[Mlir-commits] [mlir] [Affine] Parallelize loops with MemoryEffects::Free ops. (PR #172388)

Slava Zakharin llvmlistbot at llvm.org
Fri Dec 19 10:16:43 PST 2025


vzakhari wrote:

> isDeallocationOfLocallyDefined is a non-local check that does not seem straightforward and more importantly, not clearly semantically justified to me at the moment.

Thank you for taking a look!

Can you please explain what you mean by "a non-local check"?

Basically, the idea is that allocating and freeing the same memory on each iteration of the loop should not prevent parallelizing it.  Do you think this statement is wrong or just the suggested implementation is wrong?

https://github.com/llvm/llvm-project/pull/172388


More information about the Mlir-commits mailing list