On 4/9/24 16:31, Juergen Christ wrote: > Loop vectorizer can generate vector permutes with constant indexes > where all indexes are equal. Optimize this case to use vector > replicate instead of vector permute. > > gcc/ChangeLog: > > * config/s390/s390.cc (expand_perm_as_replicate): Implement. > (vectorize_vec_perm_const_1): Call new function. > * config/s390/vx-builtins.md (vec_splat<mode>): Change to... > (@vec_splat<mode>): ...this. > > gcc/testsuite/ChangeLog: > > * gcc.target/s390/vector/vec-expand-replicate.c: New test. > > Bootstrapped and regtested on s390x. Ok for trunk?
Does this also work when using the vec_perm intrinsic or would we need to define a matching RTX for that? Ok. Thanks! Andreas
