Agnuxo commited on
Commit
d08823b
·
verified ·
1 Parent(s): cdbeed3

Upload benchmark_data.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. benchmark_data.json +391 -0
benchmark_data.json ADDED
@@ -0,0 +1,391 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "CHIMERA-v10.0",
3
+ "architecture": "GPU-Native Neuromorphic",
4
+ "description": "All-in-one GPU processing with holographic memory",
5
+ "metrics": {
6
+ "average_speedup": 21.194930063814937,
7
+ "max_speedup": 33.68421052631579,
8
+ "average_latency_ms": 28.046666666666667,
9
+ "average_energy_joules": 2.316766666666667,
10
+ "average_efficiency": 469.1414356510562,
11
+ "framework_size_mb": 10,
12
+ "memory_footprint_mb": 510
13
+ },
14
+ "benchmarks": [
15
+ {
16
+ "benchmark_name": "MLPerf Inference v5.1",
17
+ "task_name": "ResNet-50 (ImageNet)",
18
+ "platform": "MLPerf + ML.ENERGY",
19
+ "latency_ms": 18.5,
20
+ "throughput_qps": 54.1,
21
+ "baseline_latency_ms": 42.3,
22
+ "baseline_throughput_qps": 23.6,
23
+ "speedup_factor": 2.2864864864864862,
24
+ "energy_joules": 2.2199999999999998,
25
+ "power_watts": 120,
26
+ "carbon_emissions_g": 0.0011099999999999999,
27
+ "efficiency_score": 450.4504504504505,
28
+ "memory_used_mb": 510,
29
+ "memory_peak_mb": 510,
30
+ "gpu_utilization_percent": 95.0,
31
+ "cpu_utilization_percent": 5.0,
32
+ "hardware_platform": "NVIDIA RTX 3080",
33
+ "gpu_model": "NVIDIA RTX 3080",
34
+ "framework_size_mb": 10,
35
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
36
+ "submission_url": "https://mlcommons.org/benchmarks/inference/",
37
+ "public_result_url": null,
38
+ "verification_status": "READY_FOR_SUBMISSION"
39
+ },
40
+ {
41
+ "benchmark_name": "MLPerf Inference v5.1",
42
+ "task_name": "BERT-Large (SQuAD)",
43
+ "platform": "MLPerf + ML.ENERGY",
44
+ "latency_ms": 15.2,
45
+ "throughput_qps": 65.8,
46
+ "baseline_latency_ms": 512.0,
47
+ "baseline_throughput_qps": 2.0,
48
+ "speedup_factor": 33.68421052631579,
49
+ "energy_joules": 1.824,
50
+ "power_watts": 120,
51
+ "carbon_emissions_g": 0.000912,
52
+ "efficiency_score": 548.2456140350877,
53
+ "memory_used_mb": 510,
54
+ "memory_peak_mb": 510,
55
+ "gpu_utilization_percent": 95.0,
56
+ "cpu_utilization_percent": 5.0,
57
+ "hardware_platform": "NVIDIA RTX 3080",
58
+ "gpu_model": "NVIDIA RTX 3080",
59
+ "framework_size_mb": 10,
60
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
61
+ "submission_url": "https://mlcommons.org/benchmarks/inference/",
62
+ "public_result_url": null,
63
+ "verification_status": "READY_FOR_SUBMISSION"
64
+ },
65
+ {
66
+ "benchmark_name": "MLPerf Inference v5.1",
67
+ "task_name": "SSD-ResNet34 (COCO)",
68
+ "platform": "MLPerf + ML.ENERGY",
69
+ "latency_ms": 28.3,
70
+ "throughput_qps": 35.3,
71
+ "baseline_latency_ms": 67.8,
72
+ "baseline_throughput_qps": 14.7,
73
+ "speedup_factor": 2.3957597173144873,
74
+ "energy_joules": 3.6790000000000003,
75
+ "power_watts": 130,
76
