Does anyone think/thought about using sse/avx technology in ccminer ? sp_ ?

I know that ccminer spends a little time @CPU comparing to GPU but still there is a room for improvment.