| Pages: 1 [2] 3 4 :: one page |
| Author |
Thread Statistics | Show CCP posts - 1 post(s) |
Jaabaa
Minmatar
Dental Drilling Corporation
   |
Posted - 2010.05.12 03:05:00 -
[31]
 Originally by: Mara Rinn
And remember: fleet fights are laggy because people keep joining them until they become laggy.
I think you are missing one small detail here.
Your opponent can keep on adding people until it becomes laggy (or a node crashes).
Until CCP removes their "false" timers and just make it a matter of raw force in local, everyone will use any tactic available, even if its hundreds of jump clones leaving/joining fleet (or maybe copying BMs) in a station(/at a POS) to force the opposition into submission (that means ... giving up).
Just because your predecessor(s) can't get a query right, doesn't meant that CCP's DBAs won't puke down the throat of any Dev trying to break their DB. I think you miss out on how much a professional DBA protects his asset (i.e. the DB).
--
EVE Mobile Skill Planner V3 !!
http://evemsp.sourceforge.net/ |
Pharago
Gallente
Piratas Leprosos Guineanos
CorsarioS.
|
Posted - 2010.05.12 03:24:00 -
[32]
Originally by: Jaabaa
Originally by: Mara Rinn
And remember: fleet fights are laggy because people keep joining them until they become laggy.
I think you are missing one small detail here.
Your opponent can keep on adding people until it becomes laggy (or a node crashes).
Until CCP removes their "false" timers and just make it a matter of raw force in local, everyone will use any tactic available, even if its hundreds of jump clones leaving/joining fleet (or maybe copying BMs) in a station(/at a POS) to force the opposition into submission (that means ... giving up).
Just because your predecessor(s) can't get a query right, doesn't meant that CCP's DBAs won't puke down the throat of any Dev trying to break their DB. I think you miss out on how much a professional DBA protects his asset (i.e. the DB).
The difference between functionality and implementation is evident, yeah, we get that.
But if there was no lag at all, and the problems that are p***ing off lots of players didn't exist, it wouldn't matter if the actual codebase was made in brainf*ck or python or c or asm.
|
altolemaeus
|
Posted - 2010.05.12 03:48:00 -
[33]
Wait a second. Wikipedia actually has a redirect from "Brain***" to the programming language? 
|
Pharago
Gallente
Piratas Leprosos Guineanos
CorsarioS.
|
Posted - 2010.05.12 04:19:00 -
[34]
Originally by: altolemaeus
Wait a second. Wikipedia actually has a redirect from "Brain***" to the programming language?
why, of course, everything concerning that language is just mad insane
|
Makar Kravchenko
|
Posted - 2010.05.12 04:50:00 -
[35]
Edited by: Makar Kravchenko on 12/05/2010 04:52:56
Originally by: Pharago
Originally by: Jaabaa
Originally by: Mara Rinn
And remember: fleet fights are laggy because people keep joining them until they become laggy.
I think you are missing one small detail here.
Your opponent can keep on adding people until it becomes laggy (or a node crashes).
Until CCP removes their "false" timers and just make it a matter of raw force in local, everyone will use any tactic available, even if its hundreds of jump clones leaving/joining fleet (or maybe copying BMs) in a station(/at a POS) to force the opposition into submission (that means ... giving up).
Just because your predecessor(s) can't get a query right, doesn't meant that CCP's DBAs won't puke down the throat of any Dev trying to break their DB. I think you miss out on how much a professional DBA protects his asset (i.e. the DB).
The difference between functionality and implementation is evident, yeah, we get that.
But if there was no lag at all, and the problems that are p***ing off lots of players didn't exist, it wouldn't matter if the actual codebase was made in brainf*ck or python or c or asm.
Confirming that we are both speculating, since neither of us actually get to look at the code, unless you are secretly a dev.
To me it just seems like trying to crack off a bunch of
Do
For i = 0 to x
If Timer(i).Enabled = False Then
Select Case(i)
Timer(i).Enabled = True
i = i + 1
End Select
Next i
DoEvents 'make sure that value gets updated amirite?
Loop Untill allDone = True
MOAR TIMERSSSS!
CRAP WHEN IS IT ALLDONE!
sorry, stupid, that was a crack at MSVB6.
I officially fail at staying on topic, and I blame falcon(alcohol).
edit, yep i seen it compile error!