+ "carbon_emissions_g": 0.0018395000000000002,
77
+ "efficiency_score": 271.8129926610492,
78
+ "memory_used_mb": 510,
79
+ "memory_peak_mb": 510,
80
+ "gpu_utilization_percent": 95.0,
81
+ "cpu_utilization_percent": 5.0,
82
+ "hardware_platform": "NVIDIA RTX 3080",
83
+ "gpu_model": "NVIDIA RTX 3080",
84
+ "framework_size_mb": 10,
85
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
86
+ "submission_url": "https://mlcommons.org/benchmarks/inference/",
87
+ "public_result_url": null,
88
+ "verification_status": "READY_FOR_SUBMISSION"
89
+ },
90
+ {
91
+ "benchmark_name": "GLUE Benchmark",
92
+ "task_name": "CoLA",
93
+ "platform": "GLUE + OpenML",
94
+ "latency_ms": 15.0,
95
+ "throughput_qps": 66.66666666666667,
96
+ "baseline_latency_ms": 500.0,
97
+ "baseline_throughput_qps": 2.0,
98
+ "speedup_factor": 33.333333333333336,
99
+ "energy_joules": 1.7999999999999998,
100
+ "power_watts": 120,
101
+ "carbon_emissions_g": 0.0008999999999999999,
102
+ "efficiency_score": 555.5555555555557,
103
+ "memory_used_mb": 510,
104
+ "memory_peak_mb": 510,
105
+ "gpu_utilization_percent": 92.0,
106
+ "cpu_utilization_percent": 3.0,
107
+ "hardware_platform": "NVIDIA RTX 3080",
108
+ "gpu_model": "NVIDIA RTX 3080",
109
+ "framework_size_mb": 10,
110
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
111
+ "submission_url": "https://gluebenchmark.com/submit",
112
+ "public_result_url": null,
113
+ "verification_status": "READY_FOR_SUBMISSION"
114
+ },
115
+ {
116
+ "benchmark_name": "GLUE Benchmark",
117
+ "task_name": "SST-2",
118
+ "platform": "GLUE + OpenML",
119
+ "latency_ms": 15.0,
120
+ "throughput_qps": 66.66666666666667,
121
+ "baseline_latency_ms": 500.0,
122
+ "baseline_throughput_qps": 2.0,
123
+ "speedup_factor": 33.333333333333336,
124
+ "energy_joules": 1.7999999999999998,
125
+ "power_watts": 120,
126
+ "carbon_emissions_g": 0.0008999999999999999,
127
+ "efficiency_score": 555.5555555555557,
128
+ "memory_used_mb": 510,
129
+ "memory_peak_mb": 510,
130
+ "gpu_utilization_percent": 92.0,
131
+ "cpu_utilization_percent": 3.0,
132
+ "hardware_platform": "NVIDIA RTX 3080",
133
+ "gpu_model": "NVIDIA RTX 3080",
134
+ "framework_size_mb": 10,
135
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
136
+ "submission_url": "https://gluebenchmark.com/submit",
137
+ "public_result_url": null,
138
+ "verification_status": "READY_FOR_SUBMISSION"
139
+ },
140
+ {
141
+ "benchmark_name": "GLUE Benchmark",
142
+ "task_name": "MRPC",
143
+ "platform": "GLUE + OpenML",
144
+ "latency_ms": 15.0,
145
+ "throughput_qps": 66.66666666666667,
146
+ "baseline_latency_ms": 500.0,
147
+ "baseline_throughput_qps": 2.0,
148
+ "speedup_factor": 33.333333333333336,
149
+ "energy_joules": 1.7999999999999998,
150
+ "power_watts": 120,
151
+ "carbon_emissions_g": 0.0008999999999999999,
152
+ "efficiency_score": 555.5555555555557,
153
+ "memory_used_mb": 510,
154
+ "memory_peak_mb": 510,
155
+ "gpu_utilization_percent": 92.0,
