| Pages: 1 2 3 [4] 5 6 7 :: one page |
| Author |
Thread Statistics | Show CCP posts - 13 post(s) |

Dufas
Amarr freelancers inc
|
Posted - 2007.08.13 05:32:00 -
[91]
blah blah blah...wake me up when you give us something useful like freighters with some kind of jump drive (only used in low sec of course) ---------------------------------------- You have failed me yet again Starscream!
|

xeom
Coagulated
|
Posted - 2007.08.13 06:23:00 -
[92]
Originally by: Dufas blah blah blah...wake me up when you give us something useful like freighters with some kind of jump drive (only used in low sec of course)
Yea yea good idea 0 risk hauling hmmm good idea man.
PS:Die. ---
-Videos- Viciously Delicious
|

Gnulpie
Minmatar Miner Tech
|
Posted - 2007.08.13 06:35:00 -
[93]
I don't understand all the people moaning around that nothing new or only bad features will be introduced.
I like the blog VERY much. Maybe people should read it again. They do exactly what is necessary. They hunt down the bugs and improve the core system so that Eve will run much better. They will fix the lag in the systems (a system can be spread about several nodes) and they will fix the desync issues. A new graphic engine will come also. Also a new UI is in work.
What else do you want? Tons of features that will clutter down Eve even more?? Oh yes and then yelling that CCP should fix the game first instead of adding new stuff. They are doing exactly this!
And I think it is very good! Great blog Kieron!
|

Hugh Ruka
Caldari Free Traders
|
Posted - 2007.08.13 06:56:00 -
[94]
Now I HOPE you give us some time (like at least 1 month) to test the new client with Trinity 2 on Sisi.
other than that, good news, thank you.
Originally by: JP Beauregard The experience with Exodus playtesting has scarred me for life. Those were bug-reports, not feature requests, you numbskulls.... 
|

Andrue
Amarr
|
Posted - 2007.08.13 07:23:00 -
[95]
Originally by: Oveur We've been adding millions of dollars of hardware every year, just recently replacing our Database cluster with new IBM X-SERIES 3950 with 64GB RAM, IBM DS4800 storage array and another magnificent beast, a TMS RAMSAN, taking us to 256GB of solid state disk storage on the database layer. We're now running at 18% CPU utilization on peak hours.
Now there's a man who knows his 'b' from his 'B'  -- (Battle hardened industrialist)
[Brackley, UK]
This is not a signature |

Don Shadow
Spectrum Solutions INC Phalanx Alliance
|
Posted - 2007.08.13 07:56:00 -
[96]
Edited by: Don Shadow on 13/08/2007 07:57:28
Quote: EVE has had 2 free expansions per year for almost 3 years now, each one bringing some new functionality. It's now time to improve what we have and address the most pressing issue of performance.
Amen! 
|

MissileRus
|
Posted - 2007.08.13 09:43:00 -
[97]
Edited by: MissileRus on 13/08/2007 09:49:10 theres no superhero greater then Oveur!
k question!
Early next year we are looking at replacing all our servers for the third time with some next-generation hardware, utilizing tech such as Infiniband so we can start doing something like RDMA. This is to enable solar systems to scale multiple CPU cores and even physical machines. Get ready to party because this means bastard systems such Jita and Saila should have better performance. We're not harboring any illusions, we know that regardless of the hardware there will be high-load systems, but it's a start.
THIS is exactly what ive been hoping for ;) better load stuff, i think thats a big problem today. thing is.. what CPUs will you use?
SUN have their new uber processor called Ultraspark T2, 8core 64thread(!!!) cpu on 1.4Ghz uber built for virtualisation n stuff.
then theres Barcelona, also "good" at virtualisation, and probably better using heavy load per thread then Ultraspark T2, but will fewer threads and less virtualisation be better then 64(!!!) threads and built in 2x10Gbit NICs and security features?
Intel are "good" too, though Barcelona will (IMO) have more benefits for something like eve online cluster, also Ultraspark T2 will probably be even more beneficial!
How will eve cluster utelize threading? is eve server sideusing massive amounts of threads? is more threading and virtualisation more beneficial then pure throughput in Gflops?
I want to think eve will benefit super threading and virtualisation, i would like to think Ultraspark T2 would make eve very very happy ;) considering its the fastest CPU for multithreading and best CPU for virtualisation. also the built in features and I/O is extremely nice.
it does cost more then competing CPUs but it equals out since so mutch is built in to the proccessor, that also increases performance and most importent latency witch eve loves right? :D
i know theres no final, and there might be new CPUs upcoming, AMD is soon making 8cores, and Intel are probably also doing so.
any "hints"? id very mutch like to understand more how eve works, how it utelize threads and virtualisation or any plans/ideas how it might be setup.
sorry for the spelling, but i think everyone knows i suck at posting by this time.. CCP ftw!
edit: simple thought 64 nodes = 64 systems on their own node would be nice right? they might not be able to handle as mutch load as other CPU threads but consider 100 or so Ultraspark T2 and you have 100x64 nodes! also if technique is possible there can be balancing so heavy loaded systems utelize more threads to up the load capacity. thats techical for me but rough thoughts 
hope for a CCP techie answer or thoghts/hints/theory about what they want and how it might be achieved 
--------------------------- 4. i like pizza |

