[Mlir-commits] [mlir] [MLIR][XeGPU] Allow load/store/prefetch uses [memref+offset] instead of tdesc (PR #150576)

Adam Siemieniuk llvmlistbot at llvm.org
Wed Jul 30 05:40:46 PDT 2025


================
@@ -644,9 +672,14 @@ LogicalResult CreateDescOp::verify() {
 //===----------------------------------------------------------------------===//
 LogicalResult PrefetchOp::verify() {
   auto tdescTy = getTensorDescType();
-  if (!tdescTy.isScattered())
+
+  if (tdescTy && !tdescTy.isScattered())
     return emitOpError("Expects a scattered TensorDesc.\n");
 
+  if (!tdescTy && getRankOf(getSource()) > 1)
+    return emitOpError(
+        "Expecting the source is a 1D memref or pointer (uint64_t).");
----------------
adam-smnk wrote:

nit: let's add invalid test cases for these 3 ops

https://github.com/llvm/llvm-project/pull/150576


More information about the Mlir-commits mailing list