Hi, 2016-04-27 14:52 GMT+02:00 Alexandra Hájková <[email protected]>: > HEVC - no speed loss > H264 - no speed loss > OPUS - 0.7 % speed up > AAC - 2.6 % speed up > DCA - 0.5 % speed up
I wouldn't have expected aac to have such a high improvement but that certainly looks nice. What were/was the sequence(s) and/or aac versions (lc/main/he/...) dca has a fairly high bitrate for the number of decoded samples, so it should have been a case where it mattered more, although not on par with those suggested by Hendrik. This reminds me of another interesting point, although not vital: generated object code size. I don't know if the patchset has that added benefit, but these are interesting numbers to me. btw, some of those codecs may or may not have used the old "unsafe" bitstream reader, ie where bitstream bound checking is left to higher level checks. I haven't checked individual patches, but is the same "safeness" used when comparing before/after in your tests? -- Christophe _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
