[Mlir-commits] [mlir] [mlir][vector] Refine Vector to LLVM lowering options (PR #159553)

[Andrzej Warzyński]

================
@@ -1489,8 +1489,8 @@ def ConvertVectorToLLVMPass : Pass<"convert-vector-to-llvm"> {
            VectorContractLoweringAttr.summary, [{::llvm::cl::values(
            clEnumValN(::mlir::vector::VectorContractLowering::Dot, "dot",
             "Progressively lower to finer grained `vector.contract` and dot-products. (default)"),
-           clEnumValN(::mlir::vector::VectorContractLowering::Matmul, "matmul",
-            "Lower to `vector.matrix_multiply`, maps 1-1 to LLVM matrix intrinsics."),
+           clEnumValN(::mlir::vector::VectorContractLowering::LLVM, "llvm",
+            "Lower directly to `llvm.intr.matrix.multiply`."),
----------------
banach-space wrote:

Good point!

Since this is about lowering to "LLVM intrinsics", and the intrinsics are:
* `llvm.intr.matrix.multiply` (i.e. **llvm.intr.<...>**)
* `llvm.intr.matrix.transpose` (i.e. **llvm.intr.<...>**),

let me rename this as `llvmintr`. 

https://github.com/llvm/llvm-project/pull/159553