[Bug target/118019] RISC-V: Performance regression in hottest function of X264

[rdapp at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118019

--- Comment #5 from Robin Dapp <rdapp at gcc dot gnu.org> ---
According to Li Pan's results this is "just" vector strict align again?
We should be vectorizing the first loop, in particular after the SLP-grouping
changes.

I realize it's annoying having to resort to strict alignment by default but
what's the alternative?  According to the "profile"-related discussions we
might leave that behind us but for the time being I'm afraid we're in that
spot.