On Jan 31, 2025, Richard Biener <richard.guent...@gmail.com> wrote: > On Thu, Jan 30, 2025 at 8:49 PM Alexandre Oliva <ol...@adacore.com> wrote:
>> Split -Ofast out of the avx conditional, so that it is passed on the >> same targets that expect the transformation. > It effectively would require -Ofast on all targets (it's vectorizing a > FP reduction), > so I'd say -Ofast should not depend on x86 only. IIUC we prefer not to modify testcases, and making -Ofast apply to other targets would do just that, because the test without -Ofast stands on its own and has been running for a while, just not with a vectorizable FP reduction. That was my reasoning to keep it guarded by a conditional. I agree it would have made sense to make it -Ofast on all targets. -- Alexandre Oliva, happy hacker https://FSFLA.org/blogs/lxo/ Free Software Activist GNU Toolchain Engineer More tolerance and less prejudice are key for inclusion and diversity Excluding neuro-others for not behaving ""normal"" is *not* inclusive