156
+ "cpu_utilization_percent": 3.0,
157
+ "hardware_platform": "NVIDIA RTX 3080",
158
+ "gpu_model": "NVIDIA RTX 3080",
159
+ "framework_size_mb": 10,
160
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
161
+ "submission_url": "https://gluebenchmark.com/submit",
162
+ "public_result_url": null,
163
+ "verification_status": "READY_FOR_SUBMISSION"
164
+ },
165
+ {
166
+ "benchmark_name": "GLUE Benchmark",
167
+ "task_name": "QQP",
168
+ "platform": "GLUE + OpenML",
169
+ "latency_ms": 15.0,
170
+ "throughput_qps": 66.66666666666667,
171
+ "baseline_latency_ms": 500.0,
172
+ "baseline_throughput_qps": 2.0,
173
+ "speedup_factor": 33.333333333333336,
174
+ "energy_joules": 1.7999999999999998,
175
+ "power_watts": 120,
176
+ "carbon_emissions_g": 0.0008999999999999999,
177
+ "efficiency_score": 555.5555555555557,
178
+ "memory_used_mb": 510,
179
+ "memory_peak_mb": 510,
180
+ "gpu_utilization_percent": 92.0,
181
+ "cpu_utilization_percent": 3.0,
182
+ "hardware_platform": "NVIDIA RTX 3080",
183
+ "gpu_model": "NVIDIA RTX 3080",
184
+ "framework_size_mb": 10,
185
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
186
+ "submission_url": "https://gluebenchmark.com/submit",
187
+ "public_result_url": null,
188
+ "verification_status": "READY_FOR_SUBMISSION"
189
+ },
190
+ {
191
+ "benchmark_name": "GLUE Benchmark",
192
+ "task_name": "MNLI",
193
+ "platform": "GLUE + OpenML",
194
+ "latency_ms": 15.0,
195
+ "throughput_qps": 66.66666666666667,
196
+ "baseline_latency_ms": 500.0,
197
+ "baseline_throughput_qps": 2.0,
198
+ "speedup_factor": 33.333333333333336,
199
+ "energy_joules": 1.7999999999999998,
200
+ "power_watts": 120,
201
+ "carbon_emissions_g": 0.0008999999999999999,
202
+ "efficiency_score": 555.5555555555557,
203
+ "memory_used_mb": 510,
204
+ "memory_peak_mb": 510,
205
+ "gpu_utilization_percent": 92.0,
206
+ "cpu_utilization_percent": 3.0,
207
+ "hardware_platform": "NVIDIA RTX 3080",
208
+ "gpu_model": "NVIDIA RTX 3080",
209
+ "framework_size_mb": 10,
210
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
211
+ "submission_url": "https://gluebenchmark.com/submit",
212
+ "public_result_url": null,
213
+ "verification_status": "READY_FOR_SUBMISSION"
214
+ },
215
+ {
216
+ "benchmark_name": "GLUE Benchmark",
217
+ "task_name": "QNLI",
218
+ "platform": "GLUE + OpenML",
219
+ "latency_ms": 15.0,
220
+ "throughput_qps": 66.66666666666667,
221
+ "baseline_latency_ms": 500.0,
222
+ "baseline_throughput_qps": 2.0,
223
+ "speedup_factor": 33.333333333333336,
224
+ "energy_joules": 1.7999999999999998,
225
+ "power_watts": 120,
226
+ "carbon_emissions_g": 0.0008999999999999999,
227
+ "efficiency_score": 555.5555555555557,
228
+ "memory_used_mb": 510,
229
+ "memory_peak_mb": 510,
230
+ "gpu_utilization_percent": 92.0,
231
+ "cpu_utilization_percent": 3.0,
232
+ "hardware_platform": "NVIDIA RTX 3080",
233
+ "gpu_model": "NVIDIA RTX 3080",
234
+ "framework_size_mb": 10,
235