****zzle |
Memorya
|
Posted - 2010.05.12 04:50:00 -
[36]
Edited by: Memorya on 12/05/2010 04:53:46
Network code is crap, but latest problem showed they didnt do nothing to protect client side or server side from exploit's (until recently, when players stod up).
CCP bouth some masive servers (speacly db one), but these servers run on old crapy code and are using 400% (if not more) more resources then actualy needed.
12 core processor and 64gb ram (lol), can easly outperformance a 10k player Lineage 2 or WoW server. Server woud be at 50-60% free resources (speacly ram).
Next thing is which is CCP doing extremly wrong...( ), they are adding all resources to high sec. and dont bother to add apropriate servers for low sec. That's why large fleets most of the time die and have lag.
Owerall... CCP is working on old age code, which will totaly die (we are already seeing the effect), once there will come mayor influx of players. I understand that CCP must feed hes investors with our money and all that, but CCP is diging its own grave and dont actualy care.
For your notice, English is my 5th. language...
------------------------
"English is a funny language; that explains why we park our car on the driveway and drive our car on the parkway."
English is my 5th. Language.
|
Captain Vampire
Caldari
No.Mercy
|
Posted - 2010.05.12 04:52:00 -
[37]
Originally by: Jaabaa
The main crux IMHO, is that CCP allows for massive parallel processing on the server side with REAL SMP support.
The only reason people have to fill in a "Fleet Fight Notification Form is because the server can't divide and balance the load on demand.
Its not Python, and judging by the RAM-SANs CCP has in place, I doubt that its the intra-server bandwidth either.
CCP has to redesign their software to enable REAL load balancing and not dedicated nodes.
The upper has to be quoted for it is probably the brightest insight in this thread. People Python is a very mature programming language and perfectly capable of serving the needs of the CCP developers for plenty of years to come.
In my opinion CCP should dedicate a Dev-blog to the problem, with in-depth information on the bottlenecks that cause the problems.
Even more so I feel that communication between CCP and it's customers could be well improved. CCP requests our patience on a fairly regular basis, which I feel they deserve, the uptime of the Tranquility cluster is beond amazing.
But I feel the patience of the masses would be easier earned when a more robust and continues steam of information would reach said masses.
And I'm not talking about a Dev-blog every two weeks here, but a Dev-blog several times per week.
|
Captain Vampire
Caldari
No.Mercy
|
Posted - 2010.05.12 05:09:00 -
[38]
Edited by: Captain Vampire on 12/05/2010 05:16:36
Originally by: Memorya
Edited by: Memorya on 12/05/2010 04:53:46
Network code is crap, but latest problem showed they didnt do nothing to protect client side or server side from exploit's (until recently, when players stod up).
In not a single MMO other than EVE is it even remotely possible for a thousand players to combat eachother.
Originally by: Memorya
12 core processor and 64gb ram (lol), can easly outperformance a 10k player Lineage 2 or WoW server. Server woud be at 50-60% free resources (speacly ram).
Source; http://www.gamespot.com/pc/rpg/worldofwarcraft/news.html?sid=6228615
At that time World of warcraft gobally had 623 realms, simple math would suggest;
Servers; 13,250 / 623 = 21 Servers per realm
CPU Cores; 75,000 / 623 = 120 CPU cores per realm
That information alone shows in what a epic manner you are overstepping your own intelligence. Please do everybody, but especially yourself a massive favor and sit back in the corner.
|
Memorya
|
Posted - 2010.05.12 05:19:00 -
[39]
Edited by: Memorya on 12/05/2010 05:23:49
Originally by: Captain Vampire
Edited by: Captain Vampire on 12/05/2010 05:16:59
In not a single MMO other than EVE is it even remotely possible for a thousand players to combat eachother.
This shows me, you have no clue what im talking about.
Originally by: Captain Vampire
Edited by: Captain Vampire on 12/05/2010 05:16:59
Source; http://www.gamespot.com/pc/rpg/worldofwarcraft/news.html?sid=6228615
At that time World of warcraft gobally had 623 realms, simple math would suggest;
Servers; 13,250 / 623 = 21 Servers per realm
CPU Cores; 75,000 / 623 = 120 CPU cores per realm
That information alone shows in what an epic manner you are overstepping your own intelligence. Please do everybody, but especially yourself a massive favor and sit back in the corner.
Same here, no clue.
------------------------
"English is a funny language; that explains why we park our car on the driveway and drive our car on the parkway."
