.0366782861328125,
105
+ 0.0366227490234375,
106
+ 0.03662161376953125,
107
+ 0.0365299658203125,
108
+ 0.03649034912109375
109
+ ],
110
+ "mean": 0.0365885927734375,
111
+ "median": 0.03662161376953125,
112
+ "std": 6.839717709323804e-05,
113
+ "min": 0.03649034912109375,
114
+ "max": 0.0366782861328125,
115
+ "p25": 0.0365299658203125,
116
+ "p75": 0.0366227490234375,
117
+ "p90": 0.0366560712890625,
118
+ "p95": 0.0366671787109375,
119
+ "p99": 0.0366760646484375,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 90.75,
125
+ "gpu_utilization_max": 93,
126
+ "gpu_utilization_min": 88,
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": "2025-10-06T16:43:48.055187",
138
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 7.81440380859375,
172
+ 7.837978515625,
173
+ 7.89311181640625,
174
+ 7.86814453125,
175
+ 7.924005859375
176
+ ],
177
+ "mean": 7.86752890625,
178
+ "median": 7.86814453125,
179
+ "std": 0.038833006856443746,
180
+ "min": 7.81440380859375,
181
+ "max": 7.924005859375,
182
+ "p25": 7.837978515625,
183
+ "p75": 7.89311181640625,
184
+ "p90": 7.9116482421875,
185
+ "p95": 7.91782705078125,
186
+ "p99": 7.92277009765625,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.038861984252929686,
193
+ 0.03815657424926758,
194
+ 0.03859772872924805,
195
+ 0.038547454833984376,
196
+ 0.038291679382324216
197
+ ],
198
+ "mean": 0.03849108428955079,
199
+ "median": 0.038547454833984376,
200
+ "std": 0.00024648600029570153,
201
+ "min": 0.03815657424926758,
202
+ "max": 0.038861984252929686,
203
+ "p25": 0.038291679382324216,
204
+ "p75": 0.03859772872924805,
205
+ "p90": 0.03875628204345703,
206
+ "p95": 0.038809133148193356,
207
+ "p99": 0.03885141403198242,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.796881559924866,
214
+ 12.758391695084406,
215
+ 12.669274466902232,
216
+ 12.709476751835055,
217
+ 12.619879613249987
218
+ ],
219
+ "mean": 12.710780817399309,
220
+ "median": 12.709476751835055,
221
+ "std": 0.06271784462459846,
222
+ "min": 12.619879613249987,
223
+ "max": 12.796881559924866,
224
+ "p25": 12.669274466902232,
225
+ "p75": 12.758391695084406,
226
+ "p90": 12.781485613988682,
227
+ "p95": 12.789183586956774,
228
+ "p99": 12.795341965331248,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.07814403808593749,
235
+ 0.07837978515625,
236
+ 0.07893111816406251,
237
+ 0.0786814453125,
238
+ 0.07924005859375
239
+ ],
240
+ "mean": 0.0786752890625,
241
+ "median": 0.0786814453125,
242
+ "std": 0.000388330068564438,
243
+ "min": 0.07814403808593749,
244
+ "max": 0.07924005859375,
245
+ "p25": 0.07837978515625,
246
+ "p75": 0.07893111816406251,
247
+ "p90": 0.079116482421875,
248
+ "p95": 0.0791782705078125,
249
+ "p99": 0.0792277009765625,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 44.62721893491124,
255
+ "gpu_utilization_max": 94,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13629.899408284024,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 169,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-10-06T16:45:15.984268",
268
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 3.411918212890625,
302
+ 3.4121884765625,
303
+ 3.4159677734375,
304
+ 3.410937255859375,
305
+ 3.412314208984375
306
+ ],
307
+ "mean": 3.4126651855468753,
308
+ "median": 3.4121884765625,
309
+ "std": 0.0017205366565487329,
310
+ "min": 3.410937255859375,
311
+ "max": 3.4159677734375,
312
+ "p25": 3.411918212890625,
313
+ "p75": 3.412314208984375,
314
+ "p90": 3.41450634765625,
315
+ "p95": 3.415237060546875,
316
+ "p99": 3.415821630859375,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.03511264038085937,
323
+ 0.03463030242919922,
324
+ 0.03458262252807617,
325
+ 0.0345720329284668,
326
+ 0.03466940689086914
327
+ ],
328
+ "mean": 0.03471340103149414,
329
+ "median": 0.03463030242919922,
330
+ "std": 0.00020264297555260458,
331
