hf-transformers-bot commited on
Commit
cae0512
·
verified ·
1 Parent(s): 3304e6a

Upload benchmark results for run 18062285374

Browse files
2025-09-27/18062285374/benchmark_results/Llama-2-7b-hf/Llama-2-7b-hf_benchmark_20250927_165003.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": "2025-09-27T16:44:01.989907",
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
+ 3.65716943359375,
42
+ 3.667003662109375,
43
+ 3.66034130859375,
44
+ 3.664185546875,
45
+ 3.6493193359375
46
+ ],
47
+ "mean": 3.6596038574218754,
48
+ "median": 3.66034130859375,
49
+ "std": 0.006131730969741667,
50
+ "min": 3.6493193359375,
51
+ "max": 3.667003662109375,
52
+ "p25": 3.65716943359375,
53
+ "p75": 3.664185546875,
54
+ "p90": 3.665876416015625,
55
+ "p95": 3.6664400390625,
56
+ "p99": 3.6668909375000003,
57
+ "unit": "seconds"
58
+ },
59
+ "time_to_first_token_seconds": {
60
+ "name": "time_to_first_token_seconds",
61
+ "measurements": [
62
+ 0.03755129623413086,
63
+ 0.0372130241394043,
64
+ 0.03703478240966797,
65
+ 0.0389780158996582,
66
+ 0.03832953643798828
67
+ ],
68
+ "mean": 0.037821331024169916,
69
+ "median": 0.03755129623413086,
70
+ "std": 0.000729082631898095,
71
+ "min": 0.03703478240966797,
72
+ "max": 0.0389780158996582,
73
+ "p25": 0.0372130241394043,
74
+ "p75": 0.03832953643798828,
75
+ "p90": 0.03871862411499023,
76
+ "p95": 0.038848320007324216,
77
+ "p99": 0.0389520767211914,
78
+ "unit": "seconds"
79
+ },
80
+ "tokens_per_second": {
81
+ "name": "tokens_per_second",
82
+ "measurements": [
83
+ 27.343551294459473,
84
+ 27.270220925407223,
85
+ 27.31985669347828,
86
+ 27.291194378866805,
87
+ 27.402370358556272
88
+ ],
89
+ "mean": 27.32543873015361,
90
+ "median": 27.31985669347828,
91
+ "std": 0.045824406431351106,
92
+ "min": 27.270220925407223,
93
+ "max": 27.402370358556272,
94
+ "p25": 27.291194378866805,
95
+ "p75": 27.343551294459473,
96
+ "p90": 27.37884273291755,
97
+ "p95": 27.39060654573691,
98
+ "p99": 27.4000175959924,
99
+ "unit": "tokens/sec"
100
+ },
101
+ "time_per_output_token_seconds": {
102
+ "name": "time_per_output_token_seconds",
103
+ "measurements": [
104
+ 0.0365716943359375,
105
+ 0.03667003662109375,
106
+ 0.0366034130859375,
107
+ 0.036641855468750005,
108
+ 0.036493193359374995
109
+ ],
110
+ "mean": 0.036596038574218746,
111
+ "median": 0.0366034130859375,
112
+ "std": 6.131730969741737e-05,
113
+ "min": 0.036493193359374995,
114
+ "max": 0.03667003662109375,
115
+ "p25": 0.0365716943359375,
116
+ "p75": 0.036641855468750005,
117
+ "p90": 0.03665876416015625,
118
+ "p95": 0.036664400390625,
119
+ "p99": 0.036668909374999996,
120
+ "unit": "seconds/token"
121
+ }
122
+ },
123
+ "gpu_metrics": {
124
+ "gpu_utilization_mean": 90.69565217391305,
125
+ "gpu_utilization_max": 94,
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": "2025-09-27T16:44:37.266467",
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
+ 7.81366259765625,
172
+ 7.89424267578125,
173
+ 7.9050439453125,
174
+ 7.92541259765625,
175
+ 7.961873046875
176
+ ],
177
