[all-commits] [llvm/llvm-project] be9e84: [mlir] [linalg] fix failure on specializing matmul...

[ziereis via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: be9e84e1a9087941f4da9259fb64bb975029b0cb
      https://github.com/llvm/llvm-project/commit/be9e84e1a9087941f4da9259fb64bb975029b0cb
  Author: ziereis <44057120+ziereis at users.noreply.github.com>
  Date:   2026-03-06 (Fri, 06 Mar 2026)

  Changed paths:
    M mlir/lib/Dialect/Linalg/Transforms/Specialize.cpp
    M mlir/test/Dialect/Linalg/specialize-generic-ops.mlir

  Log Message:
  -----------
  [mlir] [linalg] fix failure on specializing matmul with permuted loops (#184294)

This patch fixes generic specialization when the loop dimensions are
permuted in the generic w.r.t. to canonical iterator order of the named
ops by not forwarding the maps of the original generic and instead
recreating them ensuring they always follow the canonical order.

For example, the generic which is to be specialized to a matmul could
have `[parallel, reduction, parallel]` loops, specializing this as is
and just coping the indexing maps like we do now will lead to a
verification error since the dimension will not match the canonical form
the matmul named op expects

e.g. the maps could be:
```
(m, k, n) -> (m,k)
...
```
So we would have to recreate the maps to be:
```
(m,n,k) -> (m,k)
...
```

Assisted by: Claude Code



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications