low-latency router needed

[VirtualLarry]

I have a friend with a 6mbit cable connection, a PS3, and a PC. Currently, he is using a netgear wired router that I purchase for him cheap. Running speedtest.speakeasy.net, he actually benchmarks at 20Mbit/sec, I assume that's Comcast's "power boost" in action. I know he's not actually paying for 20Mbit.

Even though he can pull 20Mbit through the router, he experiences slight lag playing online games via the PS3 over the router. So as a fix, he's taken to removing the router. Recently, he stopped using the router altogether, and by plugging the PC into the cable modem directly, managed to get infected by malware twice already. I helped him with that, and now he says he will use the router all of the time, but I'm wondering if it would be worth it to spend the $100 and get a "gaming" router.

Do those routers really have lower latency and higher speeds than other routers? Or are they just standard routers with some QoS options in the config page? He doesn't use the PC much at all when he's playing the PS3 online, so the problem is NOT bandwidth contention.

It just seems (according to him) that the router in the connection is slowing him down. I would like to allow him to use a router for protection and sharing purposes, WITHOUT the slowdown.

Can someone offer some suggestions? Would it be worth it to go whole hog on a Wireless-N router, with gigabit ports? Neither his PC nor the cablemodem support gigabit, although I think that the PS3 does.

Also, is this common, that adding a router into the mix slows down online game playing? Is there any charts that benchmark the additional latency that a router adds to the connection? (That would be a good benchmark, much better than some.)

 

[thecoolnessrune]

Just get a standard router with QOS and give everyone static IPs. Set the port to the PS3 on high priority, that way any tasks occurring on the PC won't knock his lag. While a router in a technical sense does introduce latency, unless the router itself is buggy in its firmware the actual hardware-induced latency caused by a router is negligible in the grand sense of things especially with the internet itself producing 99% of it.

 

[RadiclDreamer]

Update the firmware, the router should cause much latency at all. 1-2ms at max unless there is a bug in it. Are the cables those you've made or factory created?

 

[VirtualLarry]

premade cables, those shouldn't be the problem. It just seems like perhaps the router has an older, slower CPU or something, and that causes latency with the packets.

Edit: Also, I don't really need/want QoS, it's not a contention issue. The problem is the inherent latency caused by the router itself being part of the connection. 1-2ms is too much.

 

[TheKub]

Originally posted by: VirtualLarry
1-2ms is too much.

No one can feel the difference of 1-2 ms. If they say they can they are delusional.

 

[RadiclDreamer]

Originally posted by: VirtualLarry
premade cables, those shouldn't be the problem. It just seems like perhaps the router has an older, slower CPU or something, and that causes latency with the packets.

Edit: Also, I don't really need/want QoS, it's not a contention issue. The problem is the inherent latency caused by the router itself being part of the connection. 1-2ms is too much.

You cant notice 1-2ms

 

[NickOlsen8390]

Originally posted by: TheKub

Originally posted by: VirtualLarry
1-2ms is too much.

No one can feel the difference of 1-2 ms. If they say they can they are delusional.

+1


a 1-2ms jump local is nothing. Its just water under the bridge as it moves. If you ping 60ms to a server. you wont notice anything at 58ms.
Now, if you were 60ms from you router. That would be a problem.
If your feeling lag its either the computer or the internet, Now granted the internet includes a lot of things. Like the route and the end server.....ect...

 

[JackMDS]

Routers are used to share Internet connection so the nature of the traffic depends on the whole Network activity.

However in case of a single computer connected to a Router. I find that many ?lags" complains are imaginary or if real are temporary glitches over the Internet that has nothing to do with the complainer own system.

It is hard to relate seriously to lag?s claims when they are based on human Psychology/Cognition without a multi-ping traces that measures the real timings comparing With a Router Vs. Without a Router.

In most cases when one do run such a trace you can see that the variation in the traffic of the Internet (passed the Modem out) is much larger than the total timing of the LAN with or without a Router.