+ "mean": 7.90004697265625,
178
+ "median": 7.9050439453125,
179
+ "std": 0.04896283943364418,
180
+ "min": 7.81366259765625,
181
+ "max": 7.961873046875,
182
+ "p25": 7.89424267578125,
183
+ "p75": 7.92541259765625,
184
+ "p90": 7.9472888671875,
185
+ "p95": 7.95458095703125,
186
+ "p99": 7.96041462890625,
187
+ "unit": "seconds"
188
+ },
189
+ "time_to_first_token_seconds": {
190
+ "name": "time_to_first_token_seconds",
191
+ "measurements": [
192
+ 0.04132966232299805,
193
+ 0.038022529602050784,
194
+ 0.03903193664550781,
195
+ 0.03811366271972656,
196
+ 0.0386357421875
197
+ ],
198
+ "mean": 0.039026706695556634,
199
+ "median": 0.0386357421875,
200
+ "std": 0.0012081809597756434,
201
+ "min": 0.038022529602050784,
202
+ "max": 0.04132966232299805,
203
+ "p25": 0.03811366271972656,
204
+ "p75": 0.03903193664550781,
205
+ "p90": 0.04041057205200195,
206
+ "p95": 0.0408701171875,
207
+ "p99": 0.04123775329589844,
208
+ "unit": "seconds"
209
+ },
210
+ "tokens_per_second": {
211
+ "name": "tokens_per_second",
212
+ "measurements": [
213
+ 12.798095483415874,
214
+ 12.667459578711716,
215
+ 12.650151054416032,
216
+ 12.61763962037416,
217
+ 12.559858642715932
218
+ ],
219
+ "mean": 12.658640875926743,
220
+ "median": 12.650151054416032,
221
+ "std": 0.07878637787466103,
222
+ "min": 12.559858642715932,
223
+ "max": 12.798095483415874,
224
+ "p25": 12.61763962037416,
225
+ "p75": 12.667459578711716,
226
+ "p90": 12.74584112153421,
227
+ "p95": 12.771968302475043,
228
+ "p99": 12.792870047227709,
229
+ "unit": "tokens/sec"
230
+ },
231
+ "time_per_output_token_seconds": {
232
+ "name": "time_per_output_token_seconds",
233
+ "measurements": [
234
+ 0.0781366259765625,
235
+ 0.0789424267578125,
236
+ 0.07905043945312501,
237
+ 0.0792541259765625,
238
+ 0.07961873046875
239
+ ],
240
+ "mean": 0.07900046972656251,
241
+ "median": 0.07905043945312501,
242
+ "std": 0.0004896283943364413,
243
+ "min": 0.0781366259765625,
244
+ "max": 0.07961873046875,
245
+ "p25": 0.0789424267578125,
246
+ "p75": 0.0792541259765625,
247
+ "p90": 0.079472888671875,
248
+ "p95": 0.0795458095703125,
249
+ "p99": 0.0796041462890625,
250
+ "unit": "seconds/token"
251
+ }
252
+ },
253
+ "gpu_metrics": {
254
+ "gpu_utilization_mean": 43.90588235294118,
255
+ "gpu_utilization_max": 91,
256
+ "gpu_utilization_min": 0,
257
+ "gpu_memory_used_mean": 13629.388235294118,
258
+ "gpu_memory_used_max": 13777,
259
+ "gpu_memory_used_min": 13415,
260
+ "sample_count": 170,
261
+ "gpu_monitoring_status": "success"
262
+ }
263
+ },
264
+ {
265
+ "scenario_name": "eager_sdpa_default",
266
+ "metadata": {
267
+ "timestamp": "2025-09-27T16:46:05.502480",
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
+ 3.405733154296875,
302
+ 3.410869873046875,
303
+ 3.406770751953125,
304
+ 3.40637939453125,
305
+ 3.40550830078125
306
+ ],
307
+ "mean": 3.4070522949218747,
308
+ "median": 3.40637939453125,
309
+ "std": 0.001961122842365854,
310
+ "min": 3.40550830078125,
311
+ "max": 3.410869873046875,
312
+ "p25": 3.405733154296875,
313
