Comparing the specs for both units, I’m curious why the Slate AX (IPQ6000 1.2GHz Quad-core Processor) had better Wireguard performance, while the Beryl AX (MT7981B Dual-core Processor
@1.3GHz) has better OpenVPN performance.
Both seem process intensive. Just curious if there Wireguard can use the quad core better for UDP traffic but lacks higher speed and maybe OpenVPN using TCP (or UDP?) doesn’t leverage that, or something similar – total speculation here.
Claimed. Those are the
claimed best rates, under ideal conditions. Real world is going to be a vastly different experience.
OVPN is single threaded… which a large reason WG stomps all over it. The case is even made worse against OVPN when over TCP.
I think you might want to jump over to
Van Tech Corner’s YT channel rather than let me blather on. You can thank me later.
I’m not comparing WG against OpenVpn, or care about getting maximum published rates, read again.
One router has clearly better posted rates for WG but doesn’t also have better performance for OpenVpn, instead it’s actually lower.
I would have expected better performance across the board for both, and curious what the difference might be, between the routers with respect to the different processes.
OVPN is single threaded.
OVPN likes more GHz while WireGuard is more into cores.
So email@example.comGHz ist better for OVPN while firstname.lastname@example.orgGHz is better for WG.
… unless you implement
cgroups, that is.
@admon - yes I thought that might be the case, makes sense and thanks for help to confirm.
Also I found this summary, which was helpful, I was searching “beryl” v “slate” before, so it didn’t come up
What are the exact differences? I only find cooperations to other routers.
@bring.fringe18 I’m already familiar with WG, and that’s not the question.
Please move along, the adults are talking.
Uh huh. I predate Windows, boy. Do you even predate Windows XP?
First compiler was in Fortran, but who cares
Please move along, not interested in your attitude
I’ll stick to discussion about routers here, thanks