err.txt

#map = affine_map<(d0, d1, d2, d3) -> (d0, d1, d3)>
#map1 = affine_map<(d0, d1, d2, d3) -> (d0, d2, d3)>
#map2 = affine_map<(d0, d1, d2, d3) -> (d0, d1, d2)>
#map3 = affine_map<(d0, d1, d2) -> (d0, d1, d2)>
module {
  func.func @dont_fuse_when_same_trunc_op_dispatch_0_generic_2x4096x640x640_i8xi8xi8xi32xi32(%arg0: tensor<2x4096x640xi8>, %arg1: tensor<2x640x640xi8>, %arg2: tensor<2x640x640xi8>) -> tensor<2x4096x640xf16> {
    %c0_i32 = arith.constant 0 : i32
    %0 = tensor.empty() : tensor<2x4096x640xf16>
    %1 = tensor.empty() : tensor<2x4096x640xi32>
    %2:2 = scf.forall (%arg3, %arg4, %arg5) = (0, 0, 0) to (2, 4096, 640) step (1, 64, 64) shared_outs(%arg6 = %1, %arg7 = %1) -> (tensor<2x4096x640xi32>, tensor<2x4096x640xi32>) {
      %extracted_slice = tensor.extract_slice %arg0[%arg3, %arg4, 0] [1, 64, 640] [1, 1, 1] : tensor<2x4096x640xi8> to tensor<1x64x640xi8>
      %extracted_slice_0 = tensor.extract_slice %arg2[%arg3, %arg5, 0] [1, 64, 640] [1, 1, 1] : tensor<2x640x640xi8> to tensor<1x64x640xi8>
      %extracted_slice_1 = tensor.extract_slice %arg6[%arg3, %arg4, %arg5] [1, 64, 64] [1, 1, 1] : tensor<2x4096x640xi32> to tensor<1x64x64xi32>
      %4 = linalg.fill ins(%c0_i32 : i32) outs(%extracted_slice_1 : tensor<1x64x64xi32>) -> tensor<1x64x64xi32>
      %extracted_slice_2 = tensor.extract_slice %arg7[%arg3, %arg4, %arg5] [1, 64, 64] [1, 1, 1] : tensor<2x4096x640xi32> to tensor<1x64x64xi32>
      %5 = linalg.fill ins(%c0_i32 : i32) outs(%extracted_slice_2 : tensor<1x64x64xi32>) -> tensor<1x64x64xi32>
      %6:2 = linalg.generic {indexing_maps = [#map, #map1, #map1, #map2, #map2], iterator_types = ["parallel", "parallel", "parallel", "reduction"]} ins(%extracted_slice, %extracted_slice_0, %extracted_slice_0 : tensor<1x64x640xi8>, tensor<1x64x640xi8>, tensor<1x64x640xi8>) outs(%4, %5 : tensor<1x64x64xi32>, tensor<1x64x64xi32>) {
      ^bb0(%in: i8, %in_3: i8, %in_4: i8, %out: i32, %out_5: i32):
        %7 = arith.extsi %in : i8 to i32
        %8 = arith.extsi %in_3 : i8 to i32
        %9 = arith.muli %7, %8 : i32
        %10 = arith.addi %out, %9 : i32
        %11 = arith.extsi %in_4 : i8 to i32
        %12 = arith.muli %7, %11 : i32
        %13 = arith.addi %out_5, %12 : i32
        linalg.yield %10, %13 : i32, i32
      } -> (tensor<1x64x64xi32>, tensor<1x64x64xi32>)
      scf.forall.in_parallel {
        tensor.parallel_insert_slice %6#0 into %arg6[%arg3, %arg4, %arg5] [1, 64, 64] [1, 1, 1] : tensor<1x64x64xi32> into tensor<2x4096x640xi32>
        tensor.parallel_insert_slice %6#1 into %arg7[%arg3, %arg4, %arg5] [1, 64, 64] [1, 1, 1] : tensor<1x64x64xi32> into tensor<2x4096x640xi32>
      }
    }
    %3 = linalg.generic {indexing_maps = [#map3, #map3, #map3], iterator_types = ["parallel", "parallel", "parallel"]} ins(%2#1, %2#0 : tensor<2x4096x640xi32>, tensor<2x4096x640xi32>) outs(%0 : tensor<2x4096x640x
f16>) {
    ^bb0(%in: i32, %in_0: i32, %out: f16):
      %4 = arith.sitofp %in : i32 to f32
      %5 = arith.truncf %4 : f32 to f16
      %6 = arith.sitofp %in_0 : i32 to f32
      %7 = arith.truncf %6 : f32 to f16
      %8 = arith.addf %5, %7 : f16
      linalg.yield %8 : f16
    } -> tensor<2x4096x640xf16>
    return %3 : tensor<2x4096x640xf16>
  }
  module attributes {transform.with_named_sequence} {
    transform.named_sequence @__transform_main(%arg0: !transform.any_op {transform.readonly}) {
      %0 = transform.structured.match ops{["tensor.parallel_insert_slice"]} in %arg0 : (!transform.any_op) -> !transform.any_op
      %1:2 = transform.split_handle %0 : (!transform.any_op) -> (!transform.any_op, !transform.any_op)
      %consumer, %fused_consumer = transform.test.fuse_consumer %1#0 : (!transform.any_op) -> (!transform.any_op, !transform.any_op)
      transform.yield
    }
  }
}