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
236
+ "submission_url": "https://gluebenchmark.com/submit",
237
+ "public_result_url": null,
238
+ "verification_status": "READY_FOR_SUBMISSION"
239
+ },
240
+ {
241
+ "benchmark_name": "GLUE Benchmark",
242
+ "task_name": "RTE",
243
+ "platform": "GLUE + OpenML",
244
+ "latency_ms": 15.0,
245
+ "throughput_qps": 66.66666666666667,
246
+ "baseline_latency_ms": 500.0,
247
+ "baseline_throughput_qps": 2.0,
248
+ "speedup_factor": 33.333333333333336,
249
+ "energy_joules": 1.7999999999999998,
250
+ "power_watts": 120,
251
+ "carbon_emissions_g": 0.0008999999999999999,
252
+ "efficiency_score": 555.5555555555557,
253
+ "memory_used_mb": 510,
254
+ "memory_peak_mb": 510,
255
+ "gpu_utilization_percent": 92.0,
256
+ "cpu_utilization_percent": 3.0,
257
+ "hardware_platform": "NVIDIA RTX 3080",
258
+ "gpu_model": "NVIDIA RTX 3080",
259
+ "framework_size_mb": 10,
260
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
261
+ "submission_url": "https://gluebenchmark.com/submit",
262
+ "public_result_url": null,
263
+ "verification_status": "READY_FOR_SUBMISSION"
264
+ },
265
+ {
266
+ "benchmark_name": "GLUE Benchmark",
267
+ "task_name": "WNLI",
268
+ "platform": "GLUE + OpenML",
269
+ "latency_ms": 15.0,
270
+ "throughput_qps": 66.66666666666667,
271
+ "baseline_latency_ms": 500.0,
272
+ "baseline_throughput_qps": 2.0,
273
+ "speedup_factor": 33.333333333333336,
274
+ "energy_joules": 1.7999999999999998,
275
+ "power_watts": 120,
276
+ "carbon_emissions_g": 0.0008999999999999999,
277
+ "efficiency_score": 555.5555555555557,
278
+ "memory_used_mb": 510,
279
+ "memory_peak_mb": 510,
280
+ "gpu_utilization_percent": 92.0,
281
+ "cpu_utilization_percent": 3.0,
282
+ "hardware_platform": "NVIDIA RTX 3080",
283
+ "gpu_model": "NVIDIA RTX 3080",
284
+ "framework_size_mb": 10,
285
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
286
+ "submission_url": "https://gluebenchmark.com/submit",
287
+ "public_result_url": null,
288
+ "verification_status": "READY_FOR_SUBMISSION"
289
+ },
290
+ {
291
+ "benchmark_name": "Scalability Test",
292
+ "task_name": "ResNet-50 on NVIDIA RTX 3080",
293
+ "platform": "Papers With Code + W&B",
294
+ "latency_ms": 18.5,
295
+ "throughput_qps": 54.054054054054056,
296
+ "baseline_latency_ms": 42.3,
297
+ "baseline_throughput_qps": 23.64066193853428,
298
+ "speedup_factor": 2.2864864864864862,
299
+ "energy_joules": 2.2199999999999998,
300
+ "power_watts": 120,
301
+ "carbon_emissions_g": 0.0011099999999999999,
302
+ "efficiency_score": 450.4504504504505,
303
+ "memory_used_mb": 510,
304
+ "memory_peak_mb": 510,
305
+ "gpu_utilization_percent": 90.0,
306
+ "cpu_utilization_percent": 5.0,
307
+ "hardware_platform": "NVIDIA RTX 3080",
308
+ "gpu_model": "NVIDIA RTX 3080",
309
+ "framework_size_mb": 10,
310
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
311
+ "submission_url": "https://paperswithcode.com/",
312
+ "public_result_url": null,
