VPN Speedtests, 2 out of 3 are bad

[cdikland]

I have tried using ExpressVPN and NordVPN but both services appear to have issues with 2 of my 3 PC setup. Specifically, speedtest conducted on 2 PCs, when connected to VPN, fall short of the speedtests when not connected to the VPN. The test on the 3rd PC Alienware R6) connected to the VPN returns results very close to tests without VPN. The average test results of these PC are:
PC 1 (Studio-XPS) VPN (70-80Mbps), No-VPN (240-300Mbps)
PC 2 (AlienWare R2) VPN (70-80Mbps), No-VPN (240-300Mbps)
PC 3 (AlienWare R6) VPN (200-230Mbps), No-VPN (240-300Mbps)

Note that all 3 pcs have the same speedtest results when not connected to the VPN
Support staff at both services have not been able to solve this issue.

All 3 PCs run Windows 10. All connected to the same router. PC 1 and PC 2 are several years older than PC 3 but have almost nothing running on them. I have even gone as far as to disable as much cpu activity as I can on both these PC. The results are always the same . With VPN, 1/3 download speed of the norm without VPN. Below is the PC setup in question

 

[Ralston18]

Connection protocols?

Reference:

https://www.allthingssecured.com/vpn/faq/increase-internet-speed-using-vpn/

Compare the settings of the two older computers (Alienware R2 - correct) to the Studio-XPS.

 

[cdikland]

Connection protocols?

Reference:

https://www.allthingssecured.com/vpn/faq/increase-internet-speed-using-vpn/

Compare the settings of the two older computers (Alienware R2 - correct) to the Studio-XPS.

Protocol = UDP for both VPN services. With Expressvpn I did try L2TP as well but it made hardly any difference.
I updated my post above to clarify which PC is which. The Alienware R6 is the newest and fastest of the 3. Sorry for the confusion, if any.

As for the settings, the network setting are "identical" as far as I can see, Not sure what other setting I should be looking at.

When I run each Ookla speedtests I always use the same VPN server and Ookla server on all 3 PCs to compare the results.

 

[Ralston18]

Might get "messy"...

Any differences if you can bypass the Netgear switch (model?) and try each PC directly via the router?

 

[cdikland]

Might get "messy"...

Any differences if you can bypass the Netgear switch (model?) and try each PC directly via the router?

Yes I tried that and it made no noticeable difference. However....
I was just about to give up when on a whim I decided I would try one service provider. Third time is a charm. While the two "culprit" PCs still do not get the same speeds as the R6, with this new service provider their download speeds more than doubled. In most cases the speeds exceed 180Mbps which I had originally set as my acceptable minimum.

 

[Alabalcho]

A shot in the dark: Could PC3 have (much) more powerful CPU?

 

[cdikland]

A shot in the dark: Could PC3 have (much) more powerful CPU?

You bet. I thinking that the actual client vpn software runs best on the R6 hence the faster speeds.

 

[failboat]

compare the passmark of single thread on the computers. VPN can only use 1 core. Those older PCs probably don't have AES-NI. The type of encryption can make a large impact. openvpn 256b using AES is one of the slowest. With a very high end single core and AES-NI it tops out @ 600Mbs. If you see a chacha/poly try that over AES. AES without hardware AES is going to be slow.

https://www.wireguard.com/performance/

 

[yangshijing111]

When you use the VPN service, the speed will definitely decrease. Because part of your network is used for data encryption. So if you need encryption, it means that the speed is reduced, and you need to increase the speed, then you can only choose not to encrypt.