[Mlir-commits] [mlir] [mlir][linalg] Introduce transpose semantic to 'linalg.matmul' ops. (PR #104783)

[Renato Golin]

================
@@ -891,3 +975,86 @@ func.func @fill_tensor(%f: f32, %v: vector<2x4xf32>) -> (tensor<f32>, tensor<vec
 
   return %0, %1: tensor<f32>, tensor<vector<2x4xf32>>
 }
+
+// -----
+
+// CHECK: #[[$ATTR_0:.+]] = affine_map<(d0, d1, d2) -> (d2, d0)>
+// CHECK: #[[$ATTR_1:.+]] = affine_map<(d0, d1, d2) -> (d2, d1)>
+// CHECK: #[[$ATTR_2:.+]] = affine_map<(d0, d1, d2) -> (d0, d1)>
+
+// CHECK-LABEL:   func.func @matmul_transpose_a_explicit(
+// CHECK-SAME:                                  %[[VAL_0:.*]]: memref<5x3xf32>,
+// CHECK-SAME:                                  %[[VAL_1:.*]]: memref<5x7xf32>,
+// CHECK-SAME:                                  %[[VAL_2:.*]]: memref<3x7xf32>) {
+
+// CHECK:           linalg.generic
+// CHECK:           arith.mulf
----------------
rengolin wrote:

`ATTR` variables are not used. Perhaps just check they match between `linalg.generic` and `arith.mulf`.

https://github.com/llvm/llvm-project/pull/104783