[webgpu/js] Optimize resize webgpu op & fix precision issues #23591
Conversation
xenova
changed the title
|
/azp run ONNX Runtime Web CI Pipeline,Windows GPU CI Pipeline,Linux Android Emulator QNN CI Pipeline |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run Linux CPU CI Pipeline,Linux CPU Minimal Build E2E CI Pipeline,Linux GPU CI Pipeline,Linux GPU TensorRT CI Pipeline,Linux OpenVINO CI Pipeline,Linux QNN CI Pipeline,MacOS CI Pipeline,Windows ARM64 QNN CI Pipeline,Windows CPU CI Pipeline |
|
/azp run Windows GPU TensorRT CI Pipeline,onnxruntime-binary-size-checks-ci-pipeline,orttraining-linux-ci-pipeline,orttraining-linux-gpu-ci-pipeline,orttraining-ortmodule-distributed,Windows x64 QNN CI Pipeline,Big Models |
|
Azure Pipelines could not run because the pipeline triggers exclude this branch/path. |
|
/azp run Windows GPU CUDA CI Pipeline,Windows GPU DML CI Pipeline,Windows GPU Doc Gen CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
Azure Pipelines could not run because the pipeline triggers exclude this branch/path. |
|
CI failed because I forgot to |
|
/azp run ONNX Runtime Web CI Pipeline,Windows GPU CI Pipeline,Linux Android Emulator QNN CI Pipeline |
|
/azp run Linux CPU CI Pipeline,Linux CPU Minimal Build E2E CI Pipeline,Linux GPU CI Pipeline,Linux GPU TensorRT CI Pipeline,Linux OpenVINO CI Pipeline,Linux QNN CI Pipeline,MacOS CI Pipeline,Windows ARM64 QNN CI Pipeline,Windows CPU CI Pipeline |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
Azure Pipelines could not run because the pipeline triggers exclude this branch/path. |
|
/azp run Windows GPU TensorRT CI Pipeline,onnxruntime-binary-size-checks-ci-pipeline,orttraining-linux-ci-pipeline,orttraining-linux-gpu-ci-pipeline,orttraining-ortmodule-distributed,Windows x64 QNN CI Pipeline,Big Models |
|
/azp run Windows GPU CUDA CI Pipeline,Windows GPU DML CI Pipeline,Windows GPU Doc Gen CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
Azure Pipelines could not run because the pipeline triggers exclude this branch/path. |
|
one unit test is unhappy: That test is part of the onnx test suite. You should be able to run this local: Yeah, a bit painful to make our CI happy :) |
|
4c6617a should fix it. @satyajandhyala I discovered two test cases for which the previous code would produce incorrect results, and I think it could be a good idea to add these to the CI. However, I'm unsure how to do this myself. (here's the model, same as the one in the opset10 test data). Test 1: Fractional upscale (e.g., 2.6) const feeds = {
X: new ort.Tensor(
'float32',
[1, 2, 3, 4, 5, 6, 7, 8],
[1, 1, 2, 4],
),
scales: new ort.Tensor(
'float32',
[1, 1, /* downscale */ 0.6, /* upscale */ 2.6], [4],
)
};Test 2: Specific integer upscale leading to floating point issues (300 in this case). A linear search reveals the smallest number this happens for is 37. Also happens for 41, 47, 55, 66, and many more. const feeds = {
X: new ort.Tensor(
'float32',
[1, 2, 3, 4, 5, 6, 7, 8],
[1, 1, 2, 4],
),
scales: new ort.Tensor(
'float32',
[1, 1, /* downscale */ 0.6, /* upscale */ 300], [4],
)
}; |
|
/azp run ONNX Runtime Web CI Pipeline,Windows GPU CI Pipeline,Linux Android Emulator QNN CI Pipeline |
|
/azp run Linux CPU CI Pipeline,Linux CPU Minimal Build E2E CI Pipeline,Linux GPU CI Pipeline,Linux GPU TensorRT CI Pipeline,Linux OpenVINO CI Pipeline,Linux QNN CI Pipeline,MacOS CI Pipeline,Windows ARM64 QNN CI Pipeline,Windows CPU CI Pipeline |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run Windows GPU TensorRT CI Pipeline,onnxruntime-binary-size-checks-ci-pipeline,orttraining-linux-ci-pipeline,orttraining-linux-gpu-ci-pipeline,orttraining-ortmodule-distributed,Windows x64 QNN CI Pipeline,Big Models |
|
Azure Pipelines could not run because the pipeline triggers exclude this branch/path. |
|
/azp run Windows GPU CUDA CI Pipeline,Windows GPU DML CI Pipeline,Windows GPU Doc Gen CI Pipeline, Win_TRT_Minimal_CUDA_Test_CI |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
Azure Pipelines could not run because the pipeline triggers exclude this branch/path. |
Description
This PR is a follow-up to #23488 and partially improves upon #23403. It does the following:
/decoder/decoder/generator/f0_upsamp/Resize_output_0results in differences at the end bounds due to precision issues when dividing 21600 by 72 (should be 300, but seemingly results in 299.999, which causes issues when flooring)Motivation and Context
I did a deep dive over the weekend to try fix Kokoro TTS on WebGPU and found that the above node had a large difference. Thinking this was a major issue, I spent some time fixing it. Turns out, it only happens for a small number of values, leading to high maximum error, but most values are correct (as seen here).
BEFORE:
AFTER:
So, although it has a very small impact on the final output (waveform), this bug could appear with other models in a more severe way.
BEFORE:
AFTER: