Commit 432341d8 authored by Stephen Neuendorffer's avatar Stephen Neuendorffer
Browse files

[mlir] Handle cases where transfer_read should turn into a scalar load 

The existing vector transforms reduce the dimension of transfer_read
ops.  However, beyond a certain point, the vector op actually has
to be reduced to a scalar load, since we can't load a zero-dimension
vector.  This handles this case.

Note that in the longer term, it may be preferaby to support
zero-dimension vectors.  see
https://llvm.discourse.group/t/should-we-have-0-d-vectors/3097.

Differential Revision: https://reviews.llvm.org/D103432
parent 596e61c3

mlir/lib/Dialect/Vector/VectorTransforms.cpp

+12 −0
Original line number Diff line number Diff line
@@ -2828,6 +2828,18 @@ struct TransferOpReduceRank : public OpRewritePattern<vector::TransferReadOp> { 
    // with broadasting. Otherwise we first want to permute the map.

    if (!newMap.isMinorIdentityWithBroadcasting())

      return failure();



    // TODO: support zero-dimension vectors natively.  See:

    // https://llvm.discourse.group/t/should-we-have-0-d-vectors/3097.

    // In the meantime, lower these to a scalar load when they pop up.

    if (reducedShapeRank == 0) {

      Value newRead = rewriter.create<memref::LoadOp>(

          op.getLoc(), originalVecType.getElementType(), op.source(),

          op.indices());

      rewriter.replaceOpWithNewOp<vector::BroadcastOp>(op, originalVecType,

                                                       newRead);

      return success();

    }

    SmallVector<int64_t> newShape = llvm::to_vector<4>(

        originalVecType.getShape().take_back(reducedShapeRank));

    // Vector rank cannot be zero. Handled by TransferReadToVectorLoadLowering.

mlir/test/Dialect/Vector/vector-transfer-lowering.mlir

+8 −3
Original line number Diff line number Diff line
@@ -228,6 +228,7 @@ func @transfer_broadcasting_complex(%mem : memref<10x20x30x8x8xf32>, %i : index) 
#map3 = affine_map<(d0, d1) -> (d1, d0, 0, 0)>

#map4 = affine_map<(d0, d1) -> (0, d1, 0, d0)>

#map5 = affine_map<(d0, d1, d2, d3) -> (d2, d1, d3, d0)>

#map6 = affine_map<(d0, d1) -> (0)>



// CHECK-DAG: #[[$MAP0:.*]] = affine_map<(d0, d1, d2, d3) -> (d0, d1, 0, 0)>

// CHECK-DAG: #[[$MAP1:.*]] = affine_map<(d0, d1, d2, d3) -> (d1, 0, d3)>
@@ -235,7 +236,7 @@ func @transfer_broadcasting_complex(%mem : memref<10x20x30x8x8xf32>, %i : index) 
// CHECK-LABEL: func @transfer_read_permutations

func @transfer_read_permutations(%arg0 : memref<?x?xf32>, %arg1 : memref<?x?x?x?xf32>)

    -> (vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<7x14x8x16xf32>,

       vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<7x14x8x16xf32>) {

       vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<8xf32>) {

// CHECK-DAG: %[[CF0:.*]] = constant 0.000000e+00 : f32

// CHECK-DAG: %[[C0:.*]] = constant 0 : index

  %cst = constant 0.000000e+00 : f32
@@ -275,9 +276,13 @@ func @transfer_read_permutations(%arg0 : memref<?x?xf32>, %arg1 : memref<?x?x?x? 
// CHECK: vector.transfer_read %{{.*}}[%[[C0]], %[[C0]], %[[C0]], %[[C0]]], %[[CF0]] : memref<?x?x?x?xf32>, vector<16x14x7x8xf32>

// CHECK: vector.transpose %{{.*}}, [2, 1, 3, 0] : vector<16x14x7x8xf32> to vector<7x14x8x16xf32>



  return %0, %1, %2, %3, %4, %5 : vector<7x14x8x16xf32>, vector<7x14x8x16xf32>,

  %6 = vector.transfer_read %arg0[%c0, %c0], %cst {permutation_map = #map6} : memref<?x?xf32>, vector<8xf32>

// CHECK: memref.load %{{.*}}[%[[C0]], %[[C0]]] : memref<?x?xf32>

// CHECK: vector.broadcast %{{.*}} : f32 to vector<8xf32>



  return %0, %1, %2, %3, %4, %5, %6 : vector<7x14x8x16xf32>, vector<7x14x8x16xf32>,

         vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<7x14x8x16xf32>,

         vector<7x14x8x16xf32>

         vector<7x14x8x16xf32>, vector<8xf32>

}



// -----