Ahh..new ciphers available with openvpn 2.4+

[go558a83nk 362]

Did I miss an announcement? 

 

The specs page https://airvpn.org/specs/ shows a lot more data ciphers available with openvpn > 2.4.  Some 128 bit ciphers.  That should help some weak CPU get a little better speed.

 

Does anybody know which of the 256bit ciphers is likely to be fastest with an AES-NI CPU?  Or, any way to tell which of the ciphers my CPU accelerates?

[serenacat 83]

I started thinking that where there is the same algorithm in  VeraCrypt, a quick and dirty would be to compare elapsed times and cpu/s load to en/decrypt a 2GB .mkv on a SSD, but a quick look at the VC docs indicates efforts at multicore concurrency and maybe other differences from the OVPN such as compiler code gen.

Next thought is to setup the OpenVPN code on own system(s) with localhost loop or fast ethernet for "real" results, more "professional" if getting paid.

[go558a83nk 362]

I started thinking that where there is the same algorithm in  VeraCrypt, a quick and dirty would be to compare elapsed times and cpu/s load to en/decrypt a 2GB .mkv on a SSD, but a quick look at the VC docs indicates efforts at multicore concurrency and maybe other differences from the OVPN such as compiler code gen.

Next thought is to setup the OpenVPN code on own system(s) with localhost loop or fast ethernet for "real" results, more "professional" if getting paid.

 

 

Yep, I used this method.

 

https://airvpn.org/topic/18322-how-to-quicly-test-theoretical-openvpn-throughput/?hl=%2Bopenvpn+%2Bspeed+%2Btheoretical

 

Just change the cipher from aes-256-cbc to whatever you want to test.  For me aes-256-gcm was the fastest of the 256 bit ciphers and seed-cbc.  And aes-128-gcm was no faster.  So, I just stick with 256.