English is my 5th. Language.
|
Captain Vampire
Caldari
No.Mercy
|
Posted - 2010.05.12 05:37:00 -
[40]
Edited by: Captain Vampire on 12/05/2010 05:42:17
Originally by: Memorya
This shows me, you have no clue what im talking about.
I blame the language barrier.
But it shows atleast - and quite clearly I might add - your gross understatement (by 2000%) of the hardware requirements for a World of warcraft realm.
You,
just as 99.5% of the EVE playerbase including myself have absolutly no idea of the innerworkings (the code) of the EVE cluster. My parents always learned me not to talk of things I have no knowledge of, a lesson clearly not passed on to yourself.
|
Memorya
|
Posted - 2010.05.12 06:02:00 -
[41]
Originally by: Captain Vampire
Edited by: Captain Vampire on 12/05/2010 05:42:17
Originally by: Memorya
This shows me, you have no clue what im talking about.
I blame the language barrier.
But it shows atleast - and quite clearly I might add - your gross understatement (by 2000%) of the hardware requirements for a World of warcraft realm.
You,
just as 99.5% of the EVE playerbase including myself have absolutly no idea of the innerworkings (the code) of the EVE cluster. My parents always learned me not to talk of things I have no knowledge of, a lesson clearly not passed on to yourself.
That's the problem betwen you and me. I have knowedge of eve cluster and how it work and whats wrong, while your talking on thin air and copy/paste some links here.
Act what you like i step down and let you win, becouse im not interested in pointless argue.
------------------------
"English is a funny language; that explains why we park our car on the driveway and drive our car on the parkway."
English is my 5th. Language.
|
Zeba
Minmatar
Honourable East India Trading Company
|
Posted - 2010.05.12 06:04:00 -
[42]
Just to toss in some /tinfoil on an otherwise informative thread I think the fleet lag is to keep the big fish from eating up the little fish trying to stake a claim in nullsec. It was funny that right before dominion launched there was a rather vocal majority of alliance leaders making claims that they would hotdrop back to empire anyone trying to use the new sov mechanics to even take one unused system just because they could. I assume ccp took this seriously and introduced whatever it was they introduced to make anything resembling a capital blob lag out horribly and so give the little fish enough time to build up the proper sov and attendent cyno jammers. Anyone want to take any bets that sometime after tyrannis ccp miraculously comes up with a solution to the lag problem after everyone has had their 6+ months to fortify their little corner of nullsec?
Just some /tinfoil kids, don't pay me any attention.. 
Originally by: CCP Zymurgist
Get off the forums and go kill someone!
|
Captain Vampire
Caldari
No.Mercy
|
Posted - 2010.05.12 06:08:00 -
[43]
Edited by: Captain Vampire on 12/05/2010 06:16:19
Originally by: Memorya
I have knowedge of eve cluster and how it work and whats wrong,
Ofcourse you do Boy wonder.
I do not mean to jest here but, you do yourself understand how ridiculour your claim is?
Originally by: Memorya
while your talking on thin air and copy/paste some links here.
This statement is beond belief.
You have absolutly no information on my background, and I am the posting facts (including source) yet I am the one going on thin air here?
|
Lord Bral
Minmatar
Viagra Lovers
|
Posted - 2010.05.12 06:41:00 -
[44]
*Ping*Ping*
Can we have a DEV reply to this thread kthnxbye
|
Pharago
Gallente
Piratas Leprosos Guineanos
CorsarioS.
|
Posted - 2010.05.12 06:52:00 -
[45]
Originally by: Zeba
Just to toss in some /tinfoil ...miraculously comes up with a solution...
lol, that has to be a lot of tinfoil dude, not to mention that the thing you pretend we use as proof is the actual resolution of the lag problem, or at least a return to pre-dominion performance, i'd say tinfoil is not enough
|
Liang Nuren
Parsec Flux
|
Posted - 2010.05.12 06:59:00 -
[46]
Originally by: Mara Rinn
Dear Programming Language Experts
HOLY CRAP! Someone not talking out their ass about random things. TBH, I would be surprised if we're looking at mere factor-of-ten scalability issues here. It's much more likely to be algorithmic complexity issues (n^2 vs nlog n etc) - which may be a mere 10x slower when you're looking at 100 people.... and 100x slower at 1000 people.
