Question One network, two computers, two radically different speeds

Nov 8, 2022
3
0
10
I recently had my home provider update me to 1 Gb speeds, but the actual speeds I am seeing on two different computers is bafflingly huge.

First, my network layout. I work primarily out of a home office. The physical medium (cable) enters the house inside that same office. The modem they provided also resides in that same office, and it has an integrated 4-port switch.

I have two computers connected; one a Lenovo laptop i use primarily for work, and a large desktop (I call the “big” computer) which is for personal stuff. Both sit within three feet of the modem/switch, and are both connected via short CAT8 cables (minimal extra cable). There is nothing else between the switch and either computer - just a cable. Both are Win 10.

In using Speedtest.net the Lenovo routinely exceeds 940 Mbs, so the speed is there and the provider isn’t stiffing me. The big computer however never exceeds about 40% of that speed (around 410). I tried some different speed checks that allow variable data size and the difference is even more dramatic (the Lenovo is 5-10 times faster).

Since they are both so physically close and connected to the same switch this should be the easiest thing in the world to troubleshoot, and here’s what I’ve done so far:

Swapped cables. Swapped ports. Tried all the ports. Bought a brand new, even shorter gold plated CAT8 cable. None of this changed anything, so I thought maybe my NIC was the issue and bought a new Gigabit NIC and put it in an unused slot. Still same thing.

Maybe it’s my security SW, so I rebooted in safe mode with networking. Same thing.

Made sure all drivers were up to date (they were) then pored over the NIC configs and they were identical. I always prefer auto negotiation, but for giggles tried hard settings to 1G. No change.
I checked the mobo BIOS for any “missed” settings. Nothing i can see there.

At this point I am running out of ideas and am wondering if there isn’t something buried in the Win 10 config that’s capping the speed on the big PC, but the burning question i have is are these tests reliable? Could there be anything in either PC that’s producing misleading numbers?
If not, what else is there to even check?
 
Although it is not likely your problem almost every so called cat8 is a fake. These are only really used in data centers where you need more than 40gbit and they are very expensive. The cables you see sold to end consumers do not meet the specifications to be certified cat8. Many times they do not even meet the specifications to be any kind of ethernet cables.

You need nothing better than cat5e to run gigabit connection. Buying a higher number gives you nothing other than less money in your pocket. Key is more that it the cable meets the standards which most cables sold say on amazon do not.

Some very common things that cause this. Make sure you have disabled IPv6 in the nic. Next verify you do not have any of the so called "gamer" network bloatware that comes with many motherboard or video cards. This is software that claims it can favor one kind of traffic over the other or reduce latency. It has many names but a very common one is CFOSspeed. You want to uninstall anything like that.
 
Nov 8, 2022
3
0
10
Although it is not likely your problem almost every so called cat8 is a fake. These are only really used in data centers where you need more than 40gbit and they are very expensive. The cables you see sold to end consumers do not meet the specifications to be certified cat8. Many times they do not even meet the specifications to be any kind of ethernet cables.

You need nothing better than cat5e to run gigabit connection. Buying a higher number gives you nothing other than less money in your pocket. Key is more that it the cable meets the standards which most cables sold say on amazon do not.

Some very common things that cause this. Make sure you have disabled IPv6 in the nic. Next verify you do not have any of the so called "gamer" network bloatware that comes with many motherboard or video cards. This is software that claims it can favor one kind of traffic over the other or reduce latency. It has many names but a very common one is CFOSspeed. You want to uninstall anything like that.


Thanks, and yeah, I get that. Deception in marketing is as old as the hills, so I don't doubt anything you say, but the prices are insignificant at the lengths I am using (3 ft). Literally the most expensive cable I could find at that length was $8, so if moving to CAT5 had saved me $3 it's meaningless to me.

At any rate, I did try disabling IPV6 and got 415 on the first test, but that's a small enough change I don't think it's any real improvement.

The gamer bloatware is easy: I built the PC and didn't install any, but even if I had overlooked something I would think the clean (safe) boot would have shown that to be the case, and there was no change there either.

Could something else in the config or Win10 be a culprit?
 
Could be pretty much any software on the machine.

Since you have 2 machines I would start with a old line mode program called IPERF. This should be able to get 900+mbps in both directions. Since it very small program that does not use disk it is not affected by much in the machine. It pretty much will show if the port and drivers etc are good.

If this works I would then try to copy large files between your 2 machines. Hard to say what a good number here is since the file systems on the 2 machine can have a impact.

If these work you start to suspect some web browser issue.
 
Nov 8, 2022
3
0
10
Could be pretty much any software on the machine.

Since you have 2 machines I would start with a old line mode program called IPERF. This should be able to get 900+mbps in both directions. Since it very small program that does not use disk it is not affected by much in the machine. It pretty much will show if the port and drivers etc are good.

If this works I would then try to copy large files between your 2 machines. Hard to say what a good number here is since the file systems on the 2 machine can have a impact.

If these work you start to suspect some web browser issue.


Won't a clean (safe) boot remove that as an equation?

I found testfiledownload.com to try to eliminate any web browser stuff but the difference is even worse: 25 - 27 Mb/s on the laptop, 570 - 580 Kb/s on the big PC using 512 Mb files.

I am unfamiliar with iPerf, but can have a look at it. Does it test files or sites, or does it monitor download speeds?
 
IPERF is a very old line mode command. You start it on one machine with the server option. You then on the other machines use the same command with the IP of your server machine. There are a bunch of other options related to the packet length etc but just run it default.
There are fancy gui but that is a lot of work to get installed for 1 test. Key is it does not mess with file or web sites or anything other than a simple ip address. It pretty much isolates and tests only the network and nothing else.

There are sites on the internet you can run IPERF against but I have only used it to test machine to machine in the house.

Hard to say SAFE mode still load massive amounts of drivers from microsoft.

If you want to eliminate windows you can get a linux boot image that runs from a USB stick. They generally have a simple browser preinstalled and it runs entirely from the USB stick. This also means you have to be careful what you test, there really is no place to download actual files and it is going to be kind slow being a USB stick.
Speedtest should run fine