Download speed is consistently lower than upload speed on one PC (Lumos Networks fiber Internet)

[sfx2000]

After awhile I took things into my own hands and made my own router and consolidated several devices into a PC instead. Threw a 5ge quad port card init it as that's the max speed my raid setup needed and wne from there.

Check PM - Not wanting to take this thread way off range, but it would be nice to compare notes...

 

[rmiller1959]

Are you sure your coupler is fine?

Nope! That's why I ordered another one that claims to be designed for 10 Gbps connections and cable lengths up to 328 feet (which isn't an issue here). It will be here tomorrow!

 

[rmiller1959]

If you can temporarily disable that service and retest...

I've disabled the service and managed a 200 Mbps speed increase with that adjustment alone. Judging from all the complaints on the Netgear community site about Armor and the performance hits, much more severe than mine, I doubt I'll be re-enabling or renewing it. That said, I'm not ready to turn my PC and laptop security over to Windows Defender, even though it's greatly improved, and it does nothing for my household's Android and iOS smartphones. I'm researching the security suite that will meet all my needs (McAfee is the leader in the clubhouse, although research continues).

 

[drinkingbird]

I've disabled the service and managed a 200 Mbps speed increase with that adjustment alone. Judging from all the complaints on the Netgear community site about Armor and the performance hits, much more severe than mine, I doubt I'll be re-enabling or renewing it. That said, I'm not ready to turn my PC and laptop security over to Windows Defender, even though it's greatly improved, and it does nothing for my household's Android and iOS smartphones. I'm researching the security suite that will meet all my needs (McAfee is the leader in the clubhouse, although research continues).

Not a fan of McAfee. If Bitdefender has been working for you, you can get subscriptions to that relatively cheap, sometimes even in combination with a cheap Microsoft 365 subscription as a package via Newegg (assuming you also use/want that).

 

[rmiller1959]

I am about to try my last troubleshooting steps today. The new coupler, allegedly designed for 10 Gbps speed and cable lengths of up to 328 feet, arrives today. I also ordered a shorter cable to install from the switch to the PC. The current cable is 75 feet in length, which is long, but I was planning to have an electrician friend install it in the wall and didn't know how much slack he would need to do so. Eventually, I couldn't wait for him and ran the cable indoors along the floorboards from the switch to the PC.

The reason for ordering the shorter cable is this disclaimer I found in the fine print on the mini-PC's features page related to the 2.5 Gbps LAN port:

*Minimum Category 5e network cables (within 30m range) required for 2.5Gbps operation according to ASUS testing result.

Thirty meters is 98 feet. I don't know if the switch factors into cable length, but the total run from the PC to the router is more than 98 feet. It shouldn't matter because I'm using Cat 6A cables all the way through, and they're supposed to be rated for 10 Gbps up to 328 feet. Still, I figured a shorter cable wouldn't hurt.

So with all of that, here's where I stand. The upload speed is roughly equivalent to the router speed consistently, but the download speed is 500-530 Mbps slower than the router speed, despite using the same path. Removing Netgear Armor from the router helped, and while I've disabled that feature, I may reset the router to ensure it's cleared out. I learned my PC security software exacts a slight speed toll, even Windows Defender. Once the new cable and coupler are in place, if nothing changes, the PC or the internal wiring may be a limitation.

 

[degrub]

The switch will "reset" the cable length as it generates a new signal. So 30 m + switch + 30 m ...etc.

 

[rmiller1959]

I'm finished with my testing, and I can only conclude that it's a PC or cabling issue (hmmm, where have I heard that before?). I swapped out the coupler and the longer cable with the same result:

Meanwhile, my daughter's PC, which has TWO patch cables coupled together, which are in turn coupled to the internal Ethernet wiring, gets this result:

As you can see, that's essentially the router's speed:

I replaced the coupler, the patch cable from the coupler to the 2nd switch, and the patch cable from the switch to the PC. So it can only be:

• The internal wiring running from the coupler to the initial switch
• The 2nd switch
• The PC

Let me see if I can use my laptop and the 2.5 GBps to USB 3.0 adapter to test the internal Ethernet wiring.

 

[Tech Junky]

it's a PC or cabling issue

Could be the os. Boot form a USB into Linux just to confirm. Since it will load the os into ram it rules out any other components in the system. Also, it unleashes the raw performance of the system. It's a good way to rule out gremlins caused by software.

 

[rmiller1959]

I will give it a shot and see what happens. In the meantime, I connected my laptop to the internal Ethernet cable connected directly to the 1st 2.5 Gbps switch (no coupler) using the 2.5 Gbps to USB adapter. I ran the tests with the following results:

So is it the internal wiring?

 

[drinkingbird]

I am about to try my last troubleshooting steps today. The new coupler, allegedly designed for 10 Gbps speed and cable lengths of up to 328 feet, arrives today. I also ordered a shorter cable to install from the switch to the PC. The current cable is 75 feet in length, which is long, but I was planning to have an electrician friend install it in the wall and didn't know how much slack he would need to do so. Eventually, I couldn't wait for him and ran the cable indoors along the floorboards from the switch to the PC.

The reason for ordering the shorter cable is this disclaimer I found in the fine print on the mini-PC's features page related to the 2.5 Gbps LAN port:

*Minimum Category 5e network cables (within 30m range) required for 2.5Gbps operation according to ASUS testing result.

Thirty meters is 98 feet. I don't know if the switch factors into cable length, but the total run from the PC to the router is more than 98 feet. It shouldn't matter because I'm using Cat 6A cables all the way through, and they're supposed to be rated for 10 Gbps up to 328 feet. Still, I figured a shorter cable wouldn't hurt.

So with all of that, here's where I stand. The upload speed is roughly equivalent to the router speed consistently, but the download speed is 500-530 Mbps slower than the router speed, despite using the same path. Removing Netgear Armor from the router helped, and while I've disabled that feature, I may reset the router to ensure it's cleared out. I learned my PC security software exacts a slight speed toll, even Windows Defender. Once the new cable and coupler are in place, if nothing changes, the PC or the internal wiring may be a limitation.

Cat 5e should be able to do 2.5 at 100M (~328 feet), technically pushing it but it is spec'd for it. Not sure why they say that, maybe the NIC runs some sort of lower power mode or something.

Cat 6 and above can do it easily at full 100M length.

Of course, not all cables are create equal, there are CAT8 cables out there that probably wouldn't pass 5e tests.

 

[drinkingbird]

I will give it a shot and see what happens. In the meantime, I connected my laptop to the internal Ethernet cable connected directly to the 1st 2.5 Gbps switch (no coupler) using the 2.5 Gbps to USB adapter. I ran the tests with the following results:
View attachment 49230View attachment 49231View attachment 49232

So is it the internal wiring?

Or just coincidence that USB adapter is maxing out. What happens if you plug the laptop with that USB NIC into the switch directly? If it can hit full speed then, it sounds like that cable is probably your culprit. When you ran it make sure you didn't kink it hard anywhere or staple through it (or staple it too hard, etc). Or could just be a bum cable.

 

[Tech Junky]

I will give it a shot and see what happens. In the meantime, I connected my laptop to the internal Ethernet cable connected directly to the 1st 2.5 Gbps switch (no coupler) using the 2.5 Gbps to USB adapter. I ran the tests with the following results:
View attachment 49230View attachment 49231View attachment 49232

So is it the internal wiring?

Connect to the other side of the switch aka remove it and connect to the Ethernet cable instead. Might be a switch issue. Sometimes they advertise one speed but only do less than that. Try another port on the switch as well after testing w/o it inline to the laptop.

ISP <> switch1 <eth> laptop
ISP <> switch1 <eth> switch2 <eth> different port

From the looks of it the speed is consistent no matter what up to the 2nd switch.

 

[drinkingbird]

Connect to the other side of the switch aka remove it and connect to the Ethernet cable instead. Might be a switch issue. Sometimes they advertise one speed but only do less than that. Try another port on the switch as well after testing w/o it inline to the laptop.

ISP <> switch1 <eth> laptop
ISP <> switch1 <eth> switch2 <eth> different port

From the looks of it the speed is consistent no matter what up to the 2nd switch.

I think they plugged into the same switch (the main one) as daughter's PC which is getting full 2G. But can't hurt to try it.

 

[rmiller1959]

I have run multiple tests from different parts of the house with different components, and I've practically disassembled and rebuilt my network. I'm no closer to knowing what's going on. Get ready because it's going to get weird.
First, I took the laptop and the 2.5 Gbps to USB adapter to the main switch to test the connection. I tried multiple ports and even the port where my daughter's computer was connected, and where I got a normal reading:

I made sure to get a router speed test while I was testing the direct connection to the switch:

But wait - there's more! So I think there is something wrong with the switch since the download readings are similar to the ones I'm getting when I am directly connected to it without the coupler. I get up and walk over to my wife's desktop and run a Speedtest, with this result:

My wife's PC is connected to the same switch I just tested. I ran a second test with essentially the same result (next message).

 

[rmiller1959]

(Continued)

So I go to my daughter's computer in the lower part of the house and test it, with this result:

Yes, it's connected to the same switch. I have a CalDigit TS4 docking station with a 2.5 Gb LAN port, and it is plugged into a small switch connected to the main switch. I plugged in my laptop via the Thunderbolt 4 port, and this is the reading I got - the uploads are noticeably lower, but the downloads are at speed:

Finally, since I plugged the cable that connects the mini-PC into a different port, the one where my daughter's computer was connected, and got good results, I thought I'd see if it made a difference. Nope.



I don't know where to begin. There's nothing about these readings that make sense to me. Why would the two desktop PCs have the correct readings when a direct connection to the switch via the adapter and laptop had results similar to the mini-PC? And why is the upload correct but not the download? Last reading from the mini-PC, then I'm going to bed.

 

[drinkingbird]

(Continued)

So I go to my daughter's computer in the lower part of the house and test it, with this result:

View attachment 49252
Yes, it's connected to the same switch. I have a CalDigit TS4 docking station with a 2.5 Gb LAN port, and it is plugged into a small switch connected to the main switch. I plugged in my laptop via the Thunderbolt 4 port, and this is the reading I got - the uploads are noticeably lower, but the downloads are at speed:

View attachment 49254

Finally, since I plugged the cable that connects the mini-PC into a different port, the one where my daughter's computer was connected, and got good results, I thought I'd see if it made a difference. Nope.



I don't know where to begin. There's nothing about these readings that make sense to me. Why would the two desktop PCs have the correct readings when a direct connection to the switch via the adapter and laptop had results similar to the mini-PC? And why is the upload correct but not the download? Last reading from the mini-PC, then I'm going to bed.

View attachment 49259

You're just finding various limitations of hardware at this point. The USB adapter or your laptop may not be capable of more than that. Did you ensure your antivirus and firewall are disabled during the test? The USB adapter probably doesn't have good hardware acceleration. The only thing you could try at this point is connect the daughters PC to the cable from your mini PC. Either with a long Ethernet (not to exceed 100M total) or by carrying it in there. If it gets 2G up and down, your network is fine, some of your hardware can't fully saturate it, but that's ok, that just means a bit of "inherent QOS" in your network. Except the daughter who can wreak havoc, but don't they always?

 

[rmiller1959]

You're just finding various limitations of hardware at this point. The USB adapter or your laptop may not be capable of more than that. Did you ensure your antivirus and firewall are disabled during the test? The USB adapter probably doesn't have good hardware acceleration. The only thing you could try at this point is connect the daughters PC to the cable from your mini PC. Either with a long Ethernet (not to exceed 100M total) or by carrying it in there. If it gets 2G up and down, your network is fine, some of your hardware can't fully saturate it, but that's ok, that just means a bit of "inherent QOS" in your network. Except the daughter who can wreak havoc, but don't they always?

I read your suggestion just before turning in, and I couldn't wait to try it. Since I've been removing and replacing cables, I just so happened to have a spare 75-foot Cat 6A Ethernet cable not earning its keep! I ran the cable from my daughter's PC to the internal cable supplying connectivity to my mini-PC, and here are the results (I conducted three tests just to be sure):



By the way, I had one of the old couplers in the mix; I was lazy and left her patch cable with the coupler installed and attached the long cable to the coupler. So after all that, my mini-PC doesn't have the "potatoes," as a football coach would say, to make the lift. Well, it was an excellent exercise to go through. I have a better understanding of my network; I shortened my cable length by 29 feet; I have a shiny new coupler "rated" (by the marketers, anyway!) for Multi-Gig performance; and I added a 2.5 Gbps to USB adapter for my diagnostic toolkit. And my daughter can download and upload video content to her heart's content!

I will ask ASUS if they can explain the problem and offer a solution, but I won't be annoyed. I will live with it until I decide to upgrade again, and I'll consider the form factor and the NIC when making another purchase. I noted all but two of the NICs were Realtek 2.5Gb adapters, even the one in the USB adapter. The mini-PC uses the Intel I225V, and so does the CalDigit TS4 docking station, and it's probably not a coincidence that they were the ones with the "out of spec" ratings, with the ASUS coming up about 500-530 Mbps short on download speed and the CalDigit coming up short by about the same on upload speed.

Thanks to everyone for joining me on this journey. I don't take your time for granted, and I appreciate it!

 

[Tech Junky]

I225V

There's a thread on another forum regarding this being an issue. Using the USB adapter though should have shown this to be the issue though.

