Skip to content

Commit

Permalink
[intel] Remove ClusterOpstoLLVM pass (triton-lang#995)
Browse files Browse the repository at this point in the history
The pass lowers nvidia_gpu operations, which is not needed for intel
backend.

Signed-off-by: Whitney Tsang <whitney.tsang@intel.com>
  • Loading branch information
whitneywhtsang authored Apr 29, 2024
1 parent 1ba1d23 commit 9d6d945
Show file tree
Hide file tree
Showing 4 changed files with 0 additions and 69 deletions.
1 change: 0 additions & 1 deletion third_party/intel/lib/TritonIntelGPUToLLVM/CMakeLists.txt
Original file line number Diff line number Diff line change
@@ -1,6 +1,5 @@
add_triton_library(TritonIntelGPUToLLVM
AllocateSharedMemory.cpp
ClusterOpsToLLVM.cpp
ControlFlowOpToLLVM.cpp
ConvertLayoutOpToLLVM/SharedToDotOperandDPAS.cpp
ConvertLayoutOpToLLVM.cpp
Expand Down
63 changes: 0 additions & 63 deletions third_party/intel/lib/TritonIntelGPUToLLVM/ClusterOpsToLLVM.cpp

This file was deleted.

Original file line number Diff line number Diff line change
Expand Up @@ -18,10 +18,6 @@ void populateBarrierOpToLLVMPatterns(
TritonIntelGPUToLLVMTypeConverter &typeConverter,
RewritePatternSet &patterns, PatternBenefit benefit);

void populateClusterOpsToLLVMPatterns(
TritonIntelGPUToLLVMTypeConverter &typeConverter,
RewritePatternSet &patterns, PatternBenefit benefit);

void populateConvertLayoutOpToLLVMPatterns(LLVMTypeConverter &typeConverter,
RewritePatternSet &patterns,
PatternBenefit benefit);
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -198,7 +198,6 @@ class TritonGPUToLLVMPipelineManager {

intel::populateTensorPtrOpsToLLVMPatterns(typeConverter, patterns,
benefit);
intel::populateClusterOpsToLLVMPatterns(typeConverter, patterns, benefit);
intel::populateHistogramOpToLLVMPatterns(typeConverter, patterns,
benefit);
intel::populatePrintOpToLLVMPattern(typeConverter, patterns, targetInfo,
Expand Down

0 comments on commit 9d6d945

Please sign in to comment.