+ "p75": 3.406770751953125,
314
+ "p90": 3.409230224609375,
315
+ "p95": 3.410050048828125,
316
+ "p99": 3.4107059082031252,
317
+ "unit": "seconds"
318
+ },
319
+ "time_to_first_token_seconds": {
320
+ "name": "time_to_first_token_seconds",
321
+ "measurements": [
322
+ 0.035065887451171875,
323
+ 0.03454547119140625,
324
+ 0.034573150634765626,
325
+ 0.03487811279296875,
326
+ 0.034569343566894534
327
+ ],
328
+ "mean": 0.034726393127441406,
329
+ "median": 0.034573150634765626,
330
+ "std": 0.0002093589385343254,
331
+ "min": 0.03454547119140625,
332
+ "max": 0.035065887451171875,
333
+ "p25": 0.034569343566894534,
334
+ "p75": 0.03487811279296875,
335
+ "p90": 0.034990777587890626,
336
+ "p95": 0.03502833251953125,
337
+ "p99": 0.03505837646484375,
338
+ "unit": "seconds"
339
+ },
340
+ "tokens_per_second": {
341
+ "name": "tokens_per_second",
342
+ "measurements": [
343
+ 29.36225343251983,
344
+ 29.31803432028077,
345
+ 29.35331059263947,
346
+ 29.356682981509444,
347
+ 29.364192116947486
348
+ ],
349
+ "mean": 29.3508946887794,
350
+ "median": 29.356682981509444,
351
+ "std": 0.016881868364026952,
352
+ "min": 29.31803432028077,
353
+ "max": 29.364192116947486,
354
+ "p25": 29.35331059263947,
355
+ "p75": 29.36225343251983,
356
+ "p90": 29.36341664317642,
357
+ "p95": 29.363804380061953,
358
+ "p99": 29.36411456957038,
359
+ "unit": "tokens/sec"
360
+ },
361
+ "time_per_output_token_seconds": {
362
+ "name": "time_per_output_token_seconds",
363
+ "measurements": [
364
+ 0.03405733154296875,
365
+ 0.034108698730468755,
366
+ 0.03406770751953125,
367
+ 0.034063793945312504,
368
+ 0.0340550830078125
369
+ ],
370
+ "mean": 0.03407052294921875,
371
+ "median": 0.034063793945312504,
372
+ "std": 1.9611228423659388e-05,
373
+ "min": 0.0340550830078125,
374
+ "max": 0.034108698730468755,
375
+ "p25": 0.03405733154296875,
376
+ "p75": 0.03406770751953125,
377
+ "p90": 0.034092302246093754,
378
+ "p95": 0.034100500488281255,
379
+ "p99": 0.03410705908203125,
380
+ "unit": "seconds/token"
381
+ }
382
+ },
383
+ "gpu_metrics": {
384
+ "gpu_utilization_mean": 96.96511627906976,
385
+ "gpu_utilization_max": 98,
386
+ "gpu_utilization_min": 96,
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": "2025-09-27T16:46:37.888415",
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
+ 3.5613515625,
432
+ 3.55450927734375,
433
+ 3.546437255859375,
434
+ 3.558166748046875,
435
+ 3.52887255859375
436
+ ],
437
+ "mean": 3.5498674804687504,
438
+ "median": 3.55450927734375,
439
+ "std": 0.011617757495064765,
440
+ "min": 3.52887255859375,
441
+ "max": 3.5613515625,
442
+ "p25": 3.546437255859375,
443
+ "p75": 3.558166748046875,
444
+ "p90": 3.56007763671875,
445
+ "p95": 3.560714599609375,
446
+ "p99": 3.561224169921875,
447
+ "unit": "seconds"
448
+ },
449
+ "time_to_first_token_seconds": {
450
+ "name": "time_to_first_token_seconds",
451
+ "measurements": [
452
+ 0.040314910888671875,
453
+ 0.04041721725463867,
454
+ 0.039656673431396484,
455
+ 0.04005731201171875,
456
+ 0.03805168151855469
