Question Advice on buying a gaming router ?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
By the way I forgot to tell you.
My router drops all devices from wi-fi all the time.
Is that another sign that the router is broken and causing these issues ?
I though it was my phone at first but it never drops from wi-fi on different wi-fi.
 
Sorry if I'm spamming.
So I found a spare router from my old ISP, I plugged it in just to test how internet acts even tho VOIP doesn't work and stuff.
Not that much of a difference whatsoever, didn't expect that much of a difference tbh.

By the way the moment twitch changes automatic to 360p the ping goes completely stable.
Whenever someone opens a new tab ping goes crazy too, probably taking all bandwidth to load page faster I guess.
Awaiting a friend that might be able to borrow me a gaming router and I will let you know.
 
Last edited:
So I tried bufferfloat before and after I tried the router with SQM.
Before https://www.waveform.com/tools/bufferbloat?test-id=1f29bb12-ede1-4167-9b91-2eae803c281f
After https://www.waveform.com/tools/bufferbloat?test-id=3c61edce-0149-4a7c-9238-8da36c81223e

Somehow I still get spikes when twitch streams open for example, I need to get SQM down to like 30mbps/100mbps for lag spikes to go away.
That is very strange and I'm confused. Do I need to replace the main router/modem too? Could it be causing problems to the other one?
I connected via bridge connection.
 
Last edited:
Hard to say what it is but it is not bufferbloat. The bufferbloat test shows you can run 80mbps down and 8mbps up with no issues. You are running far less than this so your traffic should never be in buffers and even if it was the sqm should fix it.

You are now at the point of guessing what to do. Although this was fixed years ago intel messed up their modem chipset.
They put out firmware that mostly fixed the problem....ie it fixed all the real world issues but if you ran a unrealistic test case you could still find it. It would be highly unlikely the ISP did not update that firmware but maybe there is some other firmware issue with the current modem.

I still suspect you are getting data loss that does not show up until the line is loaded more....but that too is purely a guess.
 
What I noticed is that twitch or heavy pages for example when they launch they take up to even the whole bandwidth for a split second and that is when I get huge spikes happen.

If I use speedtest which takes the whole bandwidth too but doesn't do it instantly I don't have issues with spikes or ping increasing.

Should I try using cake instead?
 
Last edited:
It is not likely that twitch or web pages actually can use 100mbps. You can try cake but I doubt it makes any difference.
The fact the bufferfloat tests are perfect and this still happens makes me believe I'll never solve this tbh.
I dont know what to do anymore.

What's the best way to measure ping? Any game I could try that is known to be stable and has network details in it ?

Could it be that it only happens on the game I play for some reason ? Maybe bad pathing ?
 
Last edited:
You just let a ping command run in a background window. Generally you would run 3 or more. You would leave a ping to your router ip, a ping to the very first ISP router (generally hop 2 in a trace), and then a third to say 1.1.1.1 or maybe the game server if it will respond to a ping.
 
  • Like
Reactions: Widow1
You just let a ping command run in a background window. Generally you would run 3 or more. You would leave a ping to your router ip, a ping to the very first ISP router (generally hop 2 in a trace), and then a third to say 1.1.1.1 or maybe the game server if it will respond to a ping.
I cannot trace or ping the gameserver.
I pinged the other 2 and I do indeed see ping spikes in there too.
From 4 to 10 to 44 even.

Tested the same way I test my game, opened twitch stream running in the background.
It does spike to 14 even with only the forum open tho. Should I call my ISP again ?

I tried all 3, all 3 had min/average of 4 but they had max of 15/48/59
 
Last edited:
If you see any kind of spike to your router then it is likely a issue with your pc. You generally should always see about 1ms with small variations under say 10ms.

Your largest issue is those spikes are way too small to affect a game. You need multiple in a row that are say larger than 100ms.

In addition a lot of that is testing error. Ping is a very basic function it is not something like a GPS that has precision clocks so it can measure exact time it take for signal to go from a satellite to a end station. In almost all cases ping is considered a low priority task so any router you ping will do other functions before it responds to ping. Many commecial routers also have artificial limits on how much total cpu resources it will use for stuff like ping to prevent denial of service attacks with a ping command.

