[Mlir-commits] [mlir] [mlir][scf]: Avoid inserting affine.min when tiling dynamic operation sizes if possible (PR #113819)

[Aviad Cohen]

================
@@ -186,11 +186,35 @@ static void checkSafeToTileToForall(TilingInterface op,
   }
 }
 
+/// Returns true if `size` is dynamic multiplication of `stride`.
----------------
AviadCo wrote:

@MaheshRavishankar thanks for the replay!
I tried to take a look at `scf-for-loop-canonicalization` which I was unfamiliar with, as for now I didn't manage to find the issue why it does not manage to remove the `affine.min` in this case. I updated the MR to also handle none `0` offset.

https://github.com/llvm/llvm-project/pull/113819