DCO performance improvements
Posted: Sun Jun 26, 2022 2:17 am
well, I usually don't post in fora but this is really worth saying:
good job OpenVPN team:
with my (old but trusted) M5A99X EVO (R1.0) SB950, 32G Skil, AMD 21.2.0 "AMD FX(tm)-6350 Six-Core Processor"
all running at 3.9G and reduced mem speed for thermal reasons (live in the desert, 2 of these beat the AC) I never got up to 50Mb/s in speed tests.
Was wondering what was wrong, looking where the throttling point is. No visible CPU stress or anything.
I know these consumer grade boards have heavy I/O contention internally between mem-cpu-interface (5Gb wireless with an alfa, not stressed at all).
Did install ovpn-dco. (Kali Linux Rolling 2022, Linux5.16.0-kali7-amd64 on x86_64)
VPN throughput 'measured' with the simple speed.measurementlab.net tripled:
from weak and never reached 50Mb/s download to 170.19 Mb/s.
Everything the same (my VPN server at Linode in Freemont, CA).
Good job!
Thank you!
good job OpenVPN team:
with my (old but trusted) M5A99X EVO (R1.0) SB950, 32G Skil, AMD 21.2.0 "AMD FX(tm)-6350 Six-Core Processor"
all running at 3.9G and reduced mem speed for thermal reasons (live in the desert, 2 of these beat the AC) I never got up to 50Mb/s in speed tests.
Was wondering what was wrong, looking where the throttling point is. No visible CPU stress or anything.
I know these consumer grade boards have heavy I/O contention internally between mem-cpu-interface (5Gb wireless with an alfa, not stressed at all).
Did install ovpn-dco. (Kali Linux Rolling 2022, Linux5.16.0-kali7-amd64 on x86_64)
VPN throughput 'measured' with the simple speed.measurementlab.net tripled:
from weak and never reached 50Mb/s download to 170.19 Mb/s.
Everything the same (my VPN server at Linode in Freemont, CA).
Good job!
Thank you!