You have to be very careful in how you interpret the results.
 
Could there be something I've done wrong when bridging ?
Is there something important that I need to disable on main modem/router?

Maybe I'm missing something important on edgerouter too.
 
Last edited:
When the router is in bridge mode it basically has nothing you can set after that. Its only function is to convert from the cable coming into your house to ethernet.

Maybe go back with the edgerouter and set it all to default and see if anything changes. If there is something using lots of cpu on the device it can affect traffic.
 
I did a factory reset but nothing changed.

Error: eth0 mtu must be greater than or equal to 1500 (pppoe0 mtu + 8)

Does this mean 1500 on the devices counts as 1492 ?

The max mtu I can use without dropping packets is 1492(1464+28) but everything except pppoe0 is set to 1500.

« SpeedGuide.net TCP Analyzer Results »
Tested on: 2023.07.06 13:59
IP address: 188.4.xx.xx
Client OS/browser: Windows 10 (Chrome 114.0.0.0)

TCP options string: 020405840103030801010402
MSS: 1412
MTU: 1452
TCP Window: 262400 (not multiple of MSS)
RWIN Scaling: 8 bits (2^8=256)
Unscaled RWIN : 1025
Recommended RWINs: 64952, 129904, 259808, 519616, 1039232
BDP limit (200ms): 1050 Mbps (105 Megabytes/s)
BDP limit (500ms): 420 Mbps (42 Megabytes/s)
MTU Discovery: ON
TTL: 114
Timestamps: OFF
SACKs: ON
IP ToS: 00000000 (0)

Is this legit? And if so do I have wrong mtu or does it just shows my MTU-40?
Should I maybe try lower than 1492 even tho it "works" ?

Can PC hardware cause ping spikes ?

I'm legit looking at everything right now I'm sorry I'm desperate T_T
 
Last edited:
That is strange that you get errors if you set it below 1500. It doesn't really matter as you can see in that display you posted MTU discovery is being used. This is part of the tcp setup process and it will figure out what that actual maximum mtu in the path is. This will for example lower the mtu because of overhead like pppoe or maybe even a vpn. It generally works very well except for when a firewall is blocking the messages the MTU discover uses.
Even if you were to get the mtu wrong devices in the network will just fragment the packets and the far end must reassemble them. It not optimum to fragment packets because it take cpu to reassemble them but in general it does not cause performance issues.

You can test the mtu with option on the ping command that sets the packet length and also sets a do not fragment flag.
 
That is strange that you get errors if you set it below 1500. It doesn't really matter as you can see in that display you posted MTU discovery is being used. This is part of the tcp setup process and it will figure out what that actual maximum mtu in the path is. This will for example lower the mtu because of overhead like pppoe or maybe even a vpn. It generally works very well except for when a firewall is blocking the messages the MTU discover uses.
Even if you were to get the mtu wrong devices in the network will just fragment the packets and the far end must reassemble them. It not optimum to fragment packets because it take cpu to reassemble them but in general it does not cause performance issues.

You can test the mtu with option on the ping command that sets the packet length and also sets a do not fragment flag.
I've tested it and the best was 1464. And as far as I know you add 28 to it so 1492 basically.
I'm running out of ideas.

Could it be PC hardware? I even tried netlimiter to limit chrome to 5mbps and I still get huge spikes with twitch tabs, they surpassed the limit somehow.
I dont know, sudden bandwidth drain from apps/tabs is what cause these spikes, steady download even with max speed does not.
 
Last edited:
That would be the best option if you can do it. You might not be able to do that because of the VoIP ports.
The main difference between a modem and a router is that a router processes IP addresses. VoIP means Voice over IP so it uses IP addresses.
I found something interesting today.
Yesterday I setup an access point, devices get like 40mbps of speed.
Watching twitch through my phone/tv does not increase my latency at all, yet watching from pc does.
What are your thoughts on this, I guess it eats way less bandwidth when watching from the apps ?


On the bright side I know that wi-fi wont affect my latency.
ISP told me that line is perfect and only a broken modem/router would be able to create these issues right now.
I'm getting a replacement modem from them soon, hope that solves the rest of the problems.
 