But, I have NFC how the server is coded.... but I'd be really surprised if Python is the problem.
-Liang
--
Liang Nuren - Eve Forum ***** Extraordinaire |
Ban Doga
|
Posted - 2010.05.12 07:13:00 -
[47]
This thread needs more code snippets!
Any one got some old BASIC stuff they don't need anymore??
|
Makar Kravchenko
|
Posted - 2010.05.12 07:14:00 -
[48]
This either needs to be deleted, or locked, because of the obvious consumption of alcohol that has bolstered my apparent ability to perceive, that which I could never possibly understand. Possibly to a point where I have created a topic that has invoked.. nerdcore combat. No one here actually has valid points pertaining to said topic for the simple fact that the none of us here are secret dev posters, OR are they? Is the ensuing flamegank only a troll of my original post by the devs, or is this all a lengthly meme designed to stop me from posting forever.
*Paranoid Delusional Psychosis*
Request post from real CCP Developers, on real solutions, for real problems, pertaining to real code. Which none of us have.
Comma, Period.
I just wanted to hear from the devs guys, and I had a little too much to drink. Good topic name though right?
|
Batolemaeus Junior
Free-Space-Ranger
|
Posted - 2010.05.12 07:40:00 -
[49]
Originally by: Memorya
That's the problem betwen you and me. I have knowedge of eve cluster and how it work and whats wrong, while your talking on thin air and copy/paste some links here.
Act what you like i step down and let you win, becouse im not interested in pointless argue.
Put up or shut up.
It is safe to assume that nobody in this thread has even the slightest clue about what's going on server side. Only superficial information is available to the public anyways.
The only thing that we have any authority in is game design. We know how they work, and we can predict how different gameplay mechanics would work out if they were implemented. It's the only way how the problem of lacking performance can be reasonably tackled anyways, so we might as well just discuss about how the gameplay mechanics introduced in dominion affected/worsened our problem, and what lessons could be learned from it.
|
Dan O'Connor
Cerberus Network
Dignitas.
|
Posted - 2010.05.12 10:44:00 -
[50]
Originally by: Captain Vampire
You,
just as 99.5% of the EVE playerbase including myself have absolutly no idea of the innerworkings (the code) of the EVE cluster. My parents always learned me not to talk of things I have no knowledge of, a lesson clearly not passed on to yourself.
Neither do you/we have knowledge of the inner workings of WoW. 
________________________
Apply | Achievements |
Captain Vampire
Caldari
No.Mercy
|
Posted - 2010.05.12 19:03:00 -
[51]
Edited by: Captain Vampire on 12/05/2010 19:03:21
Originally by: Dan O'Connor
Neither do you/we have knowledge of the inner workings of WoW.
Could you please quote me claiming that I had?
Though in the past I have contributed to the Mangos project.
Now,
as the Mangos project is a reverse engineered emulation I do not have any knowledge of the actual coding running on Blizzard Entertainment servers but it has left me with some knowledge atleast of how World of warcraft might operate.
Returning ontopic;
CCP has some seriously gifted developers running their crew, they know what they are doing. Some issues are just alot more complex than some of your petit intellects can seem to contemplate.
|
Gnulpie
Minmatar
Miner Tech
|
Posted - 2010.05.12 20:07:00 -
[52]
Originally by: Makar Kravchenko
stuff
Python is totally awesome for the given task at CCP. It is a really high level language with lots of easy-to-use and efficient language constructs.
Time-critical operations can be coded in C/C++ anyway.
Besides, it normally is never the language itself but almost always inefficient algorithms and datastructures which drastically reduce performance.
Another point is that execution speed of the final program is just a small fraction of the things which needs to be considered. Easy to program, easy to debug, easy to extend, well-documented behaviour, compatibility with existing code and development systems etc. are other things which a company needs to consider.
Sorry no, that is a bad idea. Waste years of developer time what for? That the program runs 10% faster? Yeah suuuuure that is a good idea ... not! Just as an example: They should invest the time to change the current gameplay that the compuation cost of n players on a grid isn't in average O(n * n) any more but only O(n * log n). Now THAT would be something totally awesome! And THAT would be well worth years of dev time! And such an algorithm would beat any compiled code by far with many enough players on a grid, regardless which (reasonable) programming language you use. Because, you see, with just 1000 people on a grid the improved algorithm would be 100 times faster and even if the language compared to C would be 20 times slower, then the result would be still a program which runs 5 times faster...
|
Tuon Peandrag
|
Posted - 2010.05.12 22:00:00 -
[53]
To my knowledge, which by this point is sorely outdated, the number one problem CCP has with increasing their capacity is stackless pythons global locks. The GIL prevents them from doing any proper threading. It actually doesn't have any threading to speak of. The way threading works in stackless is that each thread executes one after the other using a round robin scheduler. Which as everyone must realize, is not actually threading.