+ "min": 0.0345720329284668,
332
+ "max": 0.03511264038085937,
333
+ "p25": 0.03458262252807617,
334
+ "p75": 0.03466940689086914,
335
+ "p90": 0.03493534698486328,
336
+ "p95": 0.035023993682861326,
337
+ "p99": 0.035094911041259766,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.30902611387,
344
+ 29.306704681431253,
345
+ 29.2742808575649,
346
+ 29.317455144687294,
347
+ 29.305624826901134
348
+ ],
349
+ "mean": 29.302618324890915,
350
+ "median": 29.306704681431253,
351
+ "std": 0.014764419391468749,
352
+ "min": 29.2742808575649,
353
+ "max": 29.317455144687294,
354
+ "p25": 29.305624826901134,
355
+ "p75": 29.30902611387,
356
+ "p90": 29.314083532360378,
357
+ "p95": 29.315769338523836,
358
+ "p99": 29.3171179834546,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.03411918212890625,
365
+ 0.034121884765625,
366
+ 0.034159677734375,
367
+ 0.034109372558593753,
368
+ 0.03412314208984375
369
+ ],
370
+ "mean": 0.034126651855468745,
371
+ "median": 0.034121884765625,
372
+ "std": 1.720536656548591e-05,
373
+ "min": 0.034109372558593753,
374
+ "max": 0.034159677734375,
375
+ "p25": 0.03411918212890625,
376
+ "p75": 0.03412314208984375,
377
+ "p90": 0.0341450634765625,
378
+ "p95": 0.034152370605468746,
379
+ "p99": 0.034158216308593746,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.83720930232558,
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": "2025-10-06T16:45:48.524493",
398
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 3.537967041015625,
432
+ 3.555361083984375,
433
+ 3.536182861328125,
434
+ 3.541820068359375,
435
+ 3.549981689453125
436
+ ],
437
+ "mean": 3.544262548828125,
438
+ "median": 3.541820068359375,
439
+ "std": 0.007304334763113905,
440
+ "min": 3.536182861328125,
441
+ "max": 3.555361083984375,
442
+ "p25": 3.537967041015625,
443
+ "p75": 3.549981689453125,
444
+ "p90": 3.553209326171875,
445
+ "p95": 3.554285205078125,
446
+ "p99": 3.555145908203125,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.039270656585693356,
453
+ 0.03835145568847656,
454
+ 0.038335872650146485,
455
+ 0.038448734283447264,
456
+ 0.038378017425537106
457
+ ],
458
+ "mean": 0.03855694732666015,
459
+ "median": 0.038378017425537106,
460
+ "std": 0.00035894503886417517,
461
+ "min": 0.038335872650146485,
462
+ "max": 0.039270656585693356,
463
+ "p25": 0.03835145568847656,
464
+ "p75": 0.038448734283447264,
465
+ "p90": 0.03894188766479492,
466
+ "p95": 0.03910627212524414,
467
+ "p99": 0.039237779693603515,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.264819553348225,
474
+ 28.126538384656367,
475
+ 28.279080557061985,
476
+ 28.234071203487623,
477
+ 28.169159378229075
478
+ ],
479
+ "mean": 28.214733815356652,
480
+ "median": 28.234071203487623,
481
+ "std": 0.05810074390898159,
482
+ "min": 28.126538384656367,
483
+ "max": 28.279080557061985,
484
+ "p25": 28.169159378229075,
485
+ "p75": 28.264819553348225,
486
+ "p90": 28.273376155576482,
487
+ "p95": 28.276228356319233,
488
+ "p99": 28.278510116913434,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.03537967041015625,
495
+ 0.03555361083984375,
496
+ 0.03536182861328125,
497
+ 0.03541820068359375,
498
+ 0.03549981689453125
499
+ ],
500
+ "mean": 0.03544262548828125,
501
+ "median": 0.03541820068359375,
502
+ "std": 7.304334763113915e-05,
503
+ "min": 0.03536182861328125,
504
+ "max": 0.03555361083984375,
505
+ "p25": 0.03537967041015625,
506
+ "p75": 0.03549981689453125,
507
+ "p90": 0.03553209326171875,
508
+ "p95": 0.03554285205078125,
509
+ "p99": 0.03555145908203125,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 95.87640449438203,
515
+ "gpu_utilization_max": 98,
516
+ "gpu_utilization_min": 91,
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": "2025-10-06T16:46:22.051057",
528
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 3.414234375,
562
+ 3.4108564453125,
