Open enceve opened 8 years ago
The bottleneck is currently poly1305 (about 400-500 MB/s compared to ChaCha20 with 900 MB/s) The poly1305 Write uses slow non-asm code - So an asm verson for write would speed up the AEAD
The performance for the poly1305.Hash is now quite okay - so remove it from milestone 0.4 and revisit later...
The bottleneck is currently poly1305 (about 400-500 MB/s compared to ChaCha20 with 900 MB/s) The poly1305 Write uses slow non-asm code - So an asm verson for write would speed up the AEAD