Last edited:
What ethernet adapter do you have on your PC motherboard? Please tell me it's NOT Killer ethernet.

I'm late to the party, but the only 2 QOS traffic shaping algorithms that work well are based on FQ_Codel or CAKE. Make sure you set your limit to about 80-85% of your maximum bandwidth. The reason for this is it takes a second or two to normalize after opening a new connection, so ping spikes are possible unless you give it enough headroom to allow for the new connection to start and the others to slow down.

Also, are you sure you're using your ISP modem/router in bridge mode?
 
What ethernet adapter do you have on your PC motherboard? Please tell me it's NOT Killer ethernet.

I'm late to the party, but the only 2 QOS traffic shaping algorithms that work well are based on FQ_Codel or CAKE. Make sure you set your limit to about 80-85% of your maximum bandwidth. The reason for this is it takes a second or two to normalize after opening a new connection, so ping spikes are possible unless you give it enough headroom to allow for the new connection to start and the others to slow down.

Also, are you sure you're using your ISP modem/router in bridge mode?
Hey! Thanks for the response.

I have a realtek gaming 2.5gbe
The edgerouter x I have has FQ_Codel.

I've tried different maximum bandwidth limits.
Yes I've checked multiple guides to make sure I've done it correctly.
Setting up bridge/disabling firewall/disabling wi-fi.

Like the whole point was to be able to play when people are home, as I was playing fine when I was home alone.
It kinda feels like it's "laggier"/ has a choppy feeling even when I'm home alone now ?
Thinking of resetting the modem/router and trying out without the edgerouter to see if things are better when I'm home alone.

Should hwnat and ipsec both be enabled/disabled ?
 
Last edited:
I removed the edgerouter and reseted my modem/router so I get internet from it when I was home alone as I knew ping would be stable at that time and I had a perfect ping game.
Never happened with the edgerouter,ping spikes even when I'm home alone.
 
Hey! Thanks for the response.

I have a realtek gaming 2.5gbe
The edgerouter x I have has FQ_Codel.

I've tried different maximum bandwidth limits.
Yes I've checked multiple guides to make sure I've done it correctly.
Setting up bridge/disabling firewall/disabling wi-fi.

Like the whole point was to be able to play when people are home, as I was playing fine when I was home alone.
It kinda feels like it's "laggier"/ has a choppy feeling even when I'm home alone now ?
Thinking of resetting the modem/router and trying out without the edgerouter to see if things are better when I'm home alone.

Should hwnat and ipsec both be enabled/disabled ?

hwnat probably won't matter either way at 100mbps. I think for the edgerouter, if you had 400mbps more internet, hardware nat is required to reach gigabit speeds.

You can try disabling IPSEC temporarily to see if that speeds things up.
 
hwnat probably won't matter either way at 100mbps. I think for the edgerouter, if you had 400mbps more internet, hardware nat is required to reach gigabit speeds.

You can try disabling IPSEC temporarily to see if that speeds things up.
Internet has been way more stable past 2 days, most of it might be resolved.
Now I need to address a few input lag issues, working on it with some professionals and I might be able to play lag free finally!
 
Most gaming routers are designed to work alongside your existing setup. Connect the gaming router to your modem router using an Ethernet cable, and it will act as a bridge, taking over the task of managing your network traffic and optimizing it for gaming.

Some ISP-provided routers do offer basic QoS (Quality of Service) features, but they might not be as robust as those found in dedicated gaming routers.

If you're experiencing lag when others are using the internet, investing in a gaming router with advanced QoS settings makes a noticeable difference. It will allow you to allocate more bandwidth to your gaming PC while ensuring other devices don't hog all the resources.

When shopping for a gaming router, look for features like customizable QoS settings, prioritization options, and low latency.

Brands like TP-Link and Netgear offer popular gaming routers that cater to different budgets and needs. Make sure to check out customer reviews and compare the specifications to find the one that suits you best.

Did you write that with ChatGPT? Unless it's based on FQ_Codel or Cake, QOS in most gaming routers are almost worthless.
 
  • Like
Reactions: bill001g