Skip to content

[mlir] Add forall canonicalization to replace constant induction vars #112764

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

Merged
merged 4 commits into from
Oct 18, 2024
Merged

[mlir] Add forall canonicalization to replace constant induction vars #112764

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a00ac08
[mlir] Add forall canonicalization to replace constant induction vars
Max191 Oct 16, 2024
43ece6b
add non zero test
Max191 Oct 17, 2024
8350c93
address comment
Max191 Oct 17, 2024
731a481
move comment
Max191 Oct 18, 2024
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
28 changes: 27 additions & 1 deletion mlir/lib/Dialect/SCF/IR/SCF.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1767,6 +1767,31 @@ struct ForallOpSingleOrZeroIterationDimsFolder
}
};

/// Replace all induction vars with a single trip count with their lower bound.
struct ForallOpReplaceConstantInductionVar : public OpRewritePattern<ForallOp> {
using OpRewritePattern<ForallOp>::OpRewritePattern;

LogicalResult matchAndRewrite(ForallOp op,
PatternRewriter &rewriter) const override {
Location loc = op.getLoc();
bool changed = false;
for (auto [lb, ub, step, iv] :
llvm::zip(op.getMixedLowerBound(), op.getMixedUpperBound(),
op.getMixedStep(), op.getInductionVars())) {
if (iv.getUses().begin() == iv.getUses().end())
continue;
auto numIterations = constantTripCount(lb, ub, step);
if (!numIterations.has_value() || numIterations.value() != 1) {
continue;
}
rewriter.replaceAllUsesWith(
iv, getValueOrCreateConstantIndexOp(rewriter, loc, lb));
changed = true;
}
return success(changed);
}
};

struct FoldTensorCastOfOutputIntoForallOp
: public OpRewritePattern<scf::ForallOp> {
using OpRewritePattern<scf::ForallOp>::OpRewritePattern;
Expand Down Expand Up @@ -1851,7 +1876,8 @@ void ForallOp::getCanonicalizationPatterns(RewritePatternSet &results,
MLIRContext *context) {
results.add<DimOfForallOp, FoldTensorCastOfOutputIntoForallOp,
ForallOpControlOperandsFolder, ForallOpIterArgsFolder,
ForallOpSingleOrZeroIterationDimsFolder>(context);
ForallOpSingleOrZeroIterationDimsFolder,
ForallOpReplaceConstantInductionVar>(context);
}

/// Given the region at `index`, or the parent operation if `index` is None,
Expand Down
4 changes: 3 additions & 1 deletion mlir/test/Dialect/SCF/canonicalize.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1617,7 +1617,7 @@ func.func @do_not_inline_distributed_forall_loop(
%in: tensor<8x8xf32>) -> tensor<8x8xf32> {
%cst = arith.constant 0.000000e+00 : f32
%0 = tensor.empty() : tensor<8x8xf32>
%1 = scf.forall (%i, %j) = (0, 0) to (1, 1) step (8, 8)
%1 = scf.forall (%i, %j) = (0, 4) to (1, 5) step (8, 8)
shared_outs (%out_ = %0) -> (tensor<8x8xf32>) {
%slice = tensor.extract_slice %out_[%i, %j] [2, 3] [1, 1]
: tensor<8x8xf32> to tensor<2x3xf32>
Expand All @@ -1632,6 +1632,8 @@ func.func @do_not_inline_distributed_forall_loop(
}
// CHECK-LABEL: @do_not_inline_distributed_forall_loop
// CHECK: scf.forall
// CHECK: tensor.extract_slice %{{.*}}[0, 4] [2, 3] [1, 1]
// CHECK: tensor.parallel_insert_slice %{{.*}}[0, 4] [2, 3] [1, 1]

// -----

Expand Down
Loading