Closed sezero closed 2 years ago
Uses
bsr
andbsf
inline asm in watcom aux pragma syntax.Question: these optimizations are not enabled in MSVC 32 bit builds: is there a reason?
I believe that every time I've tested it, the optimized version is slower. Haven't done that in some time though, and maybe not since I switched to MSVC 2019.
I believe that every time I've tested it, the optimized version is slower.
Huh.. I see.
I believe that every time I've tested it, the optimized version is slower.
Huh.. I see.
Interesting though, mingw and unix builds have them enabled by default provided that the builtins are available. Only MSVC x86 builds are the outliers.
Anyways, will this patch go into the upcoming release or you intend to delay it for later?
I believe that every time I've tested it, the optimized version is slower.
Huh.. I see.
Interesting though, mingw and unix builds have them enabled by default provided that the builtins are available. Only MSVC x86 builds are the outliers.
Yes, that sounds familiar. Only MSVC was slower using the builtins. But I may test again now that I'm using a newer VS and am on a newer CPU. Of course, I could test on a wide range of CPUs, but the difference was not very big in any case.
Anyways, will this patch go into the upcoming release or you intend to delay it for later?
I'll pull it in right now.
Uses
bsr
andbsf
inline asm in watcom aux pragma syntax.Question: these optimizations are not enabled in MSVC 32 bit builds: is there a reason?