Hello, > > > The mmxext version being slower than the mmx one seems however to be an > > existing issue in the tree, which we should probably deal with. Unless > > of course the test is wrong. > > Its mmx, dealing with it would probably entail just deleting it. Can > leave the ordinary mmx and remove the ext version, or perhaps just > both. > > > I think there is two option for that (in swscale) Remove (at least for unscaled part), the include of template for mmxext (in swscale/x86/rgb2rgb.c) Or disable MMXExt version, only if there is a checkasm test. The previous SIMD and checkasm test (for RGBA shuffle), show same kind of result (mmxext slower than mmx).
Martin _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