563
+ 3.411046630859375,
564
+ 3.414957275390625,
565
+ 3.412583251953125
566
+ ],
567
+ "mean": 3.4127355957031247,
568
+ "median": 3.412583251953125,
569
+ "std": 0.0016485883092369948,
570
+ "min": 3.4108564453125,
571
+ "max": 3.414957275390625,
572
+ "p25": 3.411046630859375,
573
+ "p75": 3.414234375,
574
+ "p90": 3.414668115234375,
575
+ "p95": 3.4148126953125,
576
+ "p99": 3.414928359375,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.035526496887207035,
583
+ 0.03465359878540039,
584
+ 0.0347196159362793,
585
+ 0.03466038513183594,
586
+ 0.034675521850585936
587
+ ],
588
+ "mean": 0.03484712371826172,
589
+ "median": 0.034675521850585936,
590
+ "std": 0.0003404633717268198,
591
+ "min": 0.03465359878540039,
592
+ "max": 0.035526496887207035,
593
+ "p25": 0.03466038513183594,
594
+ "p75": 0.0347196159362793,
595
+ "p90": 0.03520374450683594,
596
+ "p95": 0.03536512069702149,
597
+ "p99": 0.035494221649169926,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.28914333832164,
604
+ 29.31814973844145,
605
+ 29.316515082294877,
606
+ 29.28294322176003,
607
+ 29.303314415191764
608
+ ],
609
+ "mean": 29.302013159201955,
610
+ "median": 29.303314415191764,
611
+ "std": 0.014154031465465093,
612
+ "min": 29.28294322176003,
613
+ "max": 29.31814973844145,
614
+ "p25": 29.28914333832164,
615
+ "p75": 29.316515082294877,
616
+ "p90": 29.31749587598282,
617
+ "p95": 29.317822807212135,
618
+ "p99": 29.318084352195587,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.03414234375,
625
+ 0.034108564453125004,
626
+ 0.034110466308593754,
627
+ 0.03414957275390625,
628
+ 0.03412583251953125
629
+ ],
630
+ "mean": 0.03412735595703125,
631
+ "median": 0.03412583251953125,
632
+ "std": 1.648588309236897e-05,
633
+ "min": 0.034108564453125004,
634
+ "max": 0.03414957275390625,
635
+ "p25": 0.034110466308593754,
636
+ "p75": 0.03414234375,
637
+ "p90": 0.034146681152343754,
638
+ "p95": 0.034148126953125,
639
+ "p99": 0.03414928359375,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.77906976744185,
645
+ "gpu_utilization_max": 98,
646
+ "gpu_utilization_min": 92,
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": "2025-10-06T16:46:54.493601",
658
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 3.40857421875,
692
+ 3.406154296875,
693
+ 3.4098876953125,
694
+ 3.408439697265625,
695
+ 3.408717041015625
696
+ ],
697
+ "mean": 3.40835458984375,
698
+ "median": 3.40857421875,
699
+ "std": 0.0012147882538412742,
700
+ "min": 3.406154296875,
701
+ "max": 3.4098876953125,
702
+ "p25": 3.408439697265625,
703
+ "p75": 3.408717041015625,
704
+ "p90": 3.40941943359375,
705
+ "p95": 3.4096535644531247,
706
+ "p99": 3.409840869140625,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03524095916748047,
713
+ 0.03473478317260742,
714
+ 0.03470156860351563,
715
+ 0.034695232391357425,
716
+ 0.03470147323608398
717
+ ],
718
+ "mean": 0.034814803314208984,
719
+ "median": 0.03470156860351563,
720
+ "std": 0.00021352987905428247,
721
+ "min": 0.034695232391357425,
722
+ "max": 0.03524095916748047,
723
+ "p25": 0.03470147323608398,
724
+ "p75": 0.03473478317260742,
725
+ "p90": 0.03503848876953125,
726
+ "p95": 0.03513972396850586,
727
+ "p99": 0.035220712127685544,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.33777984059042,
734
+ 29.358623034706824,
735
+ 29.326479032569864,
736
+ 29.33893771986744,
737
+ 29.33655061324922
738
+ ],
739
+ "mean": 29.33967404819676,
740
+ "median": 29.33777984059042,
741
+ "std": 0.010460005496065395,
742
+ "min": 29.326479032569864,
743
+ "max": 29.358623034706824,
744
+ "p25": 29.33655061324922,
745
+ "p75": 29.33893771986744,
746
+ "p90": 29.35074890877107,
747
+ "p95": 29.35468597173895,
748
+ "p99": 29.35783562211325,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.034085742187500004,
755
+ 0.03406154296875,
756
