[Mlir-commits] [mlir] [mlir][linalg] Vectorize directly to a named contraction (PR #147296)

[Andrzej Warzyński]

================

----------------
banach-space wrote:

Note, I have recently refactored the vectorisation tests and grouped them as discussed here:
* https://github.com/llvm/llvm-project/issues/141025

Specifically, this is what we have today:
```bash
mlir/test/Dialect/Linalg/vectorization/
├── conv.mlir
├── conv-flatten.mlir
├── conv-with-patterns.mlir
├── extract.mlir
├── extract-with-patterns.mlir
├── insert-slice.mlir
├── insert-slice-with-patterns.mlir
├── linalg-ops.mlir
├── linalg-ops-with-patterns.mlir
├── pad.mlir
├── pad-with-patterns.mlir
├── unsupported.mlir
```

So, `contraction-named.mlir` doesn't quite fit. However, one of these options would:
* `contractione.mlir` (1 file), or
* `contraction-interface.mlir` (1 file), or
* `contract.mlir`  + `matmul.mlir` + `batch_matmul.mlir` ... (multiple files)

Given that the updated logic looks at `ContractionOpInterface`, I vote for `contraction-interface.mlir`. Other names are fine too, I would just add a top-level comment to explain what this file covers (this is the first test file to focus on a specific interface, so it's worth clarifying).

https://github.com/llvm/llvm-project/pull/147296