Skip to content

[webgpu] Use workgroup_idx instead of workgroup_id.x#23696

Merged
guschmue merged 2 commits intomicrosoft:mainfrom
qjia7:fix_quantize_shader
Feb 14, 2025
Merged

[webgpu] Use workgroup_idx instead of workgroup_id.x#23696
guschmue merged 2 commits intomicrosoft:mainfrom
qjia7:fix_quantize_shader

Conversation

@qjia7
Copy link
Contributor

@qjia7 qjia7 commented Feb 14, 2025

We should always use workgroup_idx instead of workgroup_id.x in cause the dispatched workgroups are normalized.

When the input is large enough, the 1d workgroups will be normalized to 2d/3d and results incorrect result.

We should always use workgroup_idx instead of workgroup_id.x in cause the dispatched workgroups are normalized.
@qjia7
Copy link
Contributor Author

qjia7 commented Feb 14, 2025

@sushraja-msft @guschmue please take a look, thanks.

@guschmue
Copy link
Contributor

/azp run ONNX Runtime Web CI Pipeline,Windows GPU CI Pipeline,Linux Android Emulator QNN CI Pipeline

@azure-pipelines
Copy link

Azure Pipelines successfully started running 2 pipeline(s).

@guschmue
Copy link
Contributor

/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

@guschmue
Copy link
Contributor

/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

@guschmue
Copy link
Contributor

/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
Copy link

Azure Pipelines successfully started running 4 pipeline(s).

1 similar comment
@azure-pipelines
Copy link

Azure Pipelines successfully started running 4 pipeline(s).

@azure-pipelines
Copy link

Azure Pipelines successfully started running 9 pipeline(s).

@guschmue guschmue added the ep:WebGPU ort-web webgpu provider label Feb 14, 2025
@guschmue guschmue merged commit 4f66610 into microsoft:main Feb 14, 2025
76 checks passed
@qjia7 qjia7 deleted the fix_quantize_shader branch February 15, 2025 01:29
guschmue pushed a commit that referenced this pull request Mar 6, 2025
We should always use workgroup_idx instead of workgroup_id.x in cause
the dispatched workgroups are normalized.

When the input is large enough, the 1d workgroups will be normalized to
2d/3d and results incorrect result.
ashrit-ms pushed a commit that referenced this pull request Mar 17, 2025
We should always use workgroup_idx instead of workgroup_id.x in cause
the dispatched workgroups are normalized.

When the input is large enough, the 1d workgroups will be normalized to
2d/3d and results incorrect result.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ep:WebGPU ort-web webgpu provider

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants