LinalgExt ops don't support fusion

[IanWood1]

IREE currently lacks support for fusing LinalgExt operations with either other LinalgExt operations or standard Linalg operations..The current fusion implementation relies on indexing maps to determine which operations can be successfully fused. See the example below:

    %3 = tensor.empty() : tensor<4x1xi32>
    %4 = linalg.generic {indexing_maps = [#map, #map1], iterator_types = ["parallel", "parallel"]} ins(%expanded : tensor<4x1xi64>) outs(%3 : tensor<4x1xi32>) {
    ^bb0(%in: i64, %out: i32):
      %10 = arith.trunci %in : i64 to i32
      linalg.yield %10 : i32
    } -> tensor<4x1xi32>
    %5 = iree_linalg_ext.scatter dimension_map = [0] unique_indices(false) ins(%expanded_0, %4 : tensor<4x1x16x8x128xf32>, tensor<4x1xi32>) outs(%2 : tensor<8192x16x8x128xf32>) {
    ^bb0(%arg5: f32, %arg6: f32):
      iree_linalg_ext.yield %arg5 : f32
    } -> tensor<8192x16x8x128xf32>

This results in the following dispatch formation (note no fusion):

  %3 = tensor.empty() : tensor<4x1xi32>
  %4 = flow.dispatch.region -> (tensor<4x1xi32>) {
    %10 = linalg.generic {indexing_maps = [affine_map<(d0, d1) -> (d0, d1)>, affine_map<(d0, d1) -> (d0, d1)>], iterator_types = ["parallel", "parallel"]} ins(%expanded_0 : tensor<4x1xi64>) outs(%3 : tensor<4x1xi32>) {
    ^bb0(%in: i64, %out: i32):
      %11 = arith.trunci %in : i64 to i32
      linalg.yield %11 : i32
    } -> tensor<4x1xi32>
    flow.return %10 : tensor<4x1xi32>
  }
  %5 = flow.dispatch.region -> (tensor<8192x16x8x128xf32>) {
    %10 = iree_linalg_ext.scatter dimension_map = [0] unique_indices(false) ins(%expanded, %4 : tensor<4x1x16x8x128xf32>, tensor<4x1xi32>) outs(%2 : tensor<8192x16x8x128xf32>) {
    ^bb0(%arg5: f32, %arg6: f32):
      iree_linalg_ext.yield %arg5 : f32
    } -> tensor<8192x16x8x128xf32>
    flow.return %10 : tensor<8192x16x8x128xf32>
  }

---

Immediate solution

Adding functionality to FormDispatchRegions to get indexing maps for specific LinalgExt ops. This would be a quick and easy way to get indexing maps for specificLinalgExt ops. Also, it would lay the groundwork for a long-term solution.

Long term solution

• Implement LinalgInterfaces for LinalgExt ops. Add to TilingInterface
• ...

---

include "mlir/Interfaces/DestinationStyleOpInterface.td"

iree/compiler/src/iree/compiler/Dialect/Flow/Transforms/FormDispatchRegions.cpp

Line 608 in a78cee1

auto producerLinalgOp = dyn_cast<linalg::LinalgOp>(producer);

Linalg TilingInterfaceImpl
LinalgExt TilingInterfaceImpl

[MaheshRavishankar]

Thanks @IanWood1 for capturing this. Just to amend your long term solution. We will probably end up adding this to TilingInterface which already has the notion of the iteration spaces. TBD though.