[x265] [PATCH] asm: assembly code for pixel_sad_12x16

Jason Garrett-Glaser jason at x264.com
Wed Oct 30 20:29:52 CET 2013


Maybe movd+movhps would be better than movd+movq, since you'd be able
to get them into one register in two ops?  I guess it needs testing.

Jason


More information about the x265-devel mailing list