[Mlir-commits] [mlir] [mlir][linalg] Add an e2e test for linalg.matmul_transpose_a to ArmSME (PR #71644)

[Andrzej Warzyński]

================
@@ -0,0 +1,77 @@
+// RUN: mlir-opt %s \
+// RUN:   -transform-interpreter -test-transform-dialect-erase-schedule \
+// RUN:   -one-shot-bufferize="bufferize-function-boundaries" -canonicalize \
+// RUN:   -enable-arm-streaming="mode=locally enable-za" \
+// RUN:   -convert-vector-to-arm-sme -convert-arm-sme-to-scf \
+// RUN:   -convert-vector-to-scf -cse -arm-sve-legalize-vector-storage \
+// RUN:   -convert-vector-to-llvm=enable-arm-sme \
+// RUN:   -convert-vector-to-llvm=enable-arm-sve \
+// RUN:   -cse -canonicalize -allocate-arm-sme-tiles -test-lower-to-llvm | \
+// RUN: %mcr_aarch64_cmd \
+// RUN:   -e=main -entry-point-result=void \
+// RUN:   -march=aarch64 -mattr="+sve,+sme" \
+// RUN:   -shared-libs=%mlir_runner_utils,%mlir_c_runner_utils | \
+// RUN: FileCheck %s
+
+func.func @matmul_transpose_a(%A : tensor<?x?xf32>, %B : tensor<?x?xf32>, %C : tensor<?x?xf32>) {
+  %res = linalg.matmul_transpose_a ins(%A, %B: tensor<?x?xf32>, tensor<?x?xf32>)
+                                   outs(%C: tensor<?x?xf32>) -> tensor<?x?xf32>
+  %xf = tensor.cast %res : tensor<?x?xf32> to tensor<*xf32>
+  call @printMemrefF32(%xf) : (tensor<*xf32>) -> ()
+  return
+}
+
+func.func @main() {
+  %c0 = arith.constant 0 : i32
+  %c4 = arith.constant 4 : index
----------------
banach-space wrote:

Could you try some other sizes that don't fit into SME virtual tiles? 

How about AXPY? Is that a TODO? (would fine with me).

Also, could you test `i32` as well? I think that it's important that both f32 and i32 work (other types as well, but we can start with floating point vs integer).

https://github.com/llvm/llvm-project/pull/71644