Grimpak
Gallente Trinity Nova KIA Alliance
|
Posted - 2007.08.13 10:00:00 -
[98]
Originally by: MissileRus Edited by: MissileRus on 13/08/2007 09:49:10 theres no superhero greater then Oveur!
k question!
Early next year we are looking at replacing all our servers for the third time with some next-generation hardware, utilizing tech such as Infiniband so we can start doing something like RDMA. This is to enable solar systems to scale multiple CPU cores and even physical machines. Get ready to party because this means bastard systems such Jita and Saila should have better performance. We're not harboring any illusions, we know that regardless of the hardware there will be high-load systems, but it's a start.
THIS is exactly what ive been hoping for ;) better load stuff, i think thats a big problem today. thing is.. what CPUs will you use?
SUN have their new uber processor called Ultraspark T2, 8core 64thread(!!!) cpu on 1.4Ghz uber built for virtualisation n stuff.
then theres Barcelona, also "good" at virtualisation, and probably better using heavy load per thread then Ultraspark T2, but will fewer threads and less virtualisation be better then 64(!!!) threads and built in 2x10Gbit NICs and security features?
Intel are "good" too, though Barcelona will (IMO) have more benefits for something like eve online cluster, also Ultraspark T2 will probably be even more beneficial!
How will eve cluster utelize threading? is eve server sideusing massive amounts of threads? is more threading and virtualisation more beneficial then pure throughput in Gflops?
I want to think eve will benefit super threading and virtualisation, i would like to think Ultraspark T2 would make eve very very happy ;) considering its the fastest CPU for multithreading and best CPU for virtualisation. also the built in features and I/O is extremely nice.
it does cost more then competing CPUs but it equals out since so mutch is built in to the proccessor, that also increases performance and most importent latency witch eve loves right? :D
i know theres no final, and there might be new CPUs upcoming, AMD is soon making 8cores, and Intel are probably also doing so.
any "hints"? id very mutch like to understand more how eve works, how it utelize threads and virtualisation or any plans/ideas how it might be setup.
sorry for the spelling, but i think everyone knows i suck at posting by this time.. CCP ftw!
edit: simple thought 64 nodes = 64 systems on their own node would be nice right? they might not be able to handle as mutch load as other CPU threads but consider 100 or so Ultraspark T2 and you have 100x64 nodes! also if technique is possible there can be balancing so heavy loaded systems utelize more threads to up the load capacity. thats techical for me but rough thoughts 
hope for a CCP techie answer or thoghts/hints/theory about what they want and how it might be achieved 
I dunno wtf you have just write there, but sure it looks awesome ---
truth about EVE: Quote: "Guns are fine, boost players"
Quote: "Players are fine, boost guns"
|

VolCh SPb
|
Posted - 2007.08.13 10:17:00 -
[99]
Originally by: CCP Oveur
In other words, everything I talked about in the blog refers to the first release of our new engine, which only uses DirectX9 and for the new visuals requires Shade Model 3.0.
The requirements for the classic engine shouldn't change. The reason I say similar for the "classic" version is that we haven't finished the engine and currently it's the same.
I have a DX9 video card with only Shade Model 2.0 support. It means, that I can use the new DX9 engine without new visuals. I correctly have understood you?
|