457
+ ],
458
+ "mean": 0.03969955902099609,
459
+ "median": 0.04005731201171875,
460
+ "std": 0.0008647563703825356,
461
+ "min": 0.03805168151855469,
462
+ "max": 0.04041721725463867,
463
+ "p25": 0.039656673431396484,
464
+ "p75": 0.040314910888671875,
465
+ "p90": 0.04037629470825195,
466
+ "p95": 0.04039675598144531,
467
+ "p99": 0.040413125,
468
+ "unit": "seconds"
469
+ },
470
+ "tokens_per_second": {
471
+ "name": "tokens_per_second",
472
+ "measurements": [
473
+ 28.079227294763882,
474
+ 28.133278660262498,
475
+ 28.197312622627503,
476
+ 28.10436021720773,
477
+ 28.337662621585245
478
+ ],
479
+ "mean": 28.17036828328937,
480
+ "median": 28.133278660262498,
481
+ "std": 0.09246921407878403,
482
+ "min": 28.079227294763882,
483
+ "max": 28.337662621585245,
484
+ "p25": 28.10436021720773,
485
+ "p75": 28.197312622627503,
486
+ "p90": 28.28152262200215,
487
+ "p95": 28.309592621793698,
488
+ "p99": 28.332048621626935,
489
+ "unit": "tokens/sec"
490
+ },
491
+ "time_per_output_token_seconds": {
492
+ "name": "time_per_output_token_seconds",
493
+ "measurements": [
494
+ 0.035613515625,
495
+ 0.0355450927734375,
496
+ 0.03546437255859375,
497
+ 0.03558166748046875,
498
+ 0.0352887255859375
499
+ ],
500
+ "mean": 0.035498674804687505,
501
+ "median": 0.0355450927734375,
502
+ "std": 0.00011617757495064888,
503
+ "min": 0.0352887255859375,
504
+ "max": 0.035613515625,
505
+ "p25": 0.03546437255859375,
506
+ "p75": 0.03558166748046875,
507
+ "p90": 0.0356007763671875,
508
+ "p95": 0.03560714599609375,
509
+ "p99": 0.03561224169921875,
510
+ "unit": "seconds/token"
511
+ }
512
+ },
513
+ "gpu_metrics": {
514
+ "gpu_utilization_mean": 95.5111111111111,
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": 90,
521
+ "gpu_monitoring_status": "success"
522
+ }
523
+ },
524
+ {
525
+ "scenario_name": "eager_sdpa_flash_attention",
526
+ "metadata": {
527
+ "timestamp": "2025-09-27T16:47:11.554129",
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
+ 3.40582470703125,
562
+ 3.408408447265625,
563
+ 3.410392822265625,
564
+ 3.410641845703125,
565
+ 3.405829833984375
566
+ ],
567
+ "mean": 3.40821953125,
568
+ "median": 3.408408447265625,
569
+ "std": 0.0021010628962227526,
570
+ "min": 3.40582470703125,
571
+ "max": 3.410641845703125,
572
+ "p25": 3.405829833984375,
573
+ "p75": 3.410392822265625,
574
+ "p90": 3.410542236328125,
575
+ "p95": 3.410592041015625,
576
+ "p99": 3.4106318847656247,
577
+ "unit": "seconds"
578
+ },
579
+ "time_to_first_token_seconds": {
580
+ "name": "time_to_first_token_seconds",
581
+ "measurements": [
582
+ 0.03524809646606445,
583
+ 0.03458505630493164,
584
+ 0.034621086120605465,
585
+ 0.034813568115234374,
586
+ 0.03461289596557617
587
+ ],
588
+ "mean": 0.03477614059448242,
589
+ "median": 0.034621086120605465,
590
+ "std": 0.0002495384012449854,
591
+ "min": 0.03458505630493164,
592
+ "max": 0.03524809646606445,
593
+ "p25": 0.03461289596557617,
594
+ "p75": 0.034813568115234374,
595
+ "p90": 0.03507428512573242,
596
+ "p95": 0.03516119079589843,