+ 0.034098876953124996,
757
+ 0.03408439697265625,
758
+ 0.03408717041015625
759
+ ],
760
+ "mean": 0.0340835458984375,
761
+ "median": 0.034085742187500004,
762
+ "std": 1.2147882538413257e-05,
763
+ "min": 0.03406154296875,
764
+ "max": 0.034098876953124996,
765
+ "p25": 0.03408439697265625,
766
+ "p75": 0.03408717041015625,
767
+ "p90": 0.034094194335937494,
768
+ "p95": 0.034096535644531245,
769
+ "p99": 0.03409840869140625,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.83720930232558,
775
+ "gpu_utilization_max": 97,
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": "2025-10-06T16:47:26.907161",
788
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 3.6801875,
822
+ 3.689103271484375,
823
+ 3.6923779296875,
824
+ 3.69780615234375,
825
+ 3.710135986328125
826
+ ],
827
+ "mean": 3.6939221679687497,
828
+ "median": 3.6923779296875,
829
+ "std": 0.009921902839921977,
830
+ "min": 3.6801875,
831
+ "max": 3.710135986328125,
832
+ "p25": 3.689103271484375,
833
+ "p75": 3.69780615234375,
834
+ "p90": 3.7052040527343753,
835
+ "p95": 3.70767001953125,
836
+ "p99": 3.7096427929687503,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.038386558532714846,
843
+ 0.03741334533691406,
844
+ 0.03743203353881836,
845
+ 0.03743113708496094,
846
+ 0.03742540740966797
847
+ ],
848
+ "mean": 0.037617696380615226,
849
+ "median": 0.03743113708496094,
850
+ "std": 0.00038448887317062434,
851
+ "min": 0.03741334533691406,
852
+ "max": 0.038386558532714846,
853
+ "p25": 0.03742540740966797,
854
+ "p75": 0.03743203353881836,
855
+ "p90": 0.03800474853515625,
856
+ "p95": 0.03819565353393555,
857
+ "p99": 0.038348377532958985,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.172528573612077,
864
+ 27.106858399158682,
865
+ 27.08281814707504,
866
+ 27.04306171826174,
867
+ 26.953189955435768
868
+ ],
869
+ "mean": 27.07169135870866,
870
+ "median": 27.08281814707504,
871
+ "std": 0.07265109627886715,
872
+ "min": 26.953189955435768,
873
+ "max": 27.172528573612077,
874
+ "p25": 27.04306171826174,
875
+ "p75": 27.106858399158682,
876
+ "p90": 27.146260503830717,
877
+ "p95": 27.159394538721397,
878
+ "p99": 27.16990176663394,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.036801875000000005,
885
+ 0.03689103271484375,
886
+ 0.036923779296875,
887
+ 0.0369780615234375,
888
+ 0.03710135986328125
889
+ ],
890
+ "mean": 0.0369392216796875,
891
+ "median": 0.036923779296875,
892
+ "std": 9.921902839921905e-05,
893
+ "min": 0.036801875000000005,
894
+ "max": 0.03710135986328125,
895
+ "p25": 0.03689103271484375,
896
+ "p75": 0.0369780615234375,
897
+ "p90": 0.03705204052734375,
898
+ "p95": 0.0370767001953125,
899
+ "p99": 0.0370964279296875,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.55913978494624,
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": "2025-10-06T16:48:01.997740",
918
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 4.07048779296875,
952
+ 4.06689794921875,
953
+ 4.0762255859375,
954
+ 4.013225830078125,
955
+ 4.0175888671875
956
+ ],
957
+ "mean": 4.0488852050781245,
958
+ "median": 4.06689794921875,
959
+ "std": 0.027530633742572228,
960
+ "min": 4.013225830078125,
961
+ "max": 4.0762255859375,
962
+ "p25": 4.0175888671875,
963
+ "p75": 4.07048779296875,
964
+ "p90": 4.07393046875,
965
+ "p95": 4.07507802734375,
966
+ "p99": 4.07599607421875,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.043216480255126956,
973
+ 0.04252191925048828,
974
+ 0.042657535552978514,
975
+ 0.042356800079345704,
976
+ 0.04250777435302734
977
+ ],
978
+ "mean": 0.042652101898193365,
979
+ "median": 0.04252191925048828,
980
+ "std": 0.0002978367273024861,
981
+ "min": 0.042356800079345704,
982
+ "max": 0.043216480255126956,
983
+ "p25": 0.04250777435302734,
984
+ "p75": 0.042657535552978514,
985
+ "p90": 0.04299290237426758,
986
+ "p95": 0.04310469131469727,
987
