The row transform is in a macro, as I at first wanted to implement the other transform, until I noticed it was well below 0.5%.
Once optimized, this function is less than 1% of decoding time, so I considered that didn't warrant my figuring a SSE2/... variant. Christophe _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
