[Mlir-commits] [mlir] [mlir][vector] Refactor vector linearization patterns (PR #142685)
James Newling
llvmlistbot at llvm.org
Tue Jun 10 13:47:25 PDT 2025
newling wrote:
> The diff is a bit tricky to review - is this is merely moving code around?
Yes, there is no change to the patterns, type converter, or conversion target. The main change is in how the user is expected to initialize them.
> > a user can also bypass this API and mix-and-match whichever patterns they want
>
> IIUC, this would be the main change? How can a user achieve this? Could you point me in the right direction? Thanks!
Yes, that's the main change. It's shown in a test in the end goal draft PR l https://github.com/llvm/llvm-project/pull/142672. This current PR was pulled out of that larger PR.
I now see now that I've 'stacked' the bigger change in a way that makes it harder than necessary to make sense of. I will put this PR into draft mode while I reassess, hope to be back soon with something easier to digest! Thanks for your comments @banach-space and @dcaballe, I'll let you know when I'm ready for feedback again.
https://github.com/llvm/llvm-project/pull/142685
More information about the Mlir-commits
mailing list