check.mlir

#map = affine_map<(d0, d1) -> (d0, d1)>
#map1 = affine_map<(d0, d1) -> (d1, d0)>
module {
  func.func @matmul_add_transpose(%arg0: tensor<4x4xf32>, %arg1: tensor<4x4xf32>) -> tensor<4x4xf32> {
    %0 = tensor.empty() : tensor<4x4xf32>
    %c2 = arith.constant 2 : index
    %c4 = arith.constant 4 : index
    %c0 = arith.constant 0 : index
    %1 = tensor.empty() : tensor<4x4xf32>
    %2 = scf.for %arg2 = %c0 to %c4 step %c2 iter_args(%arg3 = %1) -> (tensor<4x4xf32>) {
      %4 = scf.for %arg4 = %c0 to %c4 step %c2 iter_args(%arg5 = %arg3) -> (tensor<4x4xf32>) {
        %extracted_slice = tensor.extract_slice %arg0[%arg2, 0] [2, 4] [1, 1] : tensor<4x4xf32> to tensor<2x4xf32>
        %extracted_slice_0 = tensor.extract_slice %arg1[0, %arg4] [4, 2] [1, 1] : tensor<4x4xf32> to tensor<4x2xf32>
        %extracted_slice_1 = tensor.extract_slice %arg5[%arg2, %arg4] [2, 2] [1, 1] : tensor<4x4xf32> to tensor<2x2xf32>
        %5 = linalg.matmul ins(%extracted_slice, %extracted_slice_0 : tensor<2x4xf32>, tensor<4x2xf32>) outs(%extracted_slice_1 : tensor<2x2xf32>) -> tensor<2x2xf32>
        %inserted_slice = tensor.insert_slice %5 into %arg5[%arg2, %arg4] [2, 2] [1, 1] : tensor<2x2xf32> into tensor<4x4xf32>
        scf.yield %inserted_slice : tensor<4x4xf32>
      }
      scf.yield %4 : tensor<4x4xf32>
    }
    %3 = linalg.generic {indexing_maps = [#map, #map1, #map], iterator_types = ["parallel", "parallel"]} ins(%2, %2 : tensor<4x4xf32>, tensor<4x4xf32>) outs(%0 : tensor<4x4xf32>) {
    ^bb0(%in: f32, %in_0: f32, %out: f32):
      %4 = arith.addf %in, %in_0 : f32
      linalg.yield %4 : f32
    } -> tensor<4x4xf32>
    return %3 : tensor<4x4xf32>
  }
}

module attributes {transform.with_named_sequence} {
  transform.named_sequence @__transform_main(%arg1 : !transform.any_op {transform.readonly}) {
    %yield = transform.structured.match ops{["tensor.insert_slice"]} in %arg1
      : (!transform.any_op) -> !transform.any_op
    %a, %b = transform.test.fuse_consumer %yield
      : (!transform.any_op) -> (!transform.any_op, !transform.any_op)
    transform.yield
  }
}