https://www.phoronix.com/news/FFmpeg-swscale-Rewrite-Landing This swscale rewrite is less dependent upon compiler auto vectorization and introduced a new x86 SIMD back-end.
Niklas Haas reported that the
single-threaded code was 2.1x faster overall and as much as 40.3x faster.
For the
multi-threaded usage it was
2.6x faster overall and as much as
254x faster!
Also, from that forum: "hand crafted assembler will always be faster than compiler optimized code and no programmer worth his salt that really cares about performance would rely on compiler optimizations."
Welcome
FAQ
Downloads
Wiki
