On 30/06/16 18:38, Anton Khirnov wrote:
> Nothing obviously wrong here, but the benchmarks show a significant
> slowdown which whould be investigated.

This code uses loads of read_bit() in branches and that mixes quite
badly, having some local contexts helped a bit depending on the compiler
but what would lead the best result is prefetching the bits
and then do unchecked reading. I'll play with that today if I find some
time.

lu

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to