Skip to content

[MatmulLoopPipeline] Populate LoadOp mask to PrefetchOp#4030

Merged
whitneywhtsang merged 3 commits intomainfrom
whitneywhtsang/prefetch_pipeline
Apr 28, 2025
Merged

[MatmulLoopPipeline] Populate LoadOp mask to PrefetchOp#4030
whitneywhtsang merged 3 commits intomainfrom
whitneywhtsang/prefetch_pipeline

Conversation

@whitneywhtsang
Copy link
Copy Markdown
Contributor

@whitneywhtsang whitneywhtsang commented Apr 27, 2025

This PR enhances MatmulLoopPipeline to make it create PrefetchOp operations with mask from associated LoadOp.
Benchmark CI:
https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/14697631543
https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/14716472373
(No performance regressions.)

Note: this change comes partially from #3634.

Signed-off-by: Whitney Tsang <whitney.tsang@intel.com>
@whitneywhtsang whitneywhtsang self-assigned this Apr 27, 2025
@whitneywhtsang whitneywhtsang marked this pull request as ready for review April 28, 2025 00:18
@whitneywhtsang whitneywhtsang requested review from a team, alexbaden, chengjunlu and etiotto April 28, 2025 00:21
Comment thread test/TritonIntelGPU/loop-pipeline.mlir Outdated
Comment thread test/TritonIntelGPU/loop-pipeline.mlir Outdated
Signed-off-by: Whitney Tsang <whitney.tsang@intel.com>
@whitneywhtsang whitneywhtsang requested a review from etiotto April 28, 2025 16:45
@whitneywhtsang whitneywhtsang merged commit d4699e1 into main Apr 28, 2025
10 checks passed
@whitneywhtsang whitneywhtsang deleted the whitneywhtsang/prefetch_pipeline branch April 28, 2025 21:44
@etiotto etiotto linked an issue May 2, 2025 that may be closed by this pull request
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Performance] Enable prefetching for tt.load with tensor of pointer

4 participants