On 7/10/2020 6:54 PM, Henrik Gramner wrote: > On Thu, Jul 9, 2020 at 4:54 PM James Almer <jamr...@gmail.com> wrote: >> @@ -38,7 +38,7 @@ pb_255: times 16 db 255 >> >> SECTION .text >> >> -%macro BLEND_INIT 2-3 >> +%macro BLEND_INIT 2 >> %if ARCH_X86_64 >> cglobal blend_%1, 6, 9, %2, top, top_linesize, bottom, bottom_linesize, >> dst, dst_linesize, width, end, x >> mov widthd, dword widthm > > Wont this change the output? width was previously doubled a few lines > down when called with three args (%0 means "number of arguments"). > > The existing macro is written in a somewhat obfuscated way using the > number of arguments as an implicit boolean flag, instead of just using > a boolean flag directly. > > Maybe it'd be better to change this macro to have 3 fixed arguments > and use %3 instead of %0 == 3 in combination with adding a default 0 > value in the other macros with a variable number of arguments, e.g. > "%macro BLEND_SIMPLE 2-3 0"
You're right, i had missed the %0 checks. I'm surprised the checkasm test didn't break after this, seeing it does seemingly test 16bit. Will send a new version that generates the same output. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
