The use of AVX-512 instructions in FFmpeg has accelerated some operations by 94 times.
The developers of the FFmpeg project have announced the implementation of new assembler optimizations, which have accelerated certain operations used in video decoding thanks to the use of the AVX-512 instruction set. The performance gain of the assembler code with AVX-512 instructions was 94, 44, 64, and 4.24 times compared to the baseline implementation written in C without using SIMD instructions. The assembler code with optimizations on […]
