Physical/regional servers FINALLY confirmed ??! (Y)

[Mirage]

Depends on individual ISPs too. Some ISPs have better routing agreements to certain areas of the world than others.

I would imagine Europe<->Japan would usually be less important for Europeans than Europe<->America, so it might not be unthinkable that an ISP would try to save some money by not going for the absolutely best routing partner they could get between them an Japan.

I really doubt Australians normally go via Europe to get to Japan though, unless there was a critical network error somewhere in south Asia and you had to re-route traffic. You can however check this with the tracert command.

[Antipika]

The standard for EU to JP is around 270~300ms. As Mirage is saying it all depends on routing agreements. The thing is, there isn't much traffic going from EU to JP coming from the average end-user. And ISP need to prioritize routes that are widely used by their customers.

EU>EU is very important (especially country>country and EU>London). EU>NA is important. EU>Japan? Not really.

I had several ISP in the past few years, could it be in France (Paris) or UK (London), and latency to JP was never lower than 270ms. That's always with the fastest type of line available for the average end user (read : ISP that 99% of people are using).

From work, using a very fast corporate line, I have ~230ms London to Tokyo.

[Risae]

I've been pinging the Durandal server for the past 20 minutes or so, while having the game running, photoshop open, and around a dozen or so other programs up. Yet my ping has never gone past 210ms....

This has nothing to do with latency issues.

[Ingolf]

This has nothing to do with latency issues.

Explain please.

[Mirage]

I'm not denying that there is latency issues. However, I'm very curious as to how EU people get above 200ms, and even reaching into the 300ms, when by my personal observations, they should be less than 200ms. I know there are plenty of other factors, but I'm not necessarily running my programs in my own network on high-speed internet; metropolitan area, shared network, blah blah.

You might just be lucky to have an ISP with good routing to japan. My results with a 40/40 fiber connection in northern europe are something like this.
https://dl.dropbox.com/u/1132077/images/durandal.png

I was planning on posting results for regional WoW and/or LoL servers too, just for comparison, but they apparently do not respond to ICMP ping requests. They are low, though.

[Doho]

what does having other (non-bandwidth using) programs like Photoshop open have to do with latency? i'm genuinely curious, i might close some stuff I have running in the background.

[Mirage]

They should not add to network latency, but your total latency is actually a combination of network latency and so-called input lag.

It is common for games to buffer a few frames before displaying them on your screen, because this smoothens out sudden fps drops and thus makes your framerate more stable. However, this also means that the image you see on your screen was actually rendered by your PC 3 (for example) frames ago, and that also means that whenever you react to an action, you react to an action that happened 3 frames ago.

At 60 fps, each frame equals ~16 ms, but if your framerate drops to for example 30 fps, each frame will equal ~32 ms. As you probably understand, 3 frames of input delay at 30 fps is nearly 100 ms extra lag that is added on top of your network lag, while if you had 60 fps, you would only have ~50 ms input lag. This is also why fighting games such as street fighter and tekken always strive to run at 60fps constantly, so that the input lag is always low and always constant.

So in short: non-networking programs can hurt your total lag, but only if they also cause your fps to drop ingame.

As a side note, it is usually possible to reduce the number of pre-rendered frames for your game. I know the nVidia drivers have 3 frames as the default settings, but you can turn this down to 0. At the lowest setting, you would be looking at something like 16-20 ms input lag(*) if your fps is stable at 60fps, but if your computer isn't able to keep up perfectly all the time, you might experience a less stable fps, with a few more sudden fps drops.

*) Even at 0 pre-rendered frames, it is impossible for the game to display the changes you make sooner than 1 frame after the current frame that you're looking at, which means the game will react one frame later, which is 16 ms. Then there is also a slight input lag from when you press a key on your keyboard or mouse until your is able to send this to the game process, but this is usually very low. Of course, these figures assume 0ms reaction time in the human user, which also is completely unrealistic.

[Dragonheart]

as someone said, IF in 2.0 there will be fights like Heigan, Thadius, or almost every WoW hard modes fight (taking wow as example to give a better idea), they will be impossible to do with 300ish latency...

[Dragonheart]

edit = wrong thread.

[Ingolf]

They should not add to network latency, but your total latency is actually a combination of network latency and so-called input lag.

It is common for games to buffer a few frames before displaying them on your screen, because this smoothens out sudden fps drops and thus makes your framerate more stable. However, this also means that the image you see on your screen was actually rendered by your PC 3 (for example) frames ago, and that also means that whenever you react to an action, you react to an action that happened 3 frames ago.

At 60 fps, each frame equals ~16 ms, but if your framerate drops to for example 30 fps, each frame will equal ~32 ms. As you probably understand, 3 frames of input delay at 30 fps is nearly 100 ms extra lag that is added on top of your network lag, while if you had 60 fps, you would only have ~50 ms input lag. This is also why fighting games such as street fighter and tekken always strive to run at 60fps constantly, so that the input lag is always low and always constant.

So in short: non-networking programs can hurt your total lag, but only if they also cause your fps to drop ingame.

As a side note, it is usually possible to reduce the number of pre-rendered frames for your game. I know the nVidia drivers have 3 frames as the default settings, but you can turn this down to 0. At the lowest setting, you would be looking at something like 16-20 ms input lag(*) if your fps is stable at 60fps, but if your computer isn't able to keep up perfectly all the time, you might experience a less stable fps, with a few more sudden fps drops.

*) Even at 0 pre-rendered frames, it is impossible for the game to display the changes you make sooner than 1 frame after the current frame that you're looking at, which means the game will react one frame later, which is 16 ms. Then there is also a slight input lag from when you press a key on your keyboard or mouse until your is able to send this to the game process, but this is usually very low. Of course, these figures assume 0ms reaction time in the human user, which also is completely unrealistic.

I noticed that my response times seem to have been vastly improved since I upgraded my PC and was able to keep a consistent 60fps. And yeah the nVidia drivers do have that functionality, it was like playing a completely new game, it literally halved my response times.