Staging

[Delaunay]

No description provided.

[Delaunay]

=================
Benchmark results
=================

System
------
cpu:      AMD EPYC 9654 96-Core Processor
n_cpu:    192
product:  NVIDIA H100 80GB HBM3
n_gpu:    8
memory:   81559.0

Breakdown
---------
bench                    | fail |   n | ngpu |           perf |   sem% |   std% | peak_memory |           score | weight
brax                     |    0 |   1 |    8 |      884839.68 |   0.1% |   0.5% |        5336 |       884839.68 |   1.00
diffusion-gpus           |    0 |   1 |    8 |         266.60 |   2.7% |  20.9% |       60978 |          266.60 |   1.00
diffusion-single         |    0 |   8 |    1 |          45.22 |   1.1% |  25.1% |       55354 |          365.01 |   1.00
dimenet                  |    2 |   8 |    1 |         496.00 |   0.8% |  16.2% |         nan |         2255.08 |   1.00
dinov2-giant-gpus        |    0 |   1 |    8 |         866.13 |   0.3% |   2.1% |       72172 |          866.13 |   1.00
dinov2-giant-single      |    0 |   8 |    1 |         102.75 |   0.4% |   9.8% |       74544 |          830.58 |   1.00
dqn                      |    0 |   8 |    1 | 27570955013.07 |   1.2% |  91.2% |       61900 | 220247494366.61 |   1.00
bf16                     |    0 |   8 |    1 |         782.72 |   0.2% |   5.7% |         nan |         6294.84 |   0.00
fp16                     |    0 |   8 |    1 |         799.28 |   0.2% |   6.0% |        1548 |         6431.83 |   0.00
fp32                     |    0 |   8 |    1 |          51.93 |   0.0% |   0.7% |        1932 |          415.66 |   0.00
tf32                     |    0 |   8 |    1 |         406.70 |   0.2% |   5.7% |        1932 |         3258.92 |   0.00
bert-fp16                |    0 |   8 |    1 |         456.59 |   1.0% |  15.3% |         nan |         3742.80 |   0.00
bert-fp32                |    0 |   8 |    1 |         111.15 |   0.5% |   8.1% |       21006 |          900.99 |   0.00
bert-tf32                |    0 |   8 |    1 |         243.96 |   0.6% |  10.1% |         nan |         1985.16 |   0.00
bert-tf32-fp16           |    0 |   8 |    1 |         453.70 |   1.0% |  16.1% |         nan |         3716.66 |   3.00
reformer                 |    0 |   8 |    1 |         103.42 |   0.2% |   4.9% |       25208 |          832.18 |   1.00
t5                       |    0 |   8 |    1 |          89.23 |   0.4% |   8.3% |       34222 |          720.94 |   2.00
whisper                  |    0 |   8 |    1 |         628.32 |   2.5% |  54.7% |        9134 |         5063.33 |   1.00
lightning                |    0 |   8 |    1 |        1132.50 |   1.1% |  29.9% |       27076 |         9127.99 |   1.00
lightning-gpus           |    0 |   1 |    8 |        7289.49 |   5.0% |  49.2% |       30518 |         7289.49 |   1.00
llava-single             |    0 |   8 |    1 |           3.96 |   0.4% |   9.4% |       72614 |           31.99 |   1.00
llama                    |    0 |   8 |    1 |         780.18 |   4.8% |  86.4% |       27608 |         5881.50 |   1.00
llm-full-mp-gpus         |    0 |   1 |    8 |         452.01 |   2.4% |  12.9% |       49212 |          452.01 |   1.00
llm-lora-ddp-gpus        |    0 |   1 |    8 |       29802.19 |   0.4% |   1.9% |       39352 |        29802.19 |   1.00
llm-lora-mp-gpus         |    0 |   1 |    8 |        3832.59 |   1.7% |   9.1% |       55454 |         3832.59 |   1.00
llm-lora-single          |    0 |   8 |    1 |        5053.44 |   0.2% |   3.7% |       49732 |        40477.18 |   1.00
pna                      |    0 |   8 |    1 |        6729.25 |   0.6% |  12.9% |       39556 |        53916.44 |   1.00
ppo                      |    0 |   8 |    1 |    39135353.61 |   0.5% |  57.7% |       62426 |    313082965.88 |   1.00
recursiongfn             |    0 |   8 |    1 |       12157.56 |   1.0% |  22.1% |       19124 |        97983.18 |   1.00
rlhf-gpus                |    0 |   1 |    8 |       11769.89 |   1.6% |   8.8% |       24232 |        11769.89 |   1.00
rlhf-single              |    0 |   8 |    1 |        2350.64 |   0.7% |  16.6% |       19272 |        18861.54 |   1.00
focalnet                 |    0 |   8 |    1 |         624.32 |   0.7% |  15.2% |       24524 |         5057.26 |   2.00
torchatari               |    0 |   8 |    1 |        6724.94 |   0.6% |  13.9% |        3690 |        53618.47 |   1.00
convnext_large-fp16      |    0 |   8 |    1 |         624.73 |   1.5% |  23.5% |       27248 |         5122.43 |   0.00
convnext_large-fp32      |    0 |   8 |    1 |         128.92 |   0.8% |  11.8% |       47170 |         1051.86 |   0.00
convnext_large-tf32      |    0 |   8 |    1 |         237.17 |   0.9% |  14.5% |       49506 |         1940.57 |   0.00
convnext_large-tf32-fp16 |    0 |   8 |    1 |         620.32 |   1.5% |  23.4% |       27248 |         5083.03 |   3.00
regnet_y_128gf           |    0 |   8 |    1 |         187.86 |   0.4% |   9.5% |       28740 |         1519.98 |   2.00
resnet152-ddp-gpus       |    0 |   1 |    8 |        6768.72 |   4.9% |  37.6% |       30964 |         6768.72 |   0.00
resnet50                 |    0 |   8 |    1 |        1654.01 |   1.6% |  34.5% |       14532 |        13381.48 |   1.00
resnet50-noio            |    0 |   8 |    1 |        2050.15 |   0.0% |   2.1% |       27354 |        16411.22 |   0.00
vjepa-gpus               |    0 |   1 |    8 |         238.46 |   6.2% |  49.5% |       65400 |          238.46 |   1.00
vjepa-single             |    0 |   8 |    1 |          37.62 |   1.5% |  34.1% |       58696 |          303.88 |   1.00