MissileRus
|
Posted - 2007.08.13 10:18:00 -
[100]
Originally by: Grimpak
I dunno wtf you have just write there, but sure it looks awesome
yea i know spelling is defenetly not my strong point but maybe im "lucky" and someone understands jibberish 
we all love tech talk, so giving us a bit more understanding how a "future" eve cluster might work and what kind of CPU would benefit the most is basicly what i kinda said. speculations, theory and how sutch a cluster might be built using the aviable technologies.
Ultraspark T2 is defenetly a CPU i think CCP should look up, mostly im interested what kind of technology would benefit eve the most, massive threading/virtualisation or fewer threading but more horsepower/thread etc.
as said i think 100x64 threads would benefit eve most if there was a way for them to add power when the threads dont have enough horsepower, a node/system with heavy load could use several threads, but a less loaded thread might only use 1 thread making while empty systems idles using nothing etc.
--------------------------- 4. i like pizza |

Grimpak
Gallente Trinity Nova KIA Alliance
|
Posted - 2007.08.13 10:32:00 -
[101]
Originally by: MissileRus Edited by: MissileRus on 13/08/2007 10:24:38
Originally by: Grimpak
I dunno wtf you have just write there, but sure it looks awesome
yea i know spelling is defenetly not my strong point but maybe im "lucky" and someone understands jibberish 
we all love tech talk, so giving us a bit more understanding how a "future" eve cluster might work and what kind of CPU would benefit the most is basicly what i kinda said. speculations, theory and how sutch a cluster might be built using the aviable technologies.
Ultraspark T2 is defenetly a CPU i think CCP should look up, mostly im interested what kind of technology would benefit eve the most, massive threading/virtualisation or fewer threading but more horsepower/thread etc.
as said i think 100x64 threads would benefit eve most if there was a way for them to add power when the threads dont have enough horsepower, a node/system with heavy load could use several threads, but a less loaded system might only use 1 thread while empty systems idles using nothing etc.
basicly the ultimate way to using most resources the most efficient way while having capabilety for massive loads in single/multible systems and stil having plenty of power for all other systems. virtualisation is hawt and so is massive threading, its hard though but any ideas or thoughts from CCP how it might work or how it might be possible to build a future cluster with new gen CPUs.
it was not about spelling but the hi-tech stuff.
I can understand a few things tho.
so you mean it's either massive multi-threading or bigger cpu per thread.
....isn't there a way to have both? ---
truth about EVE: Quote: "Guns are fine, boost players"
Quote: "Players are fine, boost guns"
|

MissileRus
|
Posted - 2007.08.13 11:20:00 -
[102]
Edited by: MissileRus on 13/08/2007 11:22:08 yes well kinda, Intel and AMD are not up to virtualisation as mutch as suns new CPU, they have basicly 1-2 threads/core.
example when virtualisating a quad core using 2 threads per CPU core, it could for examble handle 4x2 systems in eve witch would be less powerfull then 4 systems with their own core and 2 threads to work with.
a 8 core capable of 64 threads could handle either 8 more powerfull systems or 64 virtualised machines/systems but that are less powerfull. using the thread way (virtualising way) gives you more systems with their own "power" so more small stable systems, but they cant handle tons of load. heres where CCP comes in can they manage threading capacity to balance systems with heavy load? or will they use core capacity instead of weaker threads when balancing? (hard to explain!)
can you have another core help another core to increase it load capacity? can you have lots of virtual machines helping eatchother out using threading power instead of CPU core power helping eatchother out? information streams in thousands of threads instead of fewer but more powerfull cores basicly. is it possible to build sutch a cluster for eve? threading power getting shared and distributed rather then core power.
having lots of threads means the CPU can handle more stuff basicly, 4 cores and 16 threads or 8 cores and 64 threads? how would it be built? how would the server use the threads/cores, is it possible to load balance using threading/virtualisation or is it better to use real cores and have more horsepower but less distributed among systems? is a system per thread (virtual machine) better then a system per core? can other threads help out a fellow thread in need when its blobbed to death by tons of players or will they go for the old of having only cores helping eatchother out witch is less complex but also less efficient?
eve sounds like it needs a cluster built as a ton of towing boats helping to move an iceberg, instead of overkill moving the water then build a single super machine to move the iceberg.. kinda.. 
so is 100 CPUs eatch with 64 virtual machines handling 64 systems/nodes more effective then 100 quad core CPUs with 8-16 virtual machines handling 8-16 systems/nodes? 
how could a new eve cluster be built? how could it function? ideas/theories and possebileties 
--------------------------- 4. i like pizza |

