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

[Aviad Cohen]

AviadCo wrote:

@MaheshRavishankar per the discussion with you I decided to abandon the approach of searching patterns in arith operations as it may be too fragile like you mentioned. Instead I am avoiding the "split" before tiling and handling the `affine.min` accordingly.
I will also investigate `IntegerDivisibilityAnalysis ` &`scf-for-loop-canonicalization`  as those seems intresting for my flow, I appreciate your review!

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