On Tue, Apr 4, 2017 at 4:15 AM, Kyotaro HORIGUCHI < horiguchi.kyot...@lab.ntt.co.jp> wrote:
> The previous expression intended to avoid decimal arithmetic, but > gcc optimizes the simple division better (using cmovns-add-sar) > than the current INT64_AU32 (jmp-sar) so I changed it. This > doesn't suffer overflow. > > How does this affect non-gcc compilers? Specifically I am interested in the llvm based compilers in FreeBSD. Or is this within a gcc-specific section of the header?