[Bug tree-optimization/93721] swapping adjacent scalars could be more efficient

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93721

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Store merging and bswap should be merged - there's PRs for bswap not working
for "stores" (because those are not seeds it works from).  And bswap would
need to be enhanced to detect more permutation patterns (eventually also use
vector load + permute + store/pun to GPR).