597
+ "p99": 0.035230715332031246,
598
+ "unit": "seconds"
599
+ },
600
+ "tokens_per_second": {
601
+ "name": "tokens_per_second",
602
+ "measurements": [
603
+ 29.36146413922954,
604
+ 29.33920671397948,
605
+ 29.322135370190885,
606
+ 29.31999445382527,
607
+ 29.361419940059978
608
+ ],
609
+ "mean": 29.340844123457032,
610
+ "median": 29.33920671397948,
611
+ "std": 0.018088430344620182,
612
+ "min": 29.31999445382527,
613
+ "max": 29.36146413922954,
614
+ "p25": 29.322135370190885,
615
+ "p75": 29.361419940059978,
616
+ "p90": 29.361446459561716,
617
+ "p95": 29.361455299395626,
618
+ "p99": 29.361462371262757,
619
+ "unit": "tokens/sec"
620
+ },
621
+ "time_per_output_token_seconds": {
622
+ "name": "time_per_output_token_seconds",
623
+ "measurements": [
624
+ 0.0340582470703125,
625
+ 0.03408408447265625,
626
+ 0.034103928222656246,
627
+ 0.034106418457031246,
628
+ 0.03405829833984375
629
+ ],
630
+ "mean": 0.034082195312499994,
631
+ "median": 0.03408408447265625,
632
+ "std": 2.1010628962226553e-05,
633
+ "min": 0.0340582470703125,
634
+ "max": 0.034106418457031246,
635
+ "p25": 0.03405829833984375,
636
+ "p75": 0.034103928222656246,
637
+ "p90": 0.03410542236328125,
638
+ "p95": 0.03410592041015625,
639
+ "p99": 0.03410631884765625,
640
+ "unit": "seconds/token"
641
+ }
642
+ },
643
+ "gpu_metrics": {
644
+ "gpu_utilization_mean": 96.87209302325581,
645
+ "gpu_utilization_max": 98,
646
+ "gpu_utilization_min": 95,
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": "2025-09-27T16:47:43.960590",
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
+ 3.406966796875,
692
+ 3.410205810546875,
693
+ 3.40637890625,
694
+ 3.403314697265625,
695
+ 3.409290283203125
696
+ ],
697
+ "mean": 3.407231298828125,
698
+ "median": 3.406966796875,
699
+ "std": 0.0024175789050742257,
700
+ "min": 3.403314697265625,
701
+ "max": 3.410205810546875,
702
+ "p25": 3.40637890625,
703
+ "p75": 3.409290283203125,
704
+ "p90": 3.409839599609375,
705
+ "p95": 3.410022705078125,
706
+ "p99": 3.410169189453125,
707
+ "unit": "seconds"
708
+ },
709
+ "time_to_first_token_seconds": {
710
+ "name": "time_to_first_token_seconds",
711
+ "measurements": [
712
+ 0.03536457443237305,
713
+ 0.03473638534545898,
714
+ 0.03483903884887695,
715
+ 0.03471513748168945,
716
+ 0.03504009628295898
717
+ ],
718
+ "mean": 0.034939046478271486,
719
+ "median": 0.03483903884887695,
720
+ "std": 0.00024186500457469201,
721
+ "min": 0.03471513748168945,
722
+ "max": 0.03536457443237305,
723
+ "p25": 0.03473638534545898,
724
+ "p75": 0.03504009628295898,
725
+ "p90": 0.03523478317260742,
726
+ "p95": 0.035299678802490235,
727
+ "p99": 0.035351595306396484,
728
+ "unit": "seconds"
729
+ },
730
+ "tokens_per_second": {
731
+ "name": "tokens_per_second",
732
+ "measurements": [
733
+ 29.351621533771276,
734
+ 29.3237433619772,
735
+ 29.356687189590303,
736
+ 29.383118781329408,
737
+ 29.33161793018316
738
+ ],
739
+ "mean": 29.34935775937027,
740
+ "median": 29.351621533771276,
741
+ "std": 0.02083002985601457,
742