The main variable that establishes the stability of the timing from the Modem into the LAN is setting of the Router?s and computer?s TCP/IP MTU.

 

[hanoverphist]

Originally posted by: VirtualLarry
premade cables, those shouldn't be the problem. It just seems like perhaps the router has an older, slower CPU or something, and that causes latency with the packets.

Edit: Also, I don't really need/want QoS, it's not a contention issue. The problem is the inherent latency caused by the router itself being part of the connection. 1-2ms is too much.

i have a wrt54g wireless router hooked to a firewall/ router and then to the cable modem with 2 printers and 2 workstations wired to them. i play games on my laptop over wireless and get pings down around 49ms for the most part. if it gets up towards 100 i call it a bad conn lol. the firmware is still ver. 1.0 on both they are so old. i doubt the routers would really be the issue unless they are defective, configged wrong or the firmware is busted.

 

[JackMDS]

Originally posted by: hanoverphisti doubt the routers would really be the issue unless they are defective, configged wrong or the firmware is busted.

:thumbsup:

 

[VirtualLarry]

Well, as for the "not humanly noticable" part - my friend is a competitive-level gamer, he plays fighting games online like Tekken 5 Dark Resurrection and Mortal Kombat II on his PS3. At least for Tekken DR, there is a visible latency bar that is colored red, yellow, or green. It's really only playable when you have "green bar", and according to him, with the router installed in the connection, he rarely gets "green bar", but without the router, he gets it far more often.

So in this case, we have an actual diagnostic indicator to work with, rather than just perceptual psychology.

 

[VirtualLarry]

Originally posted by: TheKub

Originally posted by: VirtualLarry
1-2ms is too much.

No one can feel the difference of 1-2 ms. If they say they can they are delusional.

Competive gamers can tell the difference between 60fps and 80fps. Certainly they can tell the difference of a few ms as well.

 

[drebo]

So tell him the only option is to spend $400 on a Cisco ASA5505. That'll either shut him up or give you a chance to play with a really neat piece of equipment.

 

[VirtualLarry]

He's fairly poor (yeah, with a PS3, I know), so dropping $100 on a gaming router is a big deal to him, but if he can improve his online gaming it would be worth it.

Trying to look at this objectively, pinging www.google.com gives me an average latency of 25ms, +/-5.
He's on the same cable provider with the same level of service.

If the game engine runs at 60fps, that leaves 16.6ms per frame for updates. Since my ping is higher than that, I can't see how anyone could have a playable online game at 60fps. So assuming that the game engines run at 30fps, that gives us 33.3ms per frame. Depending on the ping to the gaming servers and the distance, that just might be playable. However, a worst-case latency of 30ms, leaves only 3.3ms of headroom, so if the router adds 3.3ms or worse latency, then the game will skip/miss frames.

So I guess one way to test would be to run back-to-back ping commands for a while, and check out the average latency, and test whether or not that latency is objectively (measurably) higher with the router as part of the connection. If it's not, then I can point that out to said friend, and tell him his other friend that INSISTS that he remove the router before he will even be willing to play an online game with my friend is wrong.

If the latency is higher, then I know I have to replace the router with one with less added latency.

 

[JackMDS]

Originally posted by: VirtualLarry

Competive gamers can tell the difference between 60fps and 80fps. Certainly they can tell the difference of a few ms as well.

The variation of the human's nervous system reaction time is about 30 to 50 ms.

So while 50ms variation can be sensed by some people, lower timings are beyond the capacity of the neurons.

While most people can jump few feet up, it does not mean that any one can jump up 100 feet.

 

[NickOlsen8390]

Originally posted by: VirtualLarry

Originally posted by: TheKub

Originally posted by: VirtualLarry
1-2ms is too much.

No one can feel the difference of 1-2 ms. If they say they can they are delusional.

Competive gamers can tell the difference between 60fps and 80fps. Certainly they can tell the difference of a few ms as well.

This is bull.
The human eye can't see anything faster then 60fps.
Now the differnce between 30fps and 60fps, yes.

 

[NickOlsen8390]

Originally posted by: VirtualLarry
He's fairly poor (yeah, with a PS3, I know), so dropping $100 on a gaming router is a big deal to him, but if he can improve his online gaming it would be worth it.

Trying to look at this objectively, pinging www.google.com gives me an average latency of 25ms, +/-5.
He's on the same cable provider with the same level of service.

If the game engine runs at 60fps, that leaves 16.6ms per frame for updates. Since my ping is higher than that, I can't see how anyone could have a playable online game at 60fps. So assuming that the game engines run at 30fps, that gives us 33.3ms per frame. Depending on the ping to the gaming servers and the distance, that just might be playable. However, a worst-case latency of 30ms, leaves only 3.3ms of headroom, so if the router adds 3.3ms or worse latency, then the game will skip/miss frames.

So I guess one way to test would be to run back-to-back ping commands for a while, and check out the average latency, and test whether or not that latency is objectively (measurably) higher with the router as part of the connection. If it's not, then I can point that out to said friend, and tell him his other friend that INSISTS that he remove the router before he will even be willing to play an online game with my friend is wrong.

If the latency is higher, then I know I have to replace the router with one with less added latency.

Its not going to be easy to test. If you could some how test at the same time it wouldn't be that hard. But depending on the ISP the ping could be +/- 10ms or more. Just because of whats going on with them.

And your not going to miss frames because of lag. The game is still going to run. You are going to see all the frames. Its not like a video stream, You are rendering the game in real time. The only thing going in and out is updates to the server about where the player is and what he is doing. And as said before, You can NOT tell the difference between 1-2ms, Anyone that says they can is lying. The human scenes just don't move that fast.

If you want to take it that far.
Move to newyork or atlanta or texas or any of the big hubs.
Buy fiber from every provider you can get your hands on.
BGP accross them so you get the fastest route to anywhere.
And only play on servers that are no more then 5 hops away.

 

[TheKub]

Originally posted by: VirtualLarry
Competive gamers can tell the difference between 60fps and 80fps.

Maybe, but they cannot accurately tell the difference between 60fps and 62fps.

Originally posted by: VirtualLarry
Certainly they can tell the difference of a few ms as well.

Yeah maybe 20-50, not 1-2. Remember at frames per second your are talking about somewhere is the 1/30th-1/60th of a second, and when you are talking about ms you are talking about 1/1000- 2/1000 (1/500 to you bastards) of a second.

Originally posted by: VirtualLarry
there is a visible latency bar that is colored red, yellow, or green. It's really only playable when you have "green bar", and according to him, with the router installed in the connection, he rarely gets "green bar", but without the router, he gets it far more often.

That maybe the case but I feel that your 1-2 ms claim is false. I cannot image that 25ms +/-5 would be regularly green and 25ms, +/-7 would rarely be green.

Is there a central server that all games are played on or are they peer to peer? It could be that the router is adding significant (FAR more than 1-2ms) or that just by chance he was connecting to opponents further away and got the perception that it was solely the router to blame.

 

[RadiclDreamer]

Originally posted by: VirtualLarry

Originally posted by: TheKub

Originally posted by: VirtualLarry
1-2ms is too much.

No one can feel the difference of 1-2 ms. If they say they can they are delusional.

Competive gamers can tell the difference between 60fps and 80fps. Certainly they can tell the difference of a few ms as well.

Bullshit, they can tell the difference between 60 average and 80 average, meaning there are some dips under 60 but thats it

 

[VirtualLarry]

Originally posted by: TheKub
That maybe the case but I feel that your 1-2 ms claim is false. I cannot image that 25ms +/-5 would be regularly green and 25ms, +/-7 would rarely be green.

Is there a central server that all games are played on or are they peer to peer? It could be that the router is adding significant (FAR more than 1-2ms) or that just by chance he was connecting to opponents further away and got the perception that it was solely the router to blame.

