[Mlir-commits] [mlir] [mlir][vector] Add scalable lowering for `transfer_write(transpose)` (PR #101353)

[Andrzej Warzyński]

================
@@ -987,6 +988,185 @@ struct TransferOpConversion : public VectorToSCFPattern<OpTy> {
   }
 };
 
+/// Retrieves the dimensions sizes of a mask. Currently supports CreateMaskOp
+/// and ConstantMaskOp.
+template <typename VscaleConstantBuilder>
+static FailureOr<SmallVector<OpFoldResult>>
+getMaskDimSizes(Value mask, VscaleConstantBuilder &createVscaleMultiple) {
+  if (!mask)
+    return SmallVector<OpFoldResult>{};
+  if (auto createMaskOp = mask.getDefiningOp<vector::CreateMaskOp>()) {
+    return llvm::map_to_vector(createMaskOp.getOperands(), [](Value dimSize) {
+      return OpFoldResult(dimSize);
+    });
+  }
+  if (auto constantMask = mask.getDefiningOp<vector::ConstantMaskOp>()) {
+    int dimIdx = 0;
+    VectorType maskType = constantMask.getVectorType();
+    auto indexType = IndexType::get(mask.getContext());
+    return llvm::map_to_vector(
+        constantMask.getMaskDimSizes(), [&](int64_t dimSize) {
+          // A scalable dim in a constant_mask means vscale x dimSize.
+          if (maskType.getScalableDims()[dimIdx++])
+            return OpFoldResult(createVscaleMultiple(dimSize));
+          return OpFoldResult(IntegerAttr::get(indexType, dimSize));
+        });
+  }
+  return failure();
+}
+
+/// Scalable vector lowering of transfer_write(transpose). This lowering only
+/// supports rank 2 (scalable) vectors, but can be used in conjunction with
+/// `UnrollTransferWriteConversion` to support n-D cases. The unroll conversion
+/// unrolls until the first scalable dimension.
+///
+/// Example:
+///
+/// BEFORE:
+/// ```mlir
+/// %transpose = vector.transpose %vec, [1, 0]
+///    : vector<4x[4]xf32> to vector<[4]x4xf32>
+/// vector.transfer_write %transpose, %dest[%i, %j] {in_bounds = [true, true]}
+///    : vector<[4]x4xf32>,  memref<?x?xf32>
+/// ```
+///
+/// AFTER:
+/// ```mlir
+/// %c1 = arith.constant 1 : index
+/// %c4 = arith.constant 4 : index
+/// %c0 = arith.constant 0 : index
+/// %0 = vector.extract %arg0[0] : vector<[4]xf32> from vector<4x[4]xf32>
+/// %1 = vector.extract %arg0[1] : vector<[4]xf32> from vector<4x[4]xf32>
+/// %2 = vector.extract %arg0[2] : vector<[4]xf32> from vector<4x[4]xf32>
+/// %3 = vector.extract %arg0[3] : vector<[4]xf32> from vector<4x[4]xf32>
+/// %vscale = vector.vscale
+/// %c4_vscale = arith.muli %vscale, %c4 : index
+/// scf.for %idx = %c0 to %c4_vscale step %c1 {
+///   %4 = vector.extract %0[%idx] : f32 from vector<[4]xf32>
+///   %5 = vector.extract %1[%idx] : f32 from vector<[4]xf32>
+///   %6 = vector.extract %2[%idx] : f32 from vector<[4]xf32>
+///   %7 = vector.extract %3[%idx] : f32 from vector<[4]xf32>
+///   %slice_i = affine.apply #map(%idx)[%i]
+///   %slice = vector.from_elements %4, %5, %6, %7 : vector<4xf32>
+///   vector.transfer_write %slice, %arg1[%slice_i, %j] {in_bounds = [true]}
+///     : vector<4xf32>, memref<?x?xf32>
+/// }
+/// ```
+struct ScalableTransposeTransferWriteConversion
+    : VectorToSCFPattern<vector::TransferWriteOp> {
+  using VectorToSCFPattern::VectorToSCFPattern;
+
+  LogicalResult matchAndRewrite(TransferWriteOp writeOp,
+                                PatternRewriter &rewriter) const override {
+    if (isTensorOp(writeOp) && !options.lowerTensors) {
+      return rewriter.notifyMatchFailure(
+          writeOp, "lowering tensor transfers is disabled");
+    }
----------------
banach-space wrote:

Yeah, I'd expect some logic higher up to "drive" this in some sense.

Also, since you are adding an option to `VectorTransferToSCFOptions`, will there be a similar pattern for TransferRead? Otherwise `lowerScalable` will only be used for one pattern - that's not much re-use :( 

https://github.com/llvm/llvm-project/pull/101353