search

ROCm / AMDMIGraphX

AMD's graph optimization engine.
https://rocm.docs.amd.com/projects/AMDMIGraphX/en/latest/
MIT License
185 stars 86 forks source link

mlir improvement for FP8 and WMMA #3520

Closed causten closed 1 month ago

migraphx-bot commented 1 month ago
Test Batch Rate new
58ddea		 Rate old
b9fe91		 Diff Compare
torchvision-resnet50 64 3,261.05 3,263.44 -0.07% :white_check_mark:
torchvision-resnet50_fp16 64 6,987.89 6,990.24 -0.03% :white_check_mark:
torchvision-densenet121 32 2,436.33 2,434.81 0.06% :white_check_mark:
torchvision-densenet121_fp16 32 4,068.43 4,099.99 -0.77% :white_check_mark:
torchvision-inceptionv3 32 1,638.43 1,637.40 0.06% :white_check_mark:
torchvision-inceptionv3_fp16 32 2,761.98 2,760.57 0.05% :white_check_mark:
cadene-inceptionv4 16 776.80 775.96 0.11% :white_check_mark:
cadene-resnext64x4 16 808.98 808.44 0.07% :white_check_mark:
slim-mobilenet 64 7,539.01 7,533.29 0.08% :white_check_mark:
slim-nasnetalarge 64 211.81 211.48 0.16% :white_check_mark:
slim-resnet50v2 64 3,505.72 3,499.60 0.17% :white_check_mark:
bert-mrpc-onnx 8 1,146.60 1,154.75 -0.71% :white_check_mark:
bert-mrpc-tf 1 466.81 461.78 1.09% :white_check_mark:
pytorch-examples-wlang-gru 1 423.77 435.52 -2.70% :white_check_mark:
pytorch-examples-wlang-lstm 1 386.89 482.04 -19.74% :red_circle:
torchvision-resnet50_1 1 745.71 772.17 -3.43% :red_circle:
cadene-dpn92_1 1 425.43 401.69 5.91% :high_brightness:
cadene-resnext101_1 1 384.28 382.86 0.37% :white_check_mark:
onnx-taau-downsample 1 342.76 342.99 -0.07% :white_check_mark:
dlrm-criteoterabyte 1 33.33 33.32 0.03% :white_check_mark:
dlrm-criteoterabyte_fp16 1 52.78 52.73 0.09% :white_check_mark:
agentmodel 1 8,496.90 8,434.83 0.74% :white_check_mark:
unet_fp16 2 58.97 58.82 0.25% :white_check_mark:
resnet50v1_fp16 1 938.23 900.20 4.22% :high_brightness:
resnet50v1_int8 1 982.20 987.85 -0.57% :white_check_mark:
bert_base_cased_fp16 64 1,171.41 1,171.16 0.02% :white_check_mark:
bert_large_uncased_fp16 32 363.59 363.58 0.00% :white_check_mark:
bert_large_fp16 1 197.96 195.56 1.23% :white_check_mark:
distilgpt2_fp16 16 2,200.93 2,198.71 0.10% :white_check_mark:
yolov5s 1 545.14 543.82 0.24% :white_check_mark:
tinyllama 1 43.71 43.48 0.52% :white_check_mark:
vicuna-fastchat 1 174.40 177.52 -1.76% :white_check_mark:
whisper-tiny-encoder 1 418.60 418.61 -0.00% :white_check_mark:
whisper-tiny-decoder 1 434.43 428.00 1.50% :white_check_mark:

This build is not recommended to merge :red_circle:

migraphx-bot commented 1 month ago


     :white_check_mark: bert-mrpc-onnx: PASSED: MIGraphX meets tolerance
     :white_check_mark: bert-mrpc-tf: PASSED: MIGraphX meets tolerance
     :white_check_mark: pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance
     :white_check_mark: pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance
     :white_check_mark: torchvision-resnet50_1: PASSED: MIGraphX meets tolerance
     :white_check_mark: cadene-dpn92_1: PASSED: MIGraphX meets tolerance
     :white_check_mark: cadene-resnext101_1: PASSED: MIGraphX meets tolerance
     :white_check_mark: dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance
     :white_check_mark: agentmodel: PASSED: MIGraphX meets tolerance
     :white_check_mark: unet: PASSED: MIGraphX meets tolerance
     :white_check_mark: resnet50v1: PASSED: MIGraphX meets tolerance
     :white_check_mark: bert_base_cased_fp16: PASSED: MIGraphX meets tolerance
:red_circle:bert_large_uncased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

     :white_check_mark: bert_large: PASSED: MIGraphX meets tolerance
     :white_check_mark: yolov5s: PASSED: MIGraphX meets tolerance
     :white_check_mark: tinyllama: PASSED: MIGraphX meets tolerance
     :white_check_mark: vicuna-fastchat: PASSED: MIGraphX meets tolerance
     :white_check_mark: whisper-tiny-encoder: PASSED: MIGraphX meets tolerance
     :white_check_mark: whisper-tiny-decoder: PASSED: MIGraphX meets tolerance
     :white_check_mark: distilgpt2_fp16: PASSED: MIGraphX meets tolerance

codecov[bot] commented 1 month ago

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 92.08%. Comparing base (a1e3396) to head (58ddea4). Report is 4 commits behind head on develop.

Additional details and impacted files ```diff @@ Coverage Diff @@ ## develop #3520 +/- ## ======================================== Coverage 92.08% 92.08% ======================================== Files 510 510 Lines 21094 21094 ======================================== Hits 19424 19424 Misses 1670 1670 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.