Skip to content

Enable SLPVectorization #4166

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

Enable SLPVectorization #4166

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 7 additions & 5 deletions third_party/intel/triton_xpu.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,11 +194,13 @@ void init_triton_intel(py::module &&m) {
tuningOptions.LoopUnrolling = true;
tuningOptions.LoopInterleaving = true;
tuningOptions.LoopVectorization = true;
// SLPVectorizer causes test_core.py::test_dot_mulbroadcasted to fail.
// It vectorizes @llvm.fmuladd.f32 with @llvm.fmuladd.v32f32. We can
// consider to reenable SLP vectorization when the failure is
// investigated.
tuningOptions.SLPVectorization = false;
// TODO: currently we run SLP vectorizer with an empty target machine.
// This cause the vectorizer to create larger vector which could be bad.
// Disabling it would currently cause regressions as this pass also
// applies some scheduling that helps performance in some cases. We
// should work on using NVPTX target instead and address the performance
// regressions with some scheduling solution.
tuningOptions.SLPVectorization = true;

PassBuilder pb(nullptr /*targetMachine*/, tuningOptions, std::nullopt,
instrCbPtr);
Expand Down
Loading