Rob Locksley
Causing Trouble
|
Posted - 2007.08.13 11:25:00 -
[103]
Originally by: MissileRus Edited by: MissileRus on 13/08/2007 09:49:10 SUN have their new uber processor called Ultraspark T2, 8core 64thread(!!!) cpu on 1.4Ghz uber built for virtualisation n stuff.
Doesn't the EVE cluster run Windows? If so, any SPARC processor might be out of question, as current Windows versions don't run on anything else then x86(-64) and Itanium, as far as I know.
|

Mashie Saldana
Hooligans Of War
|
Posted - 2007.08.13 11:32:00 -
[104]
Originally by: MissileRus how could a new eve cluster be built? how could it function? ideas/theories and possebileties 
Well first of all it will not be based on Sun hardware as CCP are a Microsoft developer.
10GB Ethernet will not be used as the devbog stated infiniband (much lower latency designed for clusters).
As for CPU, they currently use AMD for the nodes and Intel for the brand new database server. Guess it all depends on who has the best price/performance/support when CCP are ready to go shopping next time.
We're sorry, something happened.
|

MissileRus
|
Posted - 2007.08.13 11:36:00 -
[105]
Originally by: Rob Locksley
Originally by: MissileRus Edited by: MissileRus on 13/08/2007 09:49:10 SUN have their new uber processor called Ultraspark T2, 8core 64thread(!!!) cpu on 1.4Ghz uber built for virtualisation n stuff.
Doesn't the EVE cluster run Windows? If so, any SPARC processor might be out of question, as current Windows versions don't run on anything else then x86(-64) and Itanium, as far as I know.
the new Ultrasparc T2 is more a general processor where the first version was more of a specialized processor(like Sony Cell processor), Ultrasparc T2 will get Linux support aswell as SUNs own operating systems, they talked about having more support amongs more operating systems and hardware but i dont recall them mentioning windows..
the Ultrasparc T2 is not made as a special processor like the Cell though, its made more for general purpose, its made for easy threading and looks to be made for competing with Intel and AMD CPUs directly and not being a special processor as others have been.
Ultrasparc T2 is open source too, so basicly motherboard makers can utelize it and operating systems can be made to work with it very very easely and 3rd party makers can make them (if they have the technology too). im sure it will have windows support eventualy as its a very powerfull processor way ahead of both Intel and AMD in the new ways of server computing.
--------------------------- 4. i like pizza |

Wizzkidy
Demonic Retribution Pure.
|
Posted - 2007.08.13 11:49:00 -
[106]
Oveur,
Can you give us some kind of dates your looking at for releasing your fully re-written client?
This client has been a thorn in EVE's side for a good while now you know the main problems we get is because of the old client it just has to be, I would urge your team to get this new clientout ASAP as people are really starting to get annoyed with the way lag hits them
|

Rob Locksley
Causing Trouble
|
Posted - 2007.08.13 12:33:00 -
[107]
Originally by: MissileRus im sure it will have windows support eventualy as its a very powerfull processor way ahead of both Intel and AMD in the new ways of server computing.
I'm not so sure the T2 is so far ahead of AMD and Intel as you say. Due to its highly parallel architecture it needs highly parallel workloads to show its strength, and most current software doesn't provide that. And I don't think Microsoft will port Windows to SPARC. I'm not saying that I know what Microsoft thinks, but I know that they haven't been very eager to support other processor families in the past, so why this one?
Originally by: MissileRus edit4: SUN is actualy trying to compete with AMD and Intel with their new processor due to it being "open source" we might get to see them in our computers at home not so long in the future.
Not a chance, if you ask me. This processor might be very good for some specific server workloads, but current desktop software is not ready for this. Besides, SPARC is incompatible with x86 processors, and this alone is a reason it can't succeed in the desktop market as it stands now. A transition to another CPU architecture would be painful for everyone and there would have to be a REALLY compelling reason and broad backing by the industry for something like this to happen.
|

