Skip to content

[BACKEND][NVIDIA][NFC] Remove BarrierOpConversion conversion as dead code #5135

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
peterbell10 merged 1 commit into from
Nov 13, 2024
+0 −23
Merged

[BACKEND][NVIDIA][NFC] Remove BarrierOpConversion conversion as dead code #5135

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
23 changes: 0 additions & 23 deletions third_party/nvidia/lib/TritonNVIDIAGPUToLLVM/BarrierOpToLLVM.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,28 +33,6 @@ using namespace mlir;
using namespace mlir::triton;

namespace {
struct BarrierOpConversion
: public ConvertOpToLLVMPattern<mlir::gpu::BarrierOp> {
using ConvertOpToLLVMPattern::ConvertOpToLLVMPattern;

LogicalResult
matchAndRewrite(mlir::gpu::BarrierOp op, OpAdaptor adaptor,
ConversionPatternRewriter &rewriter) const override {
Location loc = op->getLoc();
if (op->hasAttr("bar_id")) {
// llvm.nvvm.barrier0 doesn't support bar_id and num_threads attributes,
// so we have to lower it to ptx manually.
auto barId = op->getAttrOfType<IntegerAttr>("bar_id").getInt();
auto numThreads = op->getAttrOfType<IntegerAttr>("num_threads").getInt();
barSync(rewriter, op, barId, numThreads);
rewriter.eraseOp(op);
return success();
}
// Otherwise we let the default lowering handle it
return failure();
}
};

struct FenceAsyncSharedOpConversion
: public ConvertOpToLLVMPattern<triton::nvidia_gpu::FenceAsyncSharedOp> {
using ConvertOpToLLVMPattern<
Expand Down Expand Up @@ -193,7 +171,6 @@ struct WaitBarrierOpConversion
void mlir::triton::NVIDIA::populateBarrierOpToLLVMPatterns(
LLVMTypeConverter &typeConverter, RewritePatternSet &patterns,
PatternBenefit benefit) {
patterns.add<BarrierOpConversion>(typeConverter, benefit);
patterns.add<FenceAsyncSharedOpConversion>(typeConverter, benefit);
patterns.add<InitBarrierOpConversion, InvalBarrierOpConversion>(typeConverter,
benefit);
Expand Down