Accidentally sent this reply privately, so resending Le 16 avr. 2014 13:34, "Niels Möller" <[email protected]> a écrit : > Are you thinking of correctness or performance? For performance, maybe > one can try making get_bits non-inline, profile, and try to find a test > case where it consumes a significant proportion of the total running > time.
In both cases I'm thinking of performance as it is the motivation to change something working. So I'd prefer to see what the best impact it can make is, and how much it can worsen an actually existing scenario. Your changes makes sense overall but if they change neither of the best and worst cases, then they become moot. > > I have no idea for the worst case in your proposal (more frequent reads of > > long elements ? Very high AVC intra bitrates?) > > I'm afraid I don't quite understand the question. I'm asking anyone if they can imagine an existing worst-case to your proposed changes. I know none and am throwing ideas around for other people. Best regards, Christophe _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
