[Mlir-commits] [mlir] [MLIR][XeGPU] Unroll Dpasmx Op (PR #195179)
Charitha Saumya
llvmlistbot at llvm.org
Fri May 1 11:03:36 PDT 2026
================
@@ -410,6 +410,111 @@ struct UnrollDpasOp : public UnrollPattern<xegpu::DpasOp> {
}
};
+struct UnrollDpasMxOp : public UnrollPattern<xegpu::DpasMxOp> {
+ using UnrollPattern<xegpu::DpasMxOp>::UnrollPattern;
+ LogicalResult matchAndRewrite(xegpu::DpasMxOp op,
+ PatternRewriter &rewriter) const override {
+ Location loc = op.getLoc();
+ if (llvm::any_of(op->getOperandTypes(), [&](Type type) {
+ auto vecTy = dyn_cast<VectorType>(type);
+ return !vecTy || vecTy.getRank() != 2;
+ }))
+ return failure();
----------------
charithaintc wrote:
is it possible to have non vector type in the op definition?
https://github.com/llvm/llvm-project/pull/195179
More information about the Mlir-commits
mailing list