Skip to content

[MLIR] [Vector] Disable canonicalization for vector.scatter with tensor output #168824

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

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

[MLIR] [Vector] Disable canonicalization for vector.scatter with tensor output #168824

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
6 changes: 6 additions & 0 deletions mlir/lib/Dialect/Vector/IR/VectorOps.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6087,6 +6087,9 @@ class ScatterFolder final : public OpRewritePattern<ScatterOp> {
using Base::Base;
LogicalResult matchAndRewrite(ScatterOp scatter,
PatternRewriter &rewriter) const override {
if (!isa<MemRefType>(scatter.getBase().getType()))
return failure();

switch (getMaskFormat(scatter.getMask())) {
case MaskFormat::AllTrue:
return failure(); // no unmasked equivalent
Expand All @@ -6107,6 +6110,9 @@ class FoldContiguousScatter final : public OpRewritePattern<ScatterOp> {
using Base::Base;
LogicalResult matchAndRewrite(ScatterOp op,
PatternRewriter &rewriter) const override {
if (!isa<MemRefType>(op.getBase().getType()))
return failure();

if (failed(isZeroBasedContiguousSeq(op.getIndices())))
return failure();

Expand Down