[clang] [MLIR][NVGPU] Introduce `nvgpu.wargroup.mma.store` Op for Hopper GPUs (PR #65441)
Guray Ozen via cfe-commits
cfe-commits at lists.llvm.org
Thu Oct 5 01:26:04 PDT 2023
================
@@ -53,6 +55,16 @@ static Value truncToI32(ConversionPatternRewriter &rewriter, Location loc,
return rewriter.create<LLVM::TruncOp>(loc, rewriter.getI32Type(), value);
}
+/// Returns warp-size as a value.
+static Value getWarpSizeValue(ImplicitLocOpBuilder &b) {
----------------
grypp wrote:
it was a bad idea to put a singleton. I thought with my previous compiler's logic, my bad. removed
https://github.com/llvm/llvm-project/pull/65441
More information about the cfe-commits
mailing list