[Bug middle-end/118994] GCC fails to optimize (a >> 1) + (b >> 1) + ((a | b) & 1) to PAVGB/PAVGW (or equivalent instruction)

[john_platts at hotmail dot com via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118994

--- Comment #5 from John Platts <john_platts at hotmail dot com> ---
GCC also fails to optimize (a | b) - ((a ^ b) >> 1) down to a single SSE2
PAVGB/PAVGW, NEON/SVE2 SRHADD/URHADD, AltiVec
vavgsb/vavgsh/vavgsw/vavgub/vavguh/vavguw instruction where supported on the
target, but Clang will optimize (a | b) - ((a ^ b) >> 1) down to
PAVGB/PAVGW/SRHADD/URHADD where available on the target according to a snippet
over at https://godbolt.org/z/Yz8fEW46f.