align buffer at 64-byte boundary so that memcpy can take advantage of…

[qiqian]

memcpy can take advantage of erms or avx instructions on moden cpus
these instructions run much faster (up to 20% in my test case) on 64-byte aligned address

[Liryna]

Hi @qiqian ,

This is a great change but unfortunately it will break the library - kernel API which means releasing a major version.
If that's fine for you, I will keep this change and merge it when we plan a new major version!

[qiqian]

Hi, it's ok for me, I have made a local workaround to speedup memcpy, just wish it to be merged in the master branch, so I don't have to keep maintaining a custom branch :)