[Bug target/100866] PPC: Inefficient code for vec_revb(vector unsigned short) < P9

[wschmidt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100866

--- Comment #10 from Bill Schmidt <wschmidt at gcc dot gnu.org> ---
Right, it would be a good optimization.  We've stopped focusing much on P8
optimization work at this point simply because of lack of resources.

The needed transform is to recognize load-xxlnor-vperm as a group and combine
into invload-vperm.  But this requires the loaded constant not be used
elsewhere (unlikely, but possible), or if it is, that all such uses are also
xxlnor-vperm, so dataflow analysis for reached uses is required.  Not
completely trivial.

Because it's a P8-only optimization, it's bit lower on the priority list.