Bugs in patterns under `populateVectorNarrowTypeEmulationPatterns` (1D vs 2D)

[banach-space]

All patterns in populateVectorNarrowTypeEmulationPatterns currently assume a 1-D vector load/store rather than an n-D vector load/store. This assumption is evident in ConvertVectorTransferRead, for example, here:

auto newRead = rewriter.create<vector::TransferReadOp>(
    loc, VectorType::get(numElements, newElementType), adaptor.getSource(),
    getValueOrCreateConstantIndexOp(rewriter, loc, linearizedIndices),
    newPadding);

auto bitCast = rewriter.create<vector::BitCastOp>(
    loc, VectorType::get(numElements * scale, oldElementType), newRead);

Both invocations of VectorType::get() here generate a 1-D vector.

Attempts to use these patterns with more generic cases, such as 2-D vectors, fail. For example, trying to cast the following 2-D case to i32:

func.func @vector_maskedload_2d_i8_negative(
  %idx1: index,
  %idx2: index,
  %num_elems: index,
  %passthru: vector<2x4xi8>) -> vector<2x4xi8> {

    %0 = memref.alloc() : memref<3x4xi8>
    %mask = vector.create_mask %num_elems, %num_elems : vector<2x4xi1>
    %1 = vector.maskedload %0[%idx1, %idx2], %mask, %passthru :
      memref<3x4xi8>, vector<2x4xi1>, vector<2x4xi8> into vector<2x4xi8>
    return %1 : vector<2x4xi8>

}

leads to the following error:

error: 'vector.bitcast' op failed to verify that all of {source, result} have same rank
    %1 = vector.maskedload %0[%idx1, %idx2], %mask, %passthru :
         ^

Here’s the mlir-opt invocation used:

mlir-opt --test-emulate-narrow-int="arith-compute-bitwidth=1 memref-load-bitwidth=32"

For further context, I’ve included a full list of reproductions as tests in this PR:

• [mlir][vector] Restrict narrow-type-emulation patterns #115612.

As a temporary workaround, I suggest restricting these patterns to 1-D vectors.