Is there a way to do multi core for pure simulations