On 10/16/2015 3:03 AM, Matt Oliver wrote: > Take 3. Updated the patch based on comments so that it correctly supports > msvc, icl and icc
LGTM. How does the assembly ICL generates for the x86_32 version look like? ICC seems to create a branch for the v == 0 check, so a handwritten inline assembly version using cmov for the cases where HAVE_FAST_CMOV is true may be faster. Not for this patch anyway. Just something to check and benchmark later if anyone cares enough about it. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