Stackless python CANNOT take advantage of multiple cores.
|
Pharago
Gallente
Piratas Leprosos Guineanos
CorsarioS.
|
Posted - 2010.05.12 22:21:00 -
[54]
Originally by: Tuon Peandrag
To my knowledge, which by this point is sorely outdated, the number one problem CCP has with increasing their capacity is stackless pythons global locks. The GIL prevents them from doing any proper threading. It actually doesn't have any threading to speak of. The way threading works in stackless is that each thread executes one after the other using a round robin scheduler. Which as everyone must realize, is not actually threading.
Stackless python CANNOT take advantage of multiple cores.
During the development of the last EWMH version (3.0) i had the oportunity to see quite a few threads running on the client, and believe me they actually work very well and are able to do a lot of nasty things, so yes, the client can use and will use all your cores if the OS sends its threads there.
|
Captain Vampire
Caldari
No.Mercy
|
Posted - 2010.05.12 22:55:00 -
[55]
Edited by: Captain Vampire on 12/05/2010 23:02:54
Originally by: Tuon Peandrag
It actually doesn't have any threading to speak of. The way threading works in stackless is that each thread executes one after the other using a round robin scheduler. Which as everyone must realize, is not actually threading.
Err,
Stackless Python supports threading in the form of microthreading, which allowes tasklets to be executed concurrently / simultaneously.
It doesn't allow for multi-core or multi-processor usage sure, but it is still threading.
|
Janus Riddick
|
Posted - 2010.05.12 22:58:00 -
[56]
Originally by: Ban Doga
This thread needs more code snippets!
Any one got some old BASIC stuff they don't need anymore??
10 Print "I know a little about coding"
20 Print "I know a little about coding but I disagree with you violently"
30 Print "I know MOAR than you both, and you are both wrong"
40 Print "Can I haz your stuff?"
50 Print "???"
60 Print "Profit"
70 Print "GB2W!"
80 Goto 10
Pfft! Coding is easy 
|
Tuon Peandrag
|
Posted - 2010.05.12 23:07:00 -
[57]
tasklets block each other, when one is running the others must wait for it to finish or be interrupted. The benefit of using tasklets over threads is that they don't have as much overhead so switching is faster then traditional threads. But at the end of the day, only 1 tasklet will ever be running at a time per python process.
This is fine when you are waiting on lots of IO operations. but when major fleet battles come along and all those calculations need to be done, the advantage of the fast switching is negated.
|
Captain Vampire
Caldari
No.Mercy
|
Posted - 2010.05.12 23:10:00 -
[58]
Edited by: Captain Vampire on 12/05/2010 23:11:15
Gotcha.
I will continue with taking my foot, out of my mouth.
But honestly,
I still think dynamic node support holds more merrit and will probably be something CCP is already looking at / has already looked at.
A complete rewrite of 600.000+ lines of code (as announced at the 2009 Fanfest) would be a massive and almost unmanageable undertaking.
|
Tuon Peandrag
|
Posted - 2010.05.12 23:33:00 -
[59]
I assume that when they are saying rewrite they were just talking about splitting up compartmentalizing all the code so that it wasn't a spider web of dependencies and not really a "rewrite".
You can be damn sure that they are trying hard to find a way to do some proper threading. But I they would need some heavy support form the stackless devs, and you can even see that they are very involved in it, just look at the mailing list,
Kristjßn Valur J=nsson seems to be CCPs major dev in that department, or at least the most interested one. A reply from him on their status would be very helpful
|
Ulviirala Vauryndar
Gallente
Vauryndar Dalharil
|
Posted - 2010.05.13 01:06:00 -
[60]
Posting in a thread that is full of fail and nescience.
Originally by: Makar Kravchenko
farm roids all day, you guys really have no room to speak
You can kindly die in a fire. Burn safe o/
|
| |
|
| Pages: 1 [2] 3 4 :: one page |
| First page | Previous page | Next page | Last page |