[Mlir-commits] [mlir] [mlir][vector] Remove MatrixMultiplyOp and FlatTransposeOp from Vector dialect (PR #144307)

[Adam Siemieniuk]

================
@@ -70,10 +70,17 @@ void ConvertVectorToLLVMPass::runOnOperation() {
     populateVectorBitCastLoweringPatterns(patterns);
     populateVectorBroadcastLoweringPatterns(patterns);
     populateVectorContractLoweringPatterns(patterns, vectorContractLowering);
+    if (vectorContractLowering == vector::VectorContractLowering::Matmul) {
+      populateVectorContractToMatrixMultiply(patterns);
+    }
     populateVectorMaskOpLoweringPatterns(patterns);
     populateVectorShapeCastLoweringPatterns(patterns);
     populateVectorInterleaveLoweringPatterns(patterns);
     populateVectorTransposeLoweringPatterns(patterns, vectorTransposeLowering);
+    if (vectorTransposeLowering == vector::VectorTransposeLowering::Flat) {
+      populateVectorTransposeToFlatTranspose(patterns);
+    }
+    populateVectorTransposeLoweringPatterns(patterns, vectorTransposeLowering);
----------------
adam-smnk wrote:

Why separate these pattern from their grouped counterparts?

https://github.com/llvm/llvm-project/pull/144307