The only other test I would do is plug it into one of the in spec devices and prove it's not the USB that's the issue. It would seem that the dock and mini PC have a bottleneck causing the slightly lower speed results.

None of the deficiencies would even be apparent if you didn't have 2ge service or test internally on the lan with iperf. Now, where the deficiencies are in the two devices is the next question that begs to ask if it's worth the time to pin down or just let it exist until your next upgrade of hardware.

 

[drinkingbird]

I read your suggestion just before turning in, and I couldn't wait to try it. Since I've been removing and replacing cables, I just so happened to have a spare 75-foot Cat 6A Ethernet cable not earning its keep! I ran the cable from my daughter's PC to the internal cable supplying connectivity to my mini-PC, and here are the results (I conducted three tests just to be sure):

View attachment 49260View attachment 49261View attachment 49262

By the way, I had one of the old couplers in the mix; I was lazy and left her patch cable with the coupler installed and attached the long cable to the coupler. So after all that, my mini-PC doesn't have the "potatoes," as a football coach would say, to make the lift. Well, it was an excellent exercise to go through. I have a better understanding of my network; I shortened my cable length by 29 feet; I have a shiny new coupler "rated" (by the marketers, anyway!) for Multi-Gig performance; and I added a 2.5 Gbps to USB adapter for my diagnostic toolkit. And my daughter can download and upload video content to her heart's content!

I will ask ASUS if they can explain the problem and offer a solution, but I won't be annoyed. I will live with it until I decide to upgrade again, and I'll consider the form factor and the NIC when making another purchase. I noted all but two of the NICs were Realtek 2.5Gb adapters, even the one in the USB adapter. The mini-PC uses the Intel I225V, and so does the CalDigit TS4 docking station, and it's probably not a coincidence that they were the ones with the "out of spec" ratings, with the ASUS coming up about 500-530 Mbps short on download speed and the CalDigit coming up short by about the same on upload speed.

Thanks to everyone for joining me on this journey. I don't take your time for granted, and I appreciate it!

If both onboard and USB NICs had about the same result, my guess is one of your CPU cores is pegged during the test. In windows you have to explicitly tell it to break out the cores and not just show you an average. You can look in the advanced options of the Intel NIC and make sure all the "offloading" is enabled, but in reality it may just be a cheaper NIC that relies on the CPU heavily.

You could fire off two speed tests at the same time on that PC and see if it uses two different cores, and in that case the sum of the two should be near the 2G.

But in reality, you know your internet can do what it is supposed to do, and >1G on all your PCs is excellent speed that you'll probably rarely, if ever, make use of, so just be happy that you have no concerns with bandwidth and everyone can do what they want. If anything, the daughter's PC could end up being the issue if she is able to do a huge upload or download at max speed, slowing everyone else, but not many sites out there that you can get that kind of bandwidth from.

 

[rmiller1959]

I contacted ASUS and they told me to implement the nuclear option - a full reset of Windows. And it worked!

I am not using the latest driver for the Intel I225V, and I'm almost afraid to upgrade it since it's working the way it should. I have no idea what app or setting was causing the problem, but it's not a problem anymore.

If both onboard and USB NICs had about the same result, my guess is one of your CPU cores is pegged during the test. In windows you have to explicitly tell it to break out the cores and not just show you an average. You can look in the advanced options of the Intel NIC and make sure all the "offloading" is enabled, but in reality it may just be a cheaper NIC that relies on the CPU heavily.

You could fire off two speed tests at the same time on that PC and see if it uses two different cores, and in that case the sum of the two should be near the 2G.

But in reality, you know your internet can do what it is supposed to do, and >1G on all your PCs is excellent speed that you'll probably rarely, if ever, make use of, so just be happy that you have no concerns with bandwidth and everyone can do what they want. If anything, the daughter's PC could end up being the issue if she is able to do a huge upload or download at max speed, slowing everyone else, but not many sites out there that you can get that kind of bandwidth from.

You're right that I have more speed than I'll realistically use. Still, it was hard to resist a price difference of $100 less and twice the speed of my previous connection, so I went for the 2G. Now that I'm getting close to 2Gs up and down consistently on my mini-PC, I'm happy! I've just about reinstalled all my apps, although I culled the list of a few I didn't need. After each installation, I checked my speed, and everything still works as intended. My typical download speed is 1906 Mbps, and my upload speed is around 1960 Mbps.

I'm still not upgrading the NIC driver, just in case it was the culprit (I doubt it).