Intel GPU Max Support + Gaudi

[Delaunay]

No description provided.

[Delaunay]

bench                          | fail | n |       perf |   sem% |   std% | peak_memory |      score | weight
bert-fp16                      |    0 | 8 |     264.75 |   1.0% |  15.1% |       19938 |    2172.03 |  0.00
bert-fp32                      |    0 | 8 |      44.88 |   0.6% |   9.0% |       25070 |     364.56 |  0.00
bert-tf32                      |    0 | 8 |     142.02 |   0.9% |  13.7% |       25070 |    1162.53 |  0.00
bert-tf32-fp16                 |    0 | 8 |     264.76 |   1.0% |  15.2% |       19938 |    2172.50 |  3.00
bf16                           |    0 | 8 |     293.18 |   0.3% |   7.1% |        5688 |    2361.24 |  0.00
brax                           |    0 | 1 |  726105.43 |   0.5% |   4.2% |        6448 |  726105.43 |  1.00
convnext_large-fp16            |    0 | 8 |     314.87 |   1.9% |  30.0% |       31292 |    2573.95 |  0.00
convnext_large-fp32            |    0 | 8 |      60.68 |   0.7% |  11.7% |       51200 |     495.05 |  0.00
convnext_large-tf32            |    0 | 8 |     157.86 |   1.0% |  15.3% |       53566 |    1295.69 |  0.00
convnext_large-tf32-fp16       |    0 | 8 |     327.89 |   1.6% |  25.3% |       31292 |    2683.47 |  3.00
davit_large                    |    0 | 8 |     322.97 |   0.6% |  14.1% |       34118 |    2615.24 |  1.00
davit_large-multi              |    0 | 1 |    2306.64 |   2.0% |  15.6% |       36224 |    2306.64 |  5.00
dlrm                           |    0 | 1 |  350872.23 |   3.4% |  26.3% |        7218 |  350872.23 |  1.00
focalnet                       |    0 | 8 |     365.13 |   0.8% |  17.7% |       27336 |    2956.17 |  2.00
fp16                           |    0 | 8 |     289.92 |   0.2% |   4.7% |        5688 |    2329.76 |  0.00
fp32                           |    0 | 8 |      19.13 |   0.0% |   1.1% |        6066 |     153.19 |  0.00
llama                          |    0 | 8 |     723.50 |   7.5% | 135.1% |       32166 |    5286.20 |  1.00
opt-1_3b                       |    0 | 1 |      29.98 |   0.3% |   1.5% |       41854 |      29.98 |  5.00
opt-6_7b                       |    0 | 1 |      14.01 |   3.2% |  17.0% |       60388 |      14.01 |  5.00
reformer                       |    0 | 8 |      62.22 |   0.3% |   6.1% |       29304 |     501.34 |  1.00
regnet_y_128gf                 |    0 | 8 |     115.86 |   0.5% |  10.1% |       35514 |     937.89 |  2.00
resnet152                      |    0 | 8 |     550.50 |   0.9% |  19.6% |       31204 |    4451.79 |  1.00
resnet152-ddp                  |    0 | 1 |    3098.23 |   2.3% |  17.5% |          -1 |    3098.23 |  0.00
resnet152-multi                |    0 | 1 |    3236.91 |   3.2% |  25.3% |       34020 |    3236.91 |  5.00
resnet50                       |    0 | 8 |    1155.10 |   1.0% |  22.8% |       18628 |    9362.48 |  1.00
resnet50-noio                  |    0 | 8 |    1160.48 |   0.3% |   7.4% |       31350 |    9365.89 |  0.00
rwkv                           |    0 | 8 |     430.59 |   0.2% |   3.9% |        9660 |    3446.78 |  1.00
stargan                        |    0 | 8 |      53.80 |   2.5% |  54.9% |       41036 |     427.12 |  1.00
super-slomo                    |    8 | 8 |        nan |   nan% |   nan% |       33934 |        nan |  1.00
t5                             |    0 | 8 |      51.31 |   0.4% |   9.5% |       38290 |     415.17 |  2.00
tf32                           |    0 | 8 |     146.41 |   0.2% |   4.8% |        6066 |    1176.71 |  0.00
whisper                        |    0 | 8 |     422.75 |   1.7% |  38.0% |       13178 |    3408.53 |  1.00

[Delaunay]

