[Mlir-commits] [mlir] [MLIR] Vector to XeGPU conversion: Use proper source variant for create_nd_tdesc op creation. (PR #171216)

[Jianhui Li]

================
@@ -102,18 +102,47 @@ static xegpu::CreateNdDescOp createNdDescriptor(PatternRewriter &rewriter,
                                                 xegpu::TensorDescType descType,
                                                 TypedValue<MemRefType> src) {
   MemRefType srcTy = src.getType();
+  assert(srcTy.isStrided() && "Expected strided memref type");
   auto [strides, offset] = srcTy.getStridesAndOffset();
+  bool isStatic = true;
+
+  // Memref is dynamic if any of its shape, offset or strides is dynamic.
+  if (!srcTy.hasStaticShape()) {
+    isStatic = false;
+  }
+
+  if (offset == ShapedType::kDynamic)
+    isStatic = false;
+
+  for (auto stride : strides) {
----------------
Jianhui-Li wrote:

Just wonder whether we use ShapedType::isStatic() to check both offset/shape/strides? 

https://github.com/llvm/llvm-project/pull/171216