gaspersic/triton
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[Backend] Add value cache in emitting indices calculation and some refinement (#1018)

Browse Source 
1, add explicit value cache in emitting indices calculation;
2, move the indices calculation emitting logics into
ConvertTritonGPUOpToLLVMPatternBase to avoid the redundant build cost by
templates. Refer to the discussion in this thread by @LyricZhao :
https://triton-lang.slack.com/archives/C042VBSQWNS/p1671336755922969
This commit is contained in:
goostavz
2022-12-30 03:19:59 +08:00
committed by GitHub
parent 2ba74d2729
commit 1d3029faf8
11 changed files with 355 additions and 180 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
lib/Conversion/TritonGPUToLLVM/ReduceOpToLLVM.h
View File

@@ -6,10 +6,11 @@
using namespace mlir;
using namespace mlir::triton;
void populateReduceOpToLLVMPatterns(mlir::LLVMTypeConverter &typeConverter,
RewritePatternSet &patterns, int numWarps,
AxisInfoAnalysis &axisInfoAnalysis,
const Allocation *allocation, Value smem,
PatternBenefit benefit);
void populateReduceOpToLLVMPatterns(
mlir::LLVMTypeConverter &typeConverter, RewritePatternSet &patterns,
int numWarps, AxisInfoAnalysis &axisInfoAnalysis,
const Allocation *allocation, Value smem,
ConvertTritonGPUOpToLLVMPatternBase::IndexCacheInfo &indexCacheInfo,
PatternBenefit benefit);
#endif