[Mlir-commits] [mlir] [MLIR][XeGPU] Add lowering from transfer_read/transfer_write to load_gather/store_scatter (PR #152429)

[Adam Siemieniuk]

================
@@ -221,11 +512,26 @@ struct TransferWriteLowering
     if (failed(transferPreconditions(rewriter, writeOp)))
       return failure();
 
+    // TODO:This check needs to be replaced with proper uArch capability check
+    auto chip = xegpu::getChipStr(writeOp);
+    if (chip != "pvc" && chip != "bmg") {
+      // lower to scattered load Op if the target HW doesn't have 2d block load
+      // support
----------------
adam-smnk wrote:

nit: `store` instead of `load`

https://github.com/llvm/llvm-project/pull/152429