=================
Benchmark results
=================
bench                          | fail |   n |       perf |   sem% |   std% | peak_memory |      score | weight
bert-fp16                      |    0 |   8 |     161.31 |   0.9% |  13.8% |       24600 |    1320.71 |  0.00
bert-fp32                      |    0 |   8 |      40.09 |   0.6% |   8.8% |       31564 |     325.56 |  0.00
bert-tf32                      |    0 |   8 |     130.19 |   0.9% |  14.2% |       31566 |    1066.62 |  0.00
bert-tf32-fp16                 |    0 |   8 |     161.24 |   0.9% |  13.7% |       24600 |    1319.90 |  3.00
bf16                           |    0 |   8 |     293.35 |   0.3% |   7.5% |        1788 |    2363.43 |  0.00
brax                           |    0 |   1 |  723557.96 |   0.6% |   4.3% |        2676 |  723557.96 |  1.00
convnext_large-fp16            |    0 |   8 |     308.42 |   2.1% |  32.4% |       27392 |    2520.06 |  0.00
convnext_large-fp32            |    0 |   8 |      60.80 |   0.7% |  11.3% |       47300 |     495.68 |  0.00
convnext_large-tf32            |    0 |   8 |     157.81 |   0.9% |  14.7% |       49666 |    1293.85 |  0.00
convnext_large-tf32-fp16       |    0 |   8 |     320.54 |   1.7% |  26.3% |       27392 |    2622.50 |  3.00
davit_large                    |    0 |   8 |     316.53 |   0.7% |  15.6% |       30218 |    2562.53 |  1.00
davit_large-multi              |    0 |   1 |    2318.14 |   2.0% |  15.5% |       32324 |    2318.14 |  5.00
dlrm                           |    0 |   1 |  378082.39 |   2.9% |  22.5% |        3724 |  378082.39 |  1.00
focalnet                       |    0 |   8 |     365.67 |   0.7% |  16.8% |       23616 |    2960.51 |  2.00
fp16                           |    0 |   8 |     289.44 |   0.2% |   4.9% |        1788 |    2326.17 |  0.00
fp32                           |    0 |   8 |      19.14 |   0.0% |   1.0% |        2166 |     153.26 |  0.00
llama                          |    0 |   8 |     774.58 |   7.9% | 142.0% |       28424 |    5565.06 |  1.00
opt-1_3b                       |    0 |   1 |      29.22 |   1.7% |   9.2% |       38222 |      29.22 |  5.00
opt-1_3b-multinode             |  nan | nan |        nan |   nan% |   nan% |         nan |        nan | 10.00
opt-6_7b                       |    0 |   1 |      14.62 |   2.2% |  11.7% |       57098 |      14.62 |  5.00
opt-6_7b-multinode             |  nan | nan |        nan |   nan% |   nan% |         nan |        nan | 10.00
reformer                       |    0 |   8 |      62.32 |   0.3% |   5.9% |       25404 |     502.08 |  1.00
regnet_y_128gf                 |    0 |   8 |     115.36 |   0.5% |  10.6% |       31614 |     934.34 |  2.00
resnet152                      |    0 |   8 |     535.94 |   1.0% |  22.0% |       27310 |    4332.02 |  1.00
resnet152-ddp                  |    0 |   1 |    3395.01 |   2.1% |  16.1% |          -1 |    3395.01 |  0.00
resnet152-multi                |    0 |   1 |    3440.19 |   3.4% |  27.2% |       30128 |    3440.19 |  5.00
resnet50                       |    0 |   8 |    1167.32 |   0.8% |  18.7% |       14738 |    9464.09 |  1.00
resnet50-noio                  |    0 |   8 |    1162.24 |   0.3% |   7.2% |       27450 |    9377.80 |  0.00
stargan                        |    0 |   8 |      50.62 |   2.7% |  58.9% |       37416 |     402.78 |  1.00
super-slomo                    |    0 |   8 |      36.20 |   2.0% |  44.0% |       66072 |     290.66 |  1.00
t5                             |    0 |   8 |      51.35 |   0.4% |   9.5% |       34390 |     415.44 |  2.00
tf32                           |    0 |   8 |     146.58 |   0.2% |   4.9% |        2166 |    1178.04 |  0.00
whisper                        |    0 |   8 |     198.28 |   1.5% |  34.0% |       36724 |    1598.56 |  1.00