+ "min": 29.3237433619772,
743
+ "max": 29.383118781329408,
744
+ "p25": 29.33161793018316,
745
+ "p75": 29.356687189590303,
746
+ "p90": 29.372546144633766,
747
+ "p95": 29.377832462981587,
748
+ "p99": 29.382061517659842,
749
+ "unit": "tokens/sec"
750
+ },
751
+ "time_per_output_token_seconds": {
752
+ "name": "time_per_output_token_seconds",
753
+ "measurements": [
754
+ 0.03406966796875,
755
+ 0.03410205810546875,
756
+ 0.0340637890625,
757
+ 0.034033146972656254,
758
+ 0.03409290283203125
759
+ ],
760
+ "mean": 0.034072312988281256,
761
+ "median": 0.03406966796875,
762
+ "std": 2.4175789050741746e-05,
763
+ "min": 0.034033146972656254,
764
+ "max": 0.03410205810546875,
765
+ "p25": 0.0340637890625,
766
+ "p75": 0.03409290283203125,
767
+ "p90": 0.03409839599609375,
768
+ "p95": 0.03410022705078125,
769
+ "p99": 0.03410169189453125,
770
+ "unit": "seconds/token"
771
+ }
772
+ },
773
+ "gpu_metrics": {
774
+ "gpu_utilization_mean": 96.96511627906976,
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": "2025-09-27T16:48:16.346014",
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
+ 3.6924375,
822
+ 3.688404052734375,
823
+ 3.694310791015625,
824
+ 3.691896240234375,
825
+ 3.6925673828125
826
+ ],
827
+ "mean": 3.691923193359375,
828
+ "median": 3.6924375,
829
+ "std": 0.0019372759818731698,
830
+ "min": 3.688404052734375,
831
+ "max": 3.694310791015625,
832
+ "p25": 3.691896240234375,
833
+ "p75": 3.6925673828125,
834
+ "p90": 3.693613427734375,
835
+ "p95": 3.693962109375,
836
+ "p99": 3.6942410546875,
837
+ "unit": "seconds"
838
+ },
839
+ "time_to_first_token_seconds": {
840
+ "name": "time_to_first_token_seconds",
841
+ "measurements": [
842
+ 0.03793664169311523,
843
+ 0.03739020919799805,
844
+ 0.03759347152709961,
845
+ 0.03745062255859375,
846
+ 0.03738236618041992
847
+ ],
848
+ "mean": 0.03755066223144531,
849
+ "median": 0.03745062255859375,
850
+ "std": 0.0002073110964241108,
851
+ "min": 0.03738236618041992,
852
+ "max": 0.03793664169311523,
853
+ "p25": 0.03739020919799805,
854
+ "p75": 0.03759347152709961,
855
+ "p90": 0.03779937362670898,
856
+ "p95": 0.037868007659912106,
857
+ "p99": 0.037922914886474604,
858
+ "unit": "seconds"
859
+ },
860
+ "tokens_per_second": {
861
+ "name": "tokens_per_second",
862
+ "measurements": [
863
+ 27.082381218368624,
864
+ 27.111997105052968,
865
+ 27.068648431852264,
866
+ 27.086351699215587,
867
+ 27.081428619410456
868
+ ],
869
+ "mean": 27.08616141477998,
870
+ "median": 27.082381218368624,
871
+ "std": 0.01421909215099813,
872
+ "min": 27.068648431852264,
873
+ "max": 27.111997105052968,
874
+ "p25": 27.081428619410456,
875
+ "p75": 27.086351699215587,
876
+ "p90": 27.101738942718015,
877
+ "p95": 27.106868023885493,
878
+ "p99": 27.110971288819474,
879
+ "unit": "tokens/sec"
880
+ },
881
+ "time_per_output_token_seconds": {
882
+ "name": "time_per_output_token_seconds",
883
+ "measurements": [
884
+ 0.036924375,
885
+ 0.03688404052734375,
886
+ 0.03694310791015625,
887
+ 0.03691896240234375,
888
+ 0.036925673828125
