[Mlir-commits] [mlir] [mlir][x86] Add extra patterns for AMX nano-kernel lowering (acc with zero constant) (PR #222513)

[Arun Thangamani]

================
@@ -1418,22 +1446,20 @@ struct VectorContractToAMXDotProduct
     Value srcBuffAcc;
     SmallVector<Value> indicesAcc;
 
-    llvm::TypeSwitch<Operation *>(accReadOp).Case<TransferReadOp, LoadOp>(
-        [&](auto readOp) {
-          srcBuffAcc = readOp.getOperand(0);
+    llvm::TypeSwitch<Operation *>(accWrite)
+        .Case<vector::TransferWriteOp, vector::StoreOp>([&](auto writeOp) {
+          srcBuffAcc = writeOp->getOperand(1);
----------------
arun-thmn wrote:

True. Add a validation at update `InsertionPoint` for this case at: https://github.com/arun-thmn/llvm-project/blob/371f4364574158b0e5eb15788b6632f636e58c08/mlir/lib/Dialect/X86/Transforms/VectorContractToAMXDotProduct.cpp#L1481

https://github.com/llvm/llvm-project/pull/222513