I gave you the math on this already. At 30fps, updates have to happen within 33.3ms. If the normal worst-case ping is 30ms, that leaves only 3.3ms of headroom, so in that case yes, 30ms would be "green", and 34ms would not be. A difference of only 4ms. Does the router add 4ms, or even perhaps 2ms, and leaving internet latency to a variability of 2 more ms? That's what I plan to find out.

He regularly plays with someone a few towns over, btw, so the distance isn't a variable in this equation. As far as I know, the games run through a central server (for Tekken 5 DR), because there is a game lobby, etc., and other players waiting can watch the match in progress. As for Mortal Kombat II, I have no idea if it's a direct connection between the two players or if it also runs through a central server.

 

[JackMDS]

The following Screen shot is taken by using the List feature of 3DTrace.

http://www.ezlan.net/network/trace.jpg

The table shows the result of multi ping from my computer through a Buffalo Router to Internet site 96.6.68.195*

The important columns in the table (for this discussion) are the two in the middle. I.e. Min. ms and Max ms.

The numbers there show a typical outcome. As the trace progress on the Internet the variation changes dramatically.

So while the variation on my Router is 1ms, and the connection to the direct ISP's server is 2ms, the variation later on is almost 20ms.

So what practically this means.

One should run a similar trace to his favorite Games IPs and if the first number on the table present a bigger variation than the number later own One should get a new Router.

Otherwise one should be comforted with the fact that teenage is a transient state, and with gradual maturity this issue is a Not so important anymore.


* The target IP number is an Akami server. Akami servers are probably the best server system on the Internet far surpass most the gaming servers.

 

[spidey07]

VituralLarry, Internet latency can vary as much as 100 ms or more. You aren't going to notice anything with a SOHO router unless it is overloaded. Game traffic is very small and you can't control it once it gets on the Internet...that's the whole point.

When gamers get "lag" it's because that's just the way the Internet works.

 

[VirtualLarry]

Originally posted by: spidey07
VituralLarry, Internet latency can vary as much as 100 ms or more. You aren't going to notice anything with a SOHO router unless it is overloaded. Game traffic is very small and you can't control it once it gets on the Internet...that's the whole point.

When gamers get "lag" it's because that's just the way the Internet works.

If internet latency was that bad, then online realtime game playing would be impossible. On a decent broadband ISP connection, the latency should be under 50ms at most, although it this case it's 25ms +/- 5. If my latency was 100ms, I would cancel my ISP subscription. In case you just skimmed and didn't read my math, in this particular case, a 3ms lag from the SOHO router in fact does matter, and causes the game to skip frames, making it far less playable.

This issue is consistant (according to my friend), not random internet lag. Then again, that's hard to objectively prove, but I hope to on Fri. when I go over to his house to run some extended ping tests.

 

[Crusty]

Just stop before you make more of a fool out of yourself VirtualLarry. You don't understand how a game engine works, or how the network code in your specific games work. If 1-2ms changes in latency had such drastic effects on the playability of a game you wouldn't have any online games at all, only local network play.

 

[spidey07]

Originally posted by: VirtualLarry

Originally posted by: spidey07
VituralLarry, Internet latency can vary as much as 100 ms or more. You aren't going to notice anything with a SOHO router unless it is overloaded. Game traffic is very small and you can't control it once it gets on the Internet...that's the whole point.

When gamers get "lag" it's because that's just the way the Internet works.

If internet latency was that bad, then online realtime game playing would be impossible. On a decent broadband ISP connection, the latency should be under 50ms at most, although it this case it's 25ms +/- 5. If my latency was 100ms, I would cancel my ISP subscription. In case you just skimmed and didn't read my math, in this particular case, a 3ms lag from the SOHO router in fact does matter, and causes the game to skip frames, making it far less playable.

This issue is consistant (according to my friend), not random internet lag. Then again, that's hard to objectively prove, but I hope to on Fri. when I go over to his house to run some extended ping tests.

LULZ,

Sure, because ping is a great way to measure latency...NOT.