Max191
Max191
This switches the TileAndFuse pipeline to use scf.forall distribution. Using scf.forall distribution also requires some changes to the pass ordering in the TileAndFuse pipeline, which is also handled by this...
This gist https://gist.github.com/Max191/28827ebad403bb21500bc9dc3be37e47 contains 2 func ops with very similar IR, and the only difference being that `%extracted_slice` is used as the thread forall iter_arg instead of the source of...
The `Convert1x1FilterConvToMatmul` pass currently fails when there is a non-unit batch N dimension. In such cases, the transformation is still possible, and the N dimension should be folded into the...
In a recent PR, the TopK e2e test fails in CI: https://github.com/iree-org/iree/actions/runs/11107992173/job/30867743807?pr=18634 The following test is what fails: ``` func.func @topk_2d_dim1_inverted_max() { %input_values = util.unfoldable_constant dense : tensor %input_indices =...
This PR turns on the IGEMM codegen path for convolution by default.
The current reshape propagation patterns upstream bail on multi-result operations, but the implementation seems to support such cases as far as I can tell: https://github.com/llvm/llvm-project/blob/3733528e521b7ee6af3950c65c3ff421c8fd0af6/mlir/lib/Dialect/Linalg/Transforms/ElementwiseOpFusion.cpp#L1253-L1258 This gist is an example...
bump llvm-project to https://github.com/llvm/llvm-project/commit/864902e9b4d8bc6d3f0852d5c475e3dc97dd8335
This PR adds a wrapper pass around DecomposePackUnPackOps, which adds a control function for decomposing only packs and unpacks whose reshapes can be folded with dispatch tensor loads/stores. This PR...
This fixes a bug in the tiling implementation of tensor.unpack that was causing an infinite loop when certain unpack ops get tiled and fused as a producer. The tiled implementation...