889
+ ],
890
+ "mean": 0.03691923193359375,
891
+ "median": 0.036924375,
892
+ "std": 1.937275981873061e-05,
893
+ "min": 0.03688404052734375,
894
+ "max": 0.03694310791015625,
895
+ "p25": 0.03691896240234375,
896
+ "p75": 0.036925673828125,
897
+ "p90": 0.03693613427734375,
898
+ "p95": 0.03693962109375,
899
+ "p99": 0.036942410546875,
900
+ "unit": "seconds/token"
901
+ }
902
+ },
903
+ "gpu_metrics": {
904
+ "gpu_utilization_mean": 90.44086021505376,
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": "2025-09-27T16:48:51.052581",
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
+ 3.857059814453125,
952
+ 3.84916455078125,
953
+ 3.873222900390625,
954
+ 3.91011083984375,
955
+ 3.90198974609375
956
+ ],
957
+ "mean": 3.8783095703125,
958
+ "median": 3.873222900390625,
959
+ "std": 0.02407874567864449,
960
+ "min": 3.84916455078125,
961
+ "max": 3.91011083984375,
962
+ "p25": 3.857059814453125,
963
+ "p75": 3.90198974609375,
964
+ "p90": 3.90686240234375,
965
+ "p95": 3.90848662109375,
966
+ "p99": 3.90978599609375,
967
+ "unit": "seconds"
968
+ },
969
+ "time_to_first_token_seconds": {
970
+ "name": "time_to_first_token_seconds",
971
+ "measurements": [
972
+ 0.04038566589355469,
973
+ 0.03985910415649414,
974
+ 0.04108998489379883,
975
+ 0.04018716812133789,
976
+ 0.03982169723510742
977
+ ],
978
+ "mean": 0.04026872406005859,
979
+ "median": 0.04018716812133789,
980
+ "std": 0.0004609593405539853,
981
+ "min": 0.03982169723510742,
982
+ "max": 0.04108998489379883,
983
+ "p25": 0.03985910415649414,
984
+ "p75": 0.04038566589355469,
985
+ "p90": 0.040808257293701175,
986
+ "p95": 0.04094912109375,
987
+ "p99": 0.041061812133789066,
988
+ "unit": "seconds"
989
+ },
990
+ "tokens_per_second": {
991
+ "name": "tokens_per_second",
992
+ "measurements": [
993
+ 25.926484112401184,
994
+ 25.979663555745724,
995
+ 25.818292045602316,
996
+ 25.574722583566466,
997
+ 25.627950483496065
998
+ ],
999
+ "mean": 25.78542255616235,
1000
+ "median": 25.818292045602316,
1001
+ "std": 0.1599376528685075,
1002
+ "min": 25.574722583566466,
1003
+ "max": 25.979663555745724,
1004
+ "p25": 25.627950483496065,
1005
+ "p75": 25.926484112401184,
1006
+ "p90": 25.958391778407908,
1007
+ "p95": 25.969027667076816,
1008
+ "p99": 25.97753637801194,
1009
+ "unit": "tokens/sec"
1010
+ },
1011
+ "time_per_output_token_seconds": {
1012
+ "name": "time_per_output_token_seconds",
1013
+ "measurements": [
1014
+ 0.03857059814453125,
1015
+ 0.0384916455078125,
1016
+ 0.03873222900390625,
1017
+ 0.0391011083984375,
1018
+ 0.0390198974609375
1019
+ ],
1020
+ "mean": 0.038783095703125006,
1021
+ "median": 0.03873222900390625,
1022
+ "std": 0.00024078745678644524,
1023
+ "min": 0.0384916455078125,
1024
+ "max": 0.0391011083984375,
1025
+ "p25": 0.03857059814453125,
1026
+ "p75": 0.0390198974609375,
1027
+ "p90": 0.0390686240234375,
1028
+ "p95": 0.0390848662109375,
1029
+ "p99": 0.0390978599609375,
1030
+ "unit": "seconds/token"
1031
+ }
1032
+ },
1033
+ "gpu_metrics": {
1034
+ "gpu_utilization_mean": 89.46938775510205,