+ "p99": 0.04319412246704102,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 24.567080184526603,
994
+ 24.5887655034988,
995
+ 24.53249897282139,
996
+ 24.917610977813656,
997
+ 24.89055085171138
998
+ ],
999
+ "mean": 24.699301298074367,
1000
+ "median": 24.5887655034988,
1001
+ "std": 0.16837999197903,
1002
+ "min": 24.53249897282139,
1003
+ "max": 24.917610977813656,
1004
+ "p25": 24.567080184526603,
1005
+ "p75": 24.89055085171138,
1006
+ "p90": 24.906786927372746,
1007
+ "p95": 24.9121989525932,
1008
+ "p99": 24.916528572769565,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.0407048779296875,
1015
+ 0.0406689794921875,
1016
+ 0.040762255859375006,
1017
+ 0.040132258300781244,
1018
+ 0.04017588867187501
1019
+ ],
1020
+ "mean": 0.04048885205078125,
1021
+ "median": 0.0406689794921875,
1022
+ "std": 0.00027530633742572317,
1023
+ "min": 0.040132258300781244,
1024
+ "max": 0.040762255859375006,
1025
+ "p25": 0.04017588867187501,
1026
+ "p75": 0.0407048779296875,
1027
+ "p90": 0.040739304687500005,
1028
+ "p95": 0.0407507802734375,
1029
+ "p99": 0.040759960742187504,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 85.1470588235294,
1035
+ "gpu_utilization_max": 88,
1036
+ "gpu_utilization_min": 80,
1037
+ "gpu_memory_used_mean": 13845,
1038
+ "gpu_memory_used_max": 13845,
1039
+ "gpu_memory_used_min": 13845,
1040
+ "sample_count": 102,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-10-06T16:48:44.147289",
1048
+ "commit_id": "505387c05b9e7dfcd73d8567959f7848525d8a2f",
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
+ 3.69631396484375,
1082
+ 3.702320556640625,
1083
+ 3.70993115234375,
1084
+ 3.6887802734375,
1085
+ 3.691228515625
1086
+ ],
1087
+ "mean": 3.697714892578125,
1088
+ "median": 3.69631396484375,
1089
+ "std": 0.007672186752169484,
1090
+ "min": 3.6887802734375,
1091
+ "max": 3.70993115234375,
1092
+ "p25": 3.691228515625,
1093
+ "p75": 3.702320556640625,
1094
+ "p90": 3.7068869140625003,
1095
+ "p95": 3.708409033203125,
1096
+ "p99": 3.7096267285156252,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.039849151611328126,
1103
+ 0.03747135925292969,
1104
+ 0.037561214447021485,
1105
+ 0.037438335418701174,
1106
+ 0.03761020660400391
1107
+ ],
1108
+ "mean": 0.03798605346679688,
1109
+ "median": 0.037561214447021485,
1110
+ "std": 0.0009335725946222872,
1111
+ "min": 0.037438335418701174,
1112
+ "max": 0.039849151611328126,
1113
+ "p25": 0.03747135925292969,
1114
+ "p75": 0.03761020660400391,
1115
+ "p90": 0.03895357360839844,
1116
+ "p95": 0.03940136260986328,
1117
+ "p99": 0.03975959381103516,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.053978896575465,
1124
+ 27.01008690904307,
1125
+ 26.954678104154297,
1126
+ 27.109231937746188,
1127
+ 27.091251483537036
1128
+ ],
1129
+ "mean": 27.043845466211213,
1130
+ "median": 27.053978896575465,
1131
+ "std": 0.05606401018068766,
1132
+ "min": 26.954678104154297,
1133
+ "max": 27.109231937746188,
1134
+ "p25": 27.01008690904307,
1135
+ "p75": 27.091251483537036,
1136
+ "p90": 27.102039756062528,
1137
+ "p95": 27.105635846904356,
1138
+ "p99": 27.10851271957782,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.036963139648437504,
1145
+ 0.03702320556640625,
1146
+ 0.037099311523437505,
1147
+ 0.036887802734375,
1148
+ 0.03691228515625
1149
+ ],
1150
+ "mean": 0.03697714892578126,
1151
+ "median": 0.036963139648437504,
1152
+ "std": 7.672186752169571e-05,
1153
+ "min": 0.036887802734375,
1154
+ "max": 0.037099311523437505,
1155
+ "p25": 0.03691228515625,
1156
+ "p75": 0.03702320556640625,
1157
+ "p90": 0.037068869140625,
1158
+ "p95": 0.03708409033203126,
1159
+ "p99": 0.037096267285156255,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.3225806451613,
1165
+ "gpu_utilization_max": 93,
1166
+ "gpu_utilization_min": 85,
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
+ }