| Pages: 1 [2] :: one page |
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Redblade
Caldari
Reikoku
Band of Brothers
   |
Posted - 2007.03.06 06:45:00 -
[31]
 Originally by: Akita T
Stuff
The idea in it self is nice if fleet battles is the only form of battle but it would sadly kill off what little solo / small gang combat there is, as a gang of 3-5 vs 1-2 would never work tacticly as the larger group shares the dmg so u could never take out a strategic target thus letting the larger group win by default even if the smaller group could very well have won the battle by tactics.
As i said tho the idea in it self is good and might be used as a base for something more dynamic that fits all types of combat.
|
Imperil
Northern Intelligence
Artificial Intelligence.
|
Posted - 2007.03.06 08:06:00 -
[32]
As people said before me, it is a Software and not Hardware problem.
Support POS Overhaul - Read it NOW! |
Demian Grey
|
Posted - 2007.03.06 09:03:00 -
[33]
Right now the simulation cell size is a star system. If ccp can make the cell size smaller than a fleet/blob, so that the number of players per process approaches some constant number no matter their density in space, then they can largely fix the fleet issues. I believe some other vendors have gotten there (I recall hearing a demo of mmo crowds on a 2d landscape based on a dynamic clustering a couple years ago), but it is a *very* difficult problem that would likely mean changing some of the earliest decisions made in the architecture of eve, if it's even possible given the constraints of eve's game design.
|
Adunh Slavy
Ammatar Trade Syndicate
|
Posted - 2007.03.06 09:40:00 -
[34]
Get rid of jump gates
http://oldforums.eveonline.com/?a=topic&threadID=477653
http://oldforums.eveonline.com/?a=topic&threadID=477651
Altering the granularity is also a good idea, though not sure how easy that would be to implement. The problem with the process scalability is not how many CPUs you can throw at it, but instead the way the Eve server process is forced to share the same memory address space.
If the code were written to kernel mode and a hardware mezzanine were was used, two processes, on seperate machines, could share the same space, but that is not a simple matter to ask a user mode programer to start thinkig in terms of IRPs.
A comment was made on the OS - NT can perform extreamly well, but it requires some tweaking. First step is turn off all the extra garbage MS assumes the average user is going to want to run. Oracle/Unix TPS is comperable to SQL/NT on the same hardware when both OSes are tuned to the hardware. I suspect CCP has a relationship with MS and that has already been explored.
-AS |
| |
|
| Pages: 1 [2] :: one page |
| First page | Previous page | Next page | Last page |