[llvm] [VectorUtils] Trivially vectorize ldexp, [l]lround (PR #145545)

Ramkumar Ramachandra via llvm-commits llvm-commits at lists.llvm.org
Wed Jun 25 07:49:12 PDT 2025


artagnon wrote:

> > That should be fine, assuming the cost model returns reasonable costs for them. Would be good to check if we have tests for their cost (possibly just as cost-model tests)
> 
> Thanks, I've added CostModel tests, although the scalable versions have invalid costs?

Scalable-vector versions cannot be lowered, so the invalid costs are correct. I've fixed the costs of [l]lround in https://github.com/llvm/llvm-project/pull/145714.

https://github.com/llvm/llvm-project/pull/145545


More information about the llvm-commits mailing list