Skip to content

Limit prefetch to only dense memory#4009

Merged
whitneywhtsang merged 3 commits intomainfrom
whitneywhtsang/prefetch_reduce
Apr 25, 2025
Merged

Limit prefetch to only dense memory#4009
whitneywhtsang merged 3 commits intomainfrom
whitneywhtsang/prefetch_reduce

Conversation

@whitneywhtsang
Copy link
Copy Markdown
Contributor

@whitneywhtsang whitneywhtsang commented Apr 24, 2025

This PR limits prefetch to only dense memory, to avoid polluting the cache.
Benchmark CI: https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/14647344471
No performance impact to key benchmarks.

Note: this change comes partially from #3634.

@whitneywhtsang whitneywhtsang self-assigned this Apr 24, 2025
Signed-off-by: Whitney Tsang <whitney.tsang@intel.com>
@whitneywhtsang whitneywhtsang force-pushed the whitneywhtsang/prefetch_reduce branch from 2a87e2e to 7a815f8 Compare April 24, 2025 17:30
@whitneywhtsang whitneywhtsang marked this pull request as ready for review April 24, 2025 18:34
@whitneywhtsang whitneywhtsang requested review from a team, alexbaden, chengjunlu and etiotto April 24, 2025 18:35
Comment thread third_party/intel/lib/TritonIntelGPUTransforms/Pipeliner/MatmulLoopPipeline.cpp Outdated
Comment thread third_party/intel/lib/TritonIntelGPUTransforms/Pipeliner/MatmulLoopPipeline.cpp Outdated
Signed-off-by: Whitney Tsang <whitney.tsang@intel.com>
@whitneywhtsang whitneywhtsang requested a review from etiotto April 24, 2025 20:23
@whitneywhtsang whitneywhtsang merged commit 11eff0a into main Apr 25, 2025
9 checks passed
@whitneywhtsang whitneywhtsang deleted the whitneywhtsang/prefetch_reduce branch April 25, 2025 01:33
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.

3 participants