Skip to content
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

Bump llvm-project to f66cd9e9556a53142a26a5c21a72e21f1579217c. #2466

Merged
merged 3 commits into from
Sep 19, 2023
Merged

Bump llvm-project to f66cd9e9556a53142a26a5c21a72e21f1579217c. #2466

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
ff0f964
Bump llvm-project to f66cd9e9556a53142a26a5c21a72e21f1579217c.
stellaraccident Sep 15, 2023
6f56c6a
Pin to correct stablehlo fork.
stellaraccident Sep 15, 2023
559fecf
Bump llvm-project to f66cd9e9556a53142a26a5c21a72e21f1579217c
stellaraccident Sep 15, 2023
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
2 changes: 1 addition & 1 deletion .gitmodules
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,4 +3,4 @@
url = https://github.com/llvm/llvm-project.git
[submodule "externals/stablehlo"]
path = externals/stablehlo
url = https://github.com/openxla/stablehlo.git
url = https://github.com/shark-infra/stablehlo.git
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What's the reasoning for this?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have the same confusion about this.

Copy link
Collaborator Author

@stellaraccident stellaraccident Sep 20, 2023 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wait, I'm confused and did something wrong. Let me revert this.

Reconstructing what I think was going on here: my local stablehlo git repo was reporting a commit not found and that was because my repository had a remote of a throwaway fork of stablehlo that I had used locally when doing the switch from mhlo. I saw that it was the old fork and updated, thinking that I had accidentally committed the use of the throwaway fork. I switched it to the fork we use for patching on another project, thinking I needed to land some patches on the branch, but it wasn't intended to land this way.

In any case, unintentional sequence of events matching work I was doing on a peer project. I apologize -- the change was not intentional (and that repo it is pointing to is a pure mirror of stablehlo).

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Resetting it here: https://github.com/llvm/torch-mlir/pull/2477/files

2 changes: 1 addition & 1 deletion externals/llvm-external-projects/torch-mlir-dialects/lib/Dialect/TMTensor/IR/TMTensorOps.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,7 +233,7 @@ LogicalResult AttentionOp::generateScalarImplementation(OpBuilder &b,
loc, init,
[&](OpBuilder &b, Location loc, Value elem, Value acc) {
Value x = b.create<memref::LoadOp>(loc, weight, localIVs);
Value max = b.create<arith::MaxFOp>(loc, x, acc);
Value max = b.create<arith::MaximumFOp>(loc, x, acc);
b.create<scf::ReduceReturnOp>(loc, max);
});
})
Expand Down
2 changes: 1 addition & 1 deletion externals/llvm-project
Open in desktop
Submodule llvm-project updated 4670 files
8 changes: 4 additions & 4 deletions lib/Conversion/TorchToLinalg/Reduction.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -176,8 +176,8 @@ class ConvertAtenMaxDimOp : public OpConversionPattern<AtenMaxDimOp> {

Value resultMax, predicate;
if (inElementType.isa<mlir::FloatType>()) {
resultMax =
rewriter.create<arith::MaxFOp>(nestedLoc, newValue, oldValue);
resultMax = rewriter.create<arith::MaximumFOp>(nestedLoc, newValue,
oldValue);
predicate = rewriter.create<arith::CmpFOp>(
nestedLoc, arith::CmpFPredicate::OGT, newValue, oldValue);
} else {
Expand Down Expand Up @@ -280,7 +280,7 @@ static Value createLinalgPayloadForReduceOp(OpBuilder &b, Location loc,
convertScalarToDtype(b, loc, payloadArgs[0], resultElementType);
Value result = payloadArgs[1];
if (resultElementType.isa<mlir::FloatType>())
return b.create<arith::MaxFOp>(loc, self, result);
return b.create<arith::MaximumFOp>(loc, self, result);
else if (resultElementType.isa<mlir::IntegerType>()) {
IntegerType intType = max.getSelf()
.getType()
Expand All @@ -297,7 +297,7 @@ static Value createLinalgPayloadForReduceOp(OpBuilder &b, Location loc,
convertScalarToDtype(b, loc, payloadArgs[0], resultElementType);
Value result = payloadArgs[1];
if (resultElementType.isa<mlir::FloatType>())
return b.create<arith::MinFOp>(loc, self, result);
return b.create<arith::MinimumFOp>(loc, self, result);
else if (resultElementType.isa<mlir::IntegerType>()) {
IntegerType intType = min.getSelf()
.getType()
Expand Down
4 changes: 2 additions & 2 deletions lib/Conversion/TorchToTMTensor/TorchToTMTensor.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1332,15 +1332,15 @@ class ConvertAtenScatterReduceTwoOp
if (update.getType().isa<mlir::IntegerType>()) {
result = b.create<arith::MaxSIOp>(loc, update, current);
} else if (update.getType().isa<mlir::FloatType>()) {
result = b.create<arith::MaxFOp>(loc, update, current);
result = b.create<arith::MaximumFOp>(loc, update, current);
} else {
llvm_unreachable("Only integer/float types supported!");
}
} else if (reduceEnum == torch_upstream::ReductionType::MIN) {
if (update.getType().isa<mlir::IntegerType>()) {
result = b.create<arith::MinSIOp>(loc, update, current);
} else if (update.getType().isa<mlir::FloatType>()) {
result = b.create<arith::MinFOp>(loc, update, current);
result = b.create<arith::MinimumFOp>(loc, update, current);
} else {
llvm_unreachable("Only integer/float types supported!");
}
Expand Down