https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88760
Wilco <wdijkstr at arm dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |wdijkstr at arm dot com
--- Comment #27 from Wilco <wdijkstr at arm dot com> ---
(In reply to Segher Boessenkool from comment #26)
> Yeah, and it probably should be a param (that different targets can default
> differently, per CPU probably). On most Power CPUs all loops take a minimum
> number of cycles per iteration (say, three), but that translates to a lot of
> instructions (say, >10).
>
> Small loops should probably be unrolled at -O2 already as well. Maybe fixed
> length loops only?
Agreed, there is no reason not to unroll (or vectorize) with -O2 given several
other compilers do (and beat GCC as a result). It's best to limit unrolling to
small loops and only 2x unless it's 1-2 instructions.
