L'octidi 8 brumaire, an CCXXIV, Rodger Combs a écrit : > Take a look at some results; CBC is significantly slower than CTR in both > libcrypto's and our AES code. There's some hardware limitation it hits.
Yes, that is to be expected. But I would expect something like this: time(AES-X-CBC) - time(AES-X-ECB) = time(AES-Y-CBC) - time(AES-Y-ECB) for X and Y in { 128, 192, 256 }. In other words, I expect the extra time added by CBC to be independent from the key size. But of course, having a benchmark to check that it is true is good. So patch still ok, thanks. Regards, -- Nicolas George
signature.asc
Description: Digital signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel