NS2 will wither on the vine if the server isn't fixed
Scythe
Join Date: 2002-01-25 Member: 46NS1 Playtester, Forum Moderators, Constellation, Reinforced - Silver
Join Date: 2002-01-25 Member: 46NS1 Playtester, Forum Moderators, Constellation, Reinforced - Silver
<div class="IPBDescription">Strike while the iron is hot, and other aphorisms</div>Post-pax we're going to be seeing a big chunk of new players. If they join a server and experience rubberbanding, teleporting, animation hitching and stuttery framerates, they're going to about-face and go back to Tribes/BF3/MW#. Adding the exosuit, new weapons and abilities is going to be worth little if people are turned away by poor server performance. This is the time that people need to be hooked.
This is something that has become almost transparent to most gamers these days. People, by and large, expect games to "just work", both client-side and server-side. Most people experiencing poor performance are likely to blame the client, or their own hardware.
The server application CAN perform acceptably well; provided the hardware is up to the task. Server operators are disinclined to run a servers for NS2 if it means taking two other servers offline to free up capacity.
--Scythe--
This is something that has become almost transparent to most gamers these days. People, by and large, expect games to "just work", both client-side and server-side. Most people experiencing poor performance are likely to blame the client, or their own hardware.
The server application CAN perform acceptably well; provided the hardware is up to the task. Server operators are disinclined to run a servers for NS2 if it means taking two other servers offline to free up capacity.
--Scythe--
Comments
And none of this 'its a beta' nonsense. The devs kindly provided an open beta. They did this to increase game interest and obviously sales etc. But games like this also rely on a lot of word of mouth. Having such poor performance seems to be self defeating. In a beta I would expect poor balance and unoptimized frame rates and random crashes, but all in all pretty playable for the majority of people. I think its passed that point of acceptably poor performance for a beta.
Im not dev beating, I love this game and their work. Im just a concerned fan :)
Does the server side stuff use a lua compiler like LuaJIT continuously or is it statically compiled code?
Can it be changed even though it might prevent easy modding?
That is mostly a separate issue. I say "mostly" because muddying the waters is the fact that, to an extent, server performance affects client performance. Try loading the game up in a server by yourself and see how it performs. You'll likely be surprised.
--Scythe--
more than that. i think they should shoot for 10x, or more.
Perhaps forcefully closing faulty servers that rubber band or have poor settings (too high a player limit for it's CPU) might be in order. While it'll kill off 90% of the servers, allowing new players in to see the huge rubber banding that's going on right now is a good way for people to just write the game off. Yes it's beta, but because it's open beta, it's pretty much like a window into what to expect.
Does the server side stuff use a lua compiler like LuaJIT continuously or is it statically compiled code?
Can it be changed even though it might prevent easy modding?<!--QuoteEnd--></div><!--QuoteEEnd-->
uwe has done a couple things to improve lua & the lua to c++ interface which has made performance more deterministic. but there is no luajit right now. i believe there are additional things they are going to do in this core runtime area to speed things up.
they've gotten many (if not most) speed bumps by just reworking algorithms. expect to see much more of this.
When you have 5 guys making a retail video game from scratch in 3 years, yelling at things doesn't fix them. It takes blood, sweat and tears (which they're shedding as I type this).
When you have 5 guys making a retail video game from scratch in 3 years, yelling at things doesn't fix them. It takes blood, sweat and tears (which they're shedding as I type this).<!--QuoteEnd--></div><!--QuoteEEnd-->
Your devotion to the team is admirable, I share it, but these things have to be said. Sometimes devs of all companies get their heads in the clouds and lose touch with the struggles of the common man. Server performance is the main reason I'm not playing NS2 every night, and I'm certain I'm not alone. A respectfully-provided opinion could hardly be called "yelling at things". I'm highlighting a very real issue which must not fall by the wayside. Playtesting on purpose-made, dedicated servers is hardly indicative of the average pubber experience.
--Scythe--
When you have 5 guys making a retail video game from scratch in 3 years, yelling at things doesn't fix them. It takes blood, sweat and tears (which they're shedding as I type this).<!--QuoteEnd--></div><!--QuoteEEnd-->
no pressure UWE xD
--Scythe--<!--QuoteEnd--></div><!--QuoteEEnd-->
+!
It would have been better if the team waited to really start advertising NS2 until performance was much better than it is now. I think the summer deadline in general was a bad idea, unless they really can make the game optimized by then. I hope they do, and it should be considered the absolute #1 priority next to game-breaking bugs. Either that or push back the deadline because, upon release, experienced gamers will be expecting a steady 60 FPS all the way to end-game.
Sorry to be another thorn in your guys' side, but I also really want this game to succeed instead of having new players become disappointed. This is a great FPS game, and poor performance is the only thing that is holding it back from becoming really great.
UWE does have, and we do playtest, on more reasonable server hardware specs (i.e. mid-high range Xeons). However, server performance has been steadily increasing since the engine test, even if its not really noticeable between each build. The situation with Australia Pure going down was definitely depressing, since it was the only AUS/NZ server that could handle 16-18 players.
The situation with Australia Pure going down <b>is</b> definitely depressing, since it <b>remains</b> as the only AUS/NZ server that could ever handle 16-18 players.
And except for Alchemy's server which he hosts now occasionally for a few hours from his house, our servers have trouble with 12 players let alone 18. We understand that UWE and working on it are none of us seem to be suggesting otherwise, but if any more focus and attention could be achieved through these posts then why not try :).
The issues go beyond the Australian situation though, I suspect the NS2 playerbase will grow rapidly after more exposure from PAX arrives throughout the next few weeks, so people in other regions will be forced to overflow into the not so smooth servers that to them- appear to have decent pings. I know that was the situation in Australia even when the Pure servers were up, you would often see up to 18 people playing in the alternate servers with massive rubber banding staining their impression of the game.
Problem is we don't need gradual server performance increase...
We need the servers to run nearly an order of magnitude faster and this has to happen in less then 6months, this has actually made me very pessimistic about ns2 :(
Nothing kills games faster then poor performance...
Saying it might not help, but if there was nothing to be said than nothing would of been. If we didn't care about UWE and NS2, we wouldn't be here voicing ourselves.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->Nothing kills games faster then poor performance<!--QuoteEnd--></div><!--QuoteEEnd--> especially from game reviews, if the game's review on different media is tarnished by bad performance, that will hurt the game's future a lot.
<!--quoteo(post=1923391:date=Apr 9 2012, 03:25 AM:name=Scythe)--><div class='quotetop'>QUOTE (Scythe @ Apr 9 2012, 03:25 AM) <a href="index.php?act=findpost&pid=1923391"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Try loading the game up in a server by yourself and see how it performs. You'll likely be surprised.
--Scythe--<!--QuoteEnd--></div><!--QuoteEEnd-->
Must try that.
I have to agree with people that they need to do something drastic on the server side when it comes to enhancing performance in the near future. For what I remember the goal is 32 players and now the servers barely manage 16-18 players. and here it's not much people can do about it several people here on the forum have spent lots of their own money on the best hardware they can buy and they start to hit the ceiling when it comes to what a single person can afford to purchase.
When you have 5 guys making a retail video game from scratch in 3 years, yelling at things doesn't fix them. It takes blood, sweat and tears (which they're shedding as I type this).<!--QuoteEnd--></div><!--QuoteEEnd-->
You will see alot of such topics in the next weeks, just because 1000s of new player join the game and don't know alot about it.
It is time for a new FAQ that is hilighted with flashy colors and arrows pointing at it ( same for wiki).
If you are new to the forums it is a pain in the ass to use the search function and alot of important things are written in 100 different threads. And cross thread discussion also happen alot.
Also a list of all good performing servers would be good. Also the hosts that run a 20 slot server but actually only can handel 10 slots should be contacted so that they lower the slot number. Even if it is a 4vs4, noobs could train on a lagfree server and have fun, instead a 10vs10 that feels like a Powerpoint Presentation.
NS2 runs well on any server, depends on slot numbers.
So maybe we could start a thread about server hardware and best slotnumber for it, so that server owner see that a 20 slot server will not bring the performance they wan't and switch to lower numbers.
If we can force all servers to be playable, but with way less userslots, we could atleast do some damage reduction.
I remember 700-900ms ping in those days, and also something about (structure-)entities not being deleted when they were destroyed, leading to massive memory-leaks demanding continues server-restarts. Needless to say that was a mod, and this is a commercial stand-alone, which doesn't remotely make that acceptable again.
<!--quoteo(post=0:date=:name=kabab)--><div class='quotetop'>QUOTE (kabab)</div><div class='quotemain'><!--quotec-->Problem is we don't need gradual server performance increase...
We need the servers to run nearly an order of magnitude faster and this has to happen in less then 6months, this has actually made me very pessimistic about ns2 :(<!--QuoteEnd--></div><!--QuoteEEnd-->
And this one deserves another quote I guess, as people continue bringing up the performance-increase between the alpha and now, which, while nice, isn't by far sufficient if extrapolated 5 months forward.
Now here's the really important bit: I wouldn't so much mind if they took their sweet time getting performance up, even if it ment having shoddy servers on release-day, which would be bad, but I could deal with it, but only if there was the assurance that performance would be up to spec at some point (and allow 16v16 at 60 ticks per example). The fact of the matter is that I'm not even sure it is possible, technically, to get these Lua-scripts to run at sufficient speeds. The initial crappy performance of NS1 wasn't that much to worry about because it had proven technology behind it (compiled C\C++), where-as here it is rather unique in how much Lua-script is being used (other games have used\are using it, but certainly not to the same extend). So with me it's not so much a question of when, but rather one of if...
Same for better multithreading (it uses only 2 cores right now?) and a proper linux build.
I'm afraid that the time is running out. :/ Those few months will be over quickly.
Also it might be down to luck at the end, like they said in the optimization video, you need to figure out which part of the code is taking the longest, rewrite it and see if it improved anything, which is hard to predict. So maybe they will quickly find and solve a few bottlenecks, or they might get stuck in difficult problems (like the lua vm) forever.
Same for better multithreading (it uses only 2 cores right now?) and a proper linux build.
I'm afraid that the time is running out. :/ Those few months will be over quickly.<!--QuoteEnd--></div><!--QuoteEEnd-->
+1
Time has run out; there will be no 30-40% increase in performance that people suddenly have their finger's crossed for now that 'Summer 2012' is the official release date.
New players won't have the patience for the current level of performance which is a massive shame because the core game play mechanics, although needing substantial balance tweaking, are EXCELLENT.
I hope UWE can pull something massive out of the bag here, even it means making substantial sacrifices to their current position re: ease of modding.
When you have 5 guys making a retail video game from scratch in 3 years, yelling at things doesn't fix them. It takes blood, sweat and tears (which they're shedding as I type this).<!--QuoteEnd--></div><!--QuoteEEnd-->
Couldn't have said it better. Talking about how badly a problem needs to be fixed in no way reduces the time it takes to fix it. UWE are working on server performance as hard as they can; they know it's an issue and are working at it furiously.
If you guys still aren't happy with the rate of progress, you're welcome to offer your assistance in debugging and optimizing the engine and code.