Are there improvements in this area that would allow us to keep the asm optimizations?
daemoohn opened this issue · comments
Bogdan Ungureanu commented
Brooke Vibber commented
Not the existing asm which is assembler code for specific CPUs, no.
In theory one could write some SIMD optimizations and plug that into the asm config flag as I have done for AV1 decoding, but theora is already pretty fast.