Messages

Sorry I misunderstood. Is everything connected via a wired cable? If not then I would anticipate that the differing speeds could be due to the wifi signal.

No problem.

Yes the two desktops are connected to the same gigabit switch by Cat6 cables, then to another switch* and then to the main router (one of three in my mesh group), so wifi is not an issue. All cables are at least Cat6.

Incidentally my iPhone SE is reporting 578Mbps on wifi, my wife's desktop 816Mbps and my desktop 253Mbps just now. The iphone is connected to one of the mesh nodes either through a floor or a wall depending on which one it chose.

I'm thinking of getting a wireless PCIE card for the desktop to see how that performs, although I don't see why that should make any difference when I've already tried two different gigabit ethernet ports in the system.

*My mistake - I've just checked and the upstairs switch is connected directly to one of the ports on the master mesh router, as is the downstairs switch.

A speed check on the router today gives 935 down and 111 up.

What should the speed be?
I think you need to implement traffic shaping on your upload bandwidth from the router, to the bandwidth you should be getting.
If BT are rate limiting elsewhere other than your local internet connection then the time delay of the rate limiting causes these low speeds. This happened to me with a 100MB fibre connection where we were getting 90mb down and 1.6mb up or something odd like that. As soon as I put in the traffic shaping, so that it would shape traffic outbound to the 100MB, we then got 100mb up as well as down.

I'd recommend plugging the copper internet line directly into a decent machine to test with using a cable. You wont be able to set the rate limit that way but you can get consistent test results. Then once you're happy, connect to a router than can handle that level of bandwidth and set up your traffic shaping and test using a wired cable into the router.
Then once youre happy with that, you can test with the wifi and get that nailed down in terms of channel number, channel strength, channel width. You probably have 802.11AX (or wifi 6 or wifi 7) to handle that bandwidth on the wifi.

ps. rate limiting is a hard drop of traffic that exceeds the data rate.
traffic shaping is a buffer and release of traffic at the data rate to help avoid the tcp slow start mechanism and gain more usable access to the bandwidth.

The speed at the router is fine - it's a 900-down/110-up FTTP connection with a guarantee of at least 700Mbps down, and that's what I'm getting at the router (935/111) and at one of the desktops (774/110). The issue is the large difference between two very similar desktops a few feet apart connected to the same switch.

Up speeds are all consistent, I find that interesting... and you updated drivers, swapped cables, swapped ports, so you've ruled out all media external to the PC.

Your PC was slower prior to the upgrade and while it enjoyed a bump up in speed, it's still not matching the other PC in like-for-like terms.

I wonder if there is any malware/bloatware/spyware/some other kind of you-don't-want-this-ware running on your PC. What does the task manager list look like and can you see if anything is running your network interface consistently?

I tried one more thing this morning - I removed the extra PCIE etherenet card in case it was causing some Windows-type weirdness, but as expected it made no diffierence.

OK, Malware, etc? Task manager is normal and disk and network activity are 0% or thereabouts  (e.g an occasional 0.1% on System or Firefox) when the machine is idle. There are regular spikes (every 6-10 seconds) of Ethernet activity on both machines, but these are almost all below 100Kbps and I assume are normal polling or something.

Screen grabs of both machines after a Speedtest:

My Internet connection is BT 900Mbps FTTP (I upgraded from 500Mbps yesterday for an extra £1 per month over 24 months).

My Router is a three-node Linksys Mesh system. A speed check on the router today gives 935 down and 111 up.

A speed test on my iPhone SE gives 500-720 down depending on proximity to one of the Mesh nodes.

I have two small format Shuttle XPC Windows 10 desktop PCs. Both are roughly the same vintage: one has the OS on an SSD (mine) and the other a normal HD (my wife's). Both have 8GB RAM. Both are connected via ethernet to an 8-port switch in the upstairs study (serving various pieces of kit), which itself is connected to another 8-port switch downstairs (serving a load of AV kit), which itself which is connected to the main Mesh node and on to the BT modem.

Now the annoying thing is that my wife's PC now gives 720 down and 110 up, whereas mine routinely gives 300 down and 110 up. Ping is around 5ms on both machines. Before the upgrade from 500 to 900 they were giving 500 and 200 respectively, so although both are still about 50% faster than before, mine is still about 40% of my wife's.

Things I've tried:

- Updated ethernet drivers (both)
- Different Ethernet cable
- Reset the network stack and rebooted
- Used a different ethernet port (my machine has an onboard Ethernet port and a PCIE card)
- Direct connection to one of the Mesh nodes (node connected by WiFi)
- Direct connection to one of the Mesh nodes (node connected directly to switch)

None made the slightest bit of difference (to either machine).

It's not the end of the world, but the sort of thing that annoys me greatly.

Any ideas?