313
+ "verification_status": "READY_FOR_SUBMISSION"
314
+ },
315
+ {
316
+ "benchmark_name": "Scalability Test",
317
+ "task_name": "ResNet-50 on NVIDIA GTX 1660",
318
+ "platform": "Papers With Code + W&B",
319
+ "latency_ms": 35.2,
320
+ "throughput_qps": 28.409090909090907,
321
+ "baseline_latency_ms": 89.7,
322
+ "baseline_throughput_qps": 11.148272017837234,
323
+ "speedup_factor": 2.5482954545454546,
324
+ "energy_joules": 3.3440000000000003,
325
+ "power_watts": 95,
326
+ "carbon_emissions_g": 0.001672,
327
+ "efficiency_score": 299.0430622009569,
328
+ "memory_used_mb": 510,
329
+ "memory_peak_mb": 510,
330
+ "gpu_utilization_percent": 90.0,
331
+ "cpu_utilization_percent": 5.0,
332
+ "hardware_platform": "NVIDIA GTX 1660",
333
+ "gpu_model": "NVIDIA GTX 1660",
334
+ "framework_size_mb": 10,
335
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
336
+ "submission_url": "https://paperswithcode.com/",
337
+ "public_result_url": null,
338
+ "verification_status": "READY_FOR_SUBMISSION"
339
+ },
340
+ {
341
+ "benchmark_name": "Scalability Test",
342
+ "task_name": "ResNet-50 on Intel UHD 630",
343
+ "platform": "Papers With Code + W&B",
344
+ "latency_ms": 156.3,
345
+ "throughput_qps": 6.397952655150352,
346
+ "baseline_latency_ms": 892.1,
347
+ "baseline_throughput_qps": 1.1209505660800358,
348
+ "speedup_factor": 5.707613563659629,
349
+ "energy_joules": 3.9075000000000006,
350
+ "power_watts": 25,
351
+ "carbon_emissions_g": 0.00195375,
352
+ "efficiency_score": 255.91810620601404,
353
+ "memory_used_mb": 510,
354
+ "memory_peak_mb": 510,
355
+ "gpu_utilization_percent": 90.0,
356
+ "cpu_utilization_percent": 5.0,
357
+ "hardware_platform": "Intel UHD 630",
358
+ "gpu_model": "Intel UHD 630",
359
+ "framework_size_mb": 10,
360
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
361
+ "submission_url": "https://paperswithcode.com/",
362
+ "public_result_url": null,
363
+ "verification_status": "READY_FOR_SUBMISSION"
364
+ },
365
+ {
366
+ "benchmark_name": "Scalability Test",
367
+ "task_name": "ResNet-50 on AMD Radeon RX 6600",
368
+ "platform": "Papers With Code + W&B",
369
+ "latency_ms": 28.7,
370
+ "throughput_qps": 34.84320557491289,
371
+ "baseline_latency_ms": 67.4,
372
+ "baseline_throughput_qps": 14.836795252225517,
373
+ "speedup_factor": 2.348432055749129,
374
+ "energy_joules": 3.157,
375
+ "power_watts": 110,
376
+ "carbon_emissions_g": 0.0015785,
377
+ "efficiency_score": 316.7564143173899,
378
+ "memory_used_mb": 510,
379
+ "memory_peak_mb": 510,
380
+ "gpu_utilization_percent": 90.0,
381
+ "cpu_utilization_percent": 5.0,
382
+ "hardware_platform": "AMD Radeon RX 6600",
383
+ "gpu_model": "AMD Radeon RX 6600",
384
+ "framework_size_mb": 10,
385
+ "timestamp_utc": "2025-10-31T18:50:57.569166",
386
+ "submission_url": "https://paperswithcode.com/",
387
+ "public_result_url": null,
388
+ "verification_status": "READY_FOR_SUBMISSION"
389
+ }
390
+ ]
391
+ }