1035
+ "gpu_utilization_max": 93,
1036
+ "gpu_utilization_min": 84,
1037
+ "gpu_memory_used_mean": 13845,
1038
+ "gpu_memory_used_max": 13845,
1039
+ "gpu_memory_used_min": 13845,
1040
+ "sample_count": 98,
1041
+ "gpu_monitoring_status": "success"
1042
+ }
1043
+ },
1044
+ {
1045
+ "scenario_name": "compiled_compile_max-autotune_sdpa_efficient_attention",
1046
+ "metadata": {
1047
+ "timestamp": "2025-09-27T16:49:31.887036",
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
+ 3.68543310546875,
1082
+ 3.70019140625,
1083
+ 3.7083955078125,
1084
+ 3.70324609375,
1085
+ 3.706689697265625
1086
+ ],
1087
+ "mean": 3.7007911621093745,
1088
+ "median": 3.70324609375,
1089
+ "std": 0.008183826620717195,
1090
+ "min": 3.68543310546875,
1091
+ "max": 3.7083955078125,
1092
+ "p25": 3.70019140625,
1093
+ "p75": 3.706689697265625,
1094
+ "p90": 3.70771318359375,
1095
+ "p95": 3.708054345703125,
1096
+ "p99": 3.708327275390625,
1097
+ "unit": "seconds"
1098
+ },
1099
+ "time_to_first_token_seconds": {
1100
+ "name": "time_to_first_token_seconds",
1101
+ "measurements": [
1102
+ 0.03815087890625,
1103
+ 0.037472286224365235,
1104
+ 0.03767267227172852,
1105
+ 0.03807449722290039,
1106
+ 0.03757689666748047
1107
+ ],
1108
+ "mean": 0.037789446258544926,
1109
+ "median": 0.03767267227172852,
1110
+ "std": 0.0002725037948530651,
1111
+ "min": 0.037472286224365235,
1112
+ "max": 0.03815087890625,
1113
+ "p25": 0.03757689666748047,
1114
+ "p75": 0.03807449722290039,
1115
+ "p90": 0.03812032623291015,
1116
+ "p95": 0.038135602569580075,
1117
+ "p99": 0.038147823638916015,
1118
+ "unit": "seconds"
1119
+ },
1120
+ "tokens_per_second": {
1121
+ "name": "tokens_per_second",
1122
+ "measurements": [
1123
+ 27.133852966049428,
1124
+ 27.025628952894117,
1125
+ 26.965840021467336,
1126
+ 27.003336388788977,
1127
+ 26.978249642469034
1128
+ ],
1129
+ "mean": 27.021381594333782,
1130
+ "median": 27.003336388788977,
1131
+ "std": 0.05989776380072121,
1132
+ "min": 26.965840021467336,
1133
+ "max": 27.133852966049428,
1134
+ "p25": 26.978249642469034,
1135
+ "p75": 27.025628952894117,
1136
+ "p90": 27.090563360787304,
1137
+ "p95": 27.112208163418366,
1138
+ "p99": 27.129524005523216,
1139
+ "unit": "tokens/sec"
1140
+ },
1141
+ "time_per_output_token_seconds": {
1142
+ "name": "time_per_output_token_seconds",
1143
+ "measurements": [
1144
+ 0.0368543310546875,
1145
+ 0.0370019140625,
1146
+ 0.037083955078125,
1147
+ 0.0370324609375,
1148
+ 0.03706689697265625
1149
+ ],
1150
+ "mean": 0.03700791162109375,
1151
+ "median": 0.0370324609375,
1152
+ "std": 8.1838266207171e-05,
1153
+ "min": 0.0368543310546875,
1154
+ "max": 0.037083955078125,
1155
+ "p25": 0.0370019140625,
1156
+ "p75": 0.03706689697265625,
1157
+ "p90": 0.0370771318359375,
1158
+ "p95": 0.037080543457031254,
1159
+ "p99": 0.03708327275390625,
1160
+ "unit": "seconds/token"
1161
+ }
1162
+ },
1163
+ "gpu_metrics": {
1164
+ "gpu_utilization_mean": 90.27956989247312,
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
+ }