[llvm] [X86] Lower vector integer division and remainder through float division (PR #205263)

[via llvm-commits]

Andarwinux wrote:

> As you said we only have a few days and this code doesn't look ready to me, I'll try to make a deep review soon but we shouldn't rush this to meet 23.x.

ping?

> a lot of this looks like it should be done generically in DAGCombine as well.

If also apply SAE to i32 path under strictfp, then there’s no reason to move it to DAG, right?


https://github.com/llvm/llvm-project/pull/205263