FireFoxx80
Caldari E X O D U S Imperial Republic Of the North
|
Posted - 2007.08.13 12:42:00 -
[108]
Roll on Xmas 2007 
What I do the rest of the time - Vote for a Jita bypass! |

MissileRus
|
Posted - 2007.08.13 12:45:00 -
[109]
mm your probably right.. very wierd their comparing themselves to Intel and AMD if their not even in the same "legue"..
eve might not be able to run a CPU capable of 64 threads, but 16 threads isnt too bad either right? my thoughts how CCP would use the technology is stil valid (though Ultrasparc T2 might be out of the picture).
when they build the new cluster, will they use the threading strategy and spam multible virtual machines handling systems/nodes or will they use cpu core brute power having virtualisation on the cores reducing systems/nodes but having more powerfull systems/nodes instead?
will they try to load balance on a thread or core level, how might sutch things work? "auto" or otherwice load balancing both on thread or core or even whole CPU level is all tricky and requires software built for it. will CCP focus on building software that can be balanced on thread level? that would use the msot resources aviable and give nodes/systems/functions the amount of power they need without having to mutch idle CPU power in their cluster doing nothing did that make any sence? 
--------------------------- 4. i like pizza |

Xaen
Caldari Caldari Provisions
|
Posted - 2007.08.13 14:18:00 -
[110]
Ummm...
Instead of throwing thousands of dollars worth of hardware at the problem of Saila, why not add additional equally good Level 4 Caldari agents in other areas so the load will be spread around?
I'm not trying to troll, I seriously don't understand why this isn't done. It seems like an easy fix to a very painful problem. ----------- Support fixing the EVE UI Drones should not aggro anything missiles or turrets do not. |

Xaen
Caldari Caldari Provisions
|
Posted - 2007.08.13 14:19:00 -
[111]
Originally by: Gnulpie I don't understand all the people moaning around that nothing new or only bad features will be introduced.
I like the blog VERY much. Maybe people should read it again. They do exactly what is necessary. They hunt down the bugs and improve the core system so that Eve will run much better. They will fix the lag in the systems (a system can be spread about several nodes) and they will fix the desync issues. A new graphic engine will come also. Also a new UI is in work.
What else do you want? Tons of features that will clutter down Eve even more?? Oh yes and then yelling that CCP should fix the game first instead of adding new stuff. They are doing exactly this!
And I think it is very good! Great blog Kieron!
Try reading the post linked in my sig. The post is approaching a year old and none of the problems mentioned in it have been resolved. In fact they've added more on top. ----------- Support fixing the EVE UI Drones should not aggro anything missiles or turrets do not. |

Wizzkidy
Demonic Retribution Pure.
|
Posted - 2007.08.13 15:09:00 -
[112]
Edited by: Wizzkidy on 13/08/2007 15:08:50 Things your new client need to do to become tip top
1. Reduce client lag (really)
2. Give us a decent affect when we get hit by turrets! Make all ships have a real sheild around them and when you get shot it hits the shield (like it would do) until your into armour then it hits your armour (I have never understood this with the current way as all shots hit the hull at all times)
3. New models (your doing this alrdy as u state)
4. Move stuff over to use GPU (again your doing this I presume)
5. Let us customize the damn UI a bit more, how far you go i'm not sure
6. make the map more freindly - as it stands the map is a pain the rear and using the system map seems clunky
I will update when I think of more.
|

Torothanax
|
Posted - 2007.08.13 15:49:00 -
[113]
Quote: This part alone is about 80 man-years which has been performed in the last year. That's right, 80 man-years.
So you've had 40 people working on this 24/7 for the last 2 years? That would make it 80 "man-years".
Or do you mean you've had about 40 people working on it, 40 hours a week, 46 or so weeks a year?
If it's the first, impressive. If it's the second, then cool, still a lot of work, but not what you could call "80 man-years".
|

Gnulpie
Minmatar Miner Tech
|
Posted - 2007.08.13 16:32:00 -
[114]
Originally by: Torothanax
Quote: This part alone is about 80 man-years which has been performed in the last year. That's right, 80 man-years.
So you've had 40 people working on this 24/7 for the last 2 years? That would make it 80 "man-years".
Or do you mean you've had about 40 people working on it, 40 hours a week, 46 or so weeks a year?
If it's the first, impressive. If it's the second, then cool, still a lot of work, but not what you could call "80 man-years".
A man-year is a commen term in project management and means roughly 2000 person-hours. So 80 man-years are 160.000 hours worktime. It is a measure of the amount of work. It says nothing at all how this work is done, especially not by how many people. Furthermore it says nothing if the work could be divided well to be done by many people or if lot of serial work is necessary (like in housebuilding: you cannot build roof before you have the walls and you cannot paint walls before you have a roof etc.)
In any case, 80 man-years is a nice heap of work 
|

Anwylyd Al'Vos
Minmatar Furian Alive
|
Posted - 2007.08.13 16:53:00 -
[115]
Very exciting! I like that you are endeavoring to correct functionality and lag before adding a bunch of new stuff. As someone that has come from "raiding guilds" in other games, I have always assumed I would never get into pos warfare because I assume it would take the joy out of the game, so I haven't had some of the issues others have, but I am happy for the optimizations all the same.
Great job CCP, on creating (IMO) the BEST MMO on the market. ---
: - other works |

M00dy
Killed In Action Sev3rance
|
Posted - 2007.08.13 17:09:00 -
[116]
Nice Blog! 
Pew Pew |
|

CCP Oveur

|
Posted - 2007.08.13 17:16:00 -
[117]
Originally by: Mashie Saldana
Originally by: MissileRus how could a new eve cluster be built? how could it function? ideas/theories and possebileties 
Well first of all it will not be based on Sun hardware as CCP are a Microsoft developer.
10GB Ethernet will not be used as the devbog stated infiniband (much lower latency designed for clusters).
As for CPU, they currently use AMD for the nodes and Intel for the brand new database server. Guess it all depends on who has the best price/performance/support when CCP are ready to go shopping next time.
Quite correct.
Executive Producer EVE Online
|
|
|

CCP Oveur

|
Posted - 2007.08.13 17:18:00 -
[118]
Originally by: Wizzkidy Oveur,
Can you give us some kind of dates your looking at for releasing your fully re-written client?
This client has been a thorn in EVE's side for a good while now you know the main problems we get is because of the old client it just has to be, I would urge your team to get this new clientout ASAP as people are really starting to get annoyed with the way lag hits them
It's not a rewrite of the client, it's a rewrite of the graphics engine.
As the blog states, we're looking at a November/December timeframe.
Executive Producer EVE Online
|
|
|

CCP Oveur

|
Posted - 2007.08.13 17:20:00 -
[119]
Originally by: Xaen Ummm...
Instead of throwing thousands of dollars worth of hardware at the problem of Saila, why not add additional equally good Level 4 Caldari agents in other areas so the load will be spread around?
I'm not trying to troll, I seriously don't understand why this isn't done. It seems like an easy fix to a very painful problem.
Because there are good Level 4 Caldari agents in other locations already?
Saila will be dealt with though, amongst other things, moving agents from there, creating even more agents otherwere. The main problem with Caldari is that it's the most populated race with almost the least amount of area.
Ps. The problem isn't solved with more hardware. As I stated, we require new hardware to utilize new technology which our software needs to scale. Look up the wikipedia links in the blog if you are interested in details.
Executive Producer EVE Online
|
|
|

CCP Oveur

|
Posted - 2007.08.13 17:21:00 -
[120]
Originally by: Gnulpie
Originally by: Torothanax
Quote: This part alone is about 80 man-years which has been performed in the last year. That's right, 80 man-years.
So you've had 40 people working on this 24/7 for the last 2 years? That would make it 80 "man-years".
Or do you mean you've had about 40 people working on it, 40 hours a week, 46 or so weeks a year?
If it's the first, impressive. If it's the second, then cool, still a lot of work, but not what you could call "80 man-years".
A man-year is a commen term in project management and means roughly 2000 person-hours. So 80 man-years are 160.000 hours worktime. It is a measure of the amount of work. It says nothing at all how this work is done, especially not by how many people. Furthermore it says nothing if the work could be divided well to be done by many people or if lot of serial work is necessary (like in housebuilding: you cannot build roof before you have the walls and you cannot paint walls before you have a roof etc.)
In any case, 80 man-years is a nice heap of work 
What he said 
Executive Producer EVE Online
|
|
| |
|
| Pages: 1 2 3 [4] 5 6 7 :: one page |
| First page | Previous page | Next page | Last page |