Latency [Tested] - Page 4
Forum Index > SC2 General |
FortuneSyn
1825 Posts
| ||
Jarvs
Australia639 Posts
On March 06 2010 00:53 member1987 wrote: You can't measure latency like this and its not a realistic representation. The only way to measure latency would be to use a specifically made program for SC2 that is going to measure your in-game latency and record lowest, highest and average latency. Of course you would need to test it on a new installation, with no additional programs installed. This would show the best latency you can get. Real world measurements would have to be made with old OS installation and various programs installed that may automatically connect to the internet for updates, reports and sending/receiving information. Also, since SC2 uses dedicated servers and WC3 uses user host you can't actually measure them, because your latency would be dependent on the host internet speed and location, in accordance with your location. War3 uses dedicated servers. I do agree that the conditions needed to make a totally clean test would require a complete format etc, but I really don't think such perfect information is needed. Blizzard are aware of their netcode and they would know if there is such inbuilt delay. They would also know if there would be a way to reduce this delay if there is one. | ||
thopol
Japan4560 Posts
It's like ruining a superhero movie. I don't understand how they can do it as long as they just make the comic the storyboard, and yet they always want to put a new and shitty spin on it. No need to rework what is awesome. | ||
Floydian
United Kingdom374 Posts
| ||
CowGoMoo
United States428 Posts
I would be curious what the delay on EU servers is, since I have heard from my friends the lag isn't nearly as noticeable there. | ||
NonY
8716 Posts
On March 06 2010 00:53 member1987 wrote: You can't measure latency like this and its not a realistic representation. The only way to measure latency would be to use a specifically made program for SC2 that is going to measure your in-game latency and record lowest, highest and average latency. Of course you would need to test it on a new installation, with no additional programs installed. This would show the best latency you can get. Real world measurements would have to be made with old OS installation and various programs installed that may automatically connect to the internet for updates, reports and sending/receiving information. Also, since SC2 uses dedicated servers and WC3 uses user host you can't actually measure them, because your latency would be dependent on the host internet speed and location, in accordance with your location. But latency isn't automatically as low as possible. There is a set latency. If you can go lower, it doesn't matter. If it's too low, then there is lag. So as long as there is no lag, counting frames between issuing a command and the response to the command is a reliable method for determining what the latency is set at. The problem is that the set latency is much higher than it needs to be. edit: tec27's post below explains better | ||
Nuttyguy
United Kingdom1526 Posts
| ||
Mohdoo
United States15082 Posts
Though I do not doubt the results of Gibybo, it just seems like there is more to it than that. | ||
Paperkat
United Kingdom47 Posts
| ||
Mohdoo
United States15082 Posts
On March 06 2010 03:10 Paperkat wrote: i dont really know, but i dont usually feel any lag (i play on eu) and i gave a beta key to a friend who lives in New zealand to play with me on EU servers and he lags in every other game (other titles) he plays when he wants to play with me, but doesnt lag on starcraft 2 eu servers and i can play against him fine Very interesting. I would venture to guess that it is possible Blizzard accidentally set the latency on the Americas region server too high on accident, perhaps by force of habit when setting up other Battle.net servers. For the latency to be identical to Bnet1.0 is a bit suspicious. I would be more inclined to believe it is them failing to do it well enough if it was say...Higher...Or even a bit lower than Bnet1.0. But the fact that it is the same makes me think it just an accident. Especially considering people from Europe are saying they have almost no lag. I live roughly 900 miles (1 448.4096 kilometers) from the Americas Battle.net server (In Anaheim, California if it is in the same place USWest is), and my delay, although lower than Bnet1.0, is certainly more than HoN. I would imagine that most people playing in Europe aren't this close to their Bnet server, so I think that supports my theory that this is an accident. Can anyone run similar tests on the Euro or even the Asia server? Artosis? You have access to both Asia and Americas server, right? | ||
TheYango
United States47024 Posts
On March 06 2010 01:18 Jarvs wrote: War3 uses dedicated servers. From the Warcraft III Battle.net FAQ: Does Warcraft III have a Client/Server Model? No. Due to the amount of units and the number of players in a game of Warcraft III, it is not possible to implement a straight Client/Server architecture as in Diablo II. We are using a variation of the peer-to-peer model that allows us to eliminate some of the abuses found in StarCraft games. | ||
tec27
United States3673 Posts
On March 06 2010 00:53 member1987 wrote: You can't measure latency like this and its not a realistic representation. The only way to measure latency would be to use a specifically made program for SC2 that is going to measure your in-game latency and record lowest, highest and average latency. Of course you would need to test it on a new installation, with no additional programs installed. This would show the best latency you can get. Real world measurements would have to be made with old OS installation and various programs installed that may automatically connect to the internet for updates, reports and sending/receiving information. Also, since SC2 uses dedicated servers and WC3 uses user host you can't actually measure them, because your latency would be dependent on the host internet speed and location, in accordance with your location. You can in fact measure latency this way and get a realistic representation, so long as the game is not lagging because the latency is set too low. In the case that we are attempting to measure, latency is not ping time (which is what you seem to think), it is essentially the delay between sending update packets. We can look at BW to see how this works. Latency settings in BW is basically a table of numbers specifying how long to buffer packets for. Then the client basically sees if CurrentTick % LatencySetting = 0, and if so, it sends all the commands it has buffered. At normal Battle.net settings, this is about 5 ticks, or 210ms on fastest. This means that any command you issue on the first tick does not get sent until the 5th. If your client was to execute commands immediately, but buffer commands sent across the network, desyncs would occur. For instance, if you were to order and SCV to move and you client did it immediately, it would have 4 extra ticks to move before the same SCV moved on your opponents screen. This is obviously undesirable. Therefore, not only does the client buffer commands sent across the network, it also buffers commands you perform locally for the same period of time. This is why you can measure latency settings by looking at how long it takes for an action you perform to occur. Now that we understand that, another important question: why does lag and slowdown occur? Lag and slowdown in BW is a result of packets not arriving when they should. Essentially, if your packets to the other person take longer to arrive than the delay between each packets' send, you will experience slowdown as the game freezes to wait for the packets that are missing. So in even simpler terms, if LatencySetting > Ping, you'll get slowdown. This is an important point, because it means that it should be possible to get a decent latency setting automatically based on ping between two players. And it should also be possible to make automatic adjustments to this latency based on the number of times the game has to slow down. Why Blizzard does not try to get the minimum amount of latency is beyond me, there are definitely ways of doing it that have little negative effect on the userbase. | ||
EtherealDeath
United States8366 Posts
| ||
Myrmidon
United States9452 Posts
Ping time is maybe misleading because that's generally used to represent round-trip time. You just need the input latency to be set where all the packets from one destination gets to the other under that threshold. Assuming the forward and reverse links are symmetrical or close to it, this one-way transmission propagation delay is about half the ping time. | ||
FortuneSyn
1825 Posts
On March 06 2010 03:16 Mohdoo wrote: Very interesting. I would venture to guess that it is possible Blizzard accidentally set the latency on the Americas region server too high on accident, perhaps by force of habit when setting up other Battle.net servers. For the latency to be identical to Bnet1.0 is a bit suspicious. I would be more inclined to believe it is them failing to do it well enough if it was say...Higher...Or even a bit lower than Bnet1.0. But the fact that it is the same makes me think it just an accident. Especially considering people from Europe are saying they have almost no lag. I live roughly 900 miles (1 448.4096 kilometers) from the Americas Battle.net server (In Anaheim, California if it is in the same place USWest is), and my delay, although lower than Bnet1.0, is certainly more than HoN. I would imagine that most people playing in Europe aren't this close to their Bnet server, so I think that supports my theory that this is an accident. Can anyone run similar tests on the Euro or even the Asia server? Artosis? You have access to both Asia and Americas server, right? This is false. There are many many euro players reporting long delay times. The fact of the matter is Bnet 2.0 does not have LAN latency. Therefore, it is causing higher delay times to everyone (almost). Some people can't notice it because they're either used to non-iccup delay or they just live very close to the server. | ||
semantics
10040 Posts
| ||
Gibybo
United States229 Posts
On March 06 2010 03:09 Mohdoo wrote: Out of curiosity, does anyone playing on the West coast of the USA feel that the delay in SC2 and WC3/SC1 are equal? I certainly do not. It is noticeably higher delay then HoN, but it seems WAY better than WC3/SC1. Though I do not doubt the results of Gibybo, it just seems like there is more to it than that. I am in Arizona actually. However it would be nice if other people would perform these tests too, especially in other locations and on Europe/Asia servers. It did 'feel' better than WC3/SC1 when I was playing too, and HoN actually felt much better than ICCUP, but tests reveal that my initial feelings weren't quite right. Frankly i don't like how this test was done it should have been done relative ms with a high speed camera pointing at the screen and the mouse and when the mouse is fully engaged then you get your latency till movement, As the animations are different from game to game and although sc b.net compared to sc lan would be valid in that test that is only because it's the same game. Yep that would be better, but I don't have a high speed camera I am hoping someone else will step it up The click animations do appear to start at the same time in every game though. BTW, I did post it in the feedback forum: http://forums.battle.net/thread.html?topicId=23425708841&sid=5000 | ||
Jarvs
Australia639 Posts
On March 06 2010 03:18 TheYango wrote: From the Warcraft III Battle.net FAQ: Without derailing this thread too much, I really find this news hard to believe (but its in the FAQ alright). I play on the USWest bnet server of which I have (give or take) 180ms to. Every ladder game I have the exact same ping, regardless of who my opponent or where they are from. I do not have SC2beta and I'm only speculating but the war3 bnet and bnet2.0 function incredibly similarly. | ||
CynanMachae
Canada1459 Posts
But yea... set latency could be so much lower >< | ||
member1987
141 Posts
I don't think lower latency would be the perfect or even best choice. In fact it really depends on your internet connection relative to the location of the server and relative to the other person. Essentially lower latency should work better when all parties are close and have pretty much same internet connection or should i say similar packet loss(in the lower ranks) While higher latency would be better when more people with different internet connections at different and maybe not so close locations are playing. So I think that setting the latency as high as possible in the beta start is the best way to determine the lowest latency settings by the end of beta. So if Blizzard get statistical data of what latency works best for most of the people, that will be the set latency at game release. Of course they could change that after the game is released and having stats from millions of players. So my point is that the high latency should not worry anyone, especially in this stage of development. Blizzard is testing the servers and will do so for the whole time this beta lasts. In fact as I've explained above, having high latency in this beta stage is the best possible scenario for Blizzard to actually determine what the lowest latency can be and just drop it little by little. | ||
| ||