[x264-devel] [PATCH 1/4] x264_intra_sad_x3_4x4_armv6

George Stephanos gaf.stephanos at gmail.com
Wed Feb 1 18:20:24 CET 2012


Couldn't merge the third block since it depends on a value calculated in the loop (r6).
Only other way I can think of would be loading the data into seperate GPRs and reusing them.
This would require 2 cycles (push) and 2 cycles (pop) to release 4 cycles(load). Am I thinking
wrong ? Any other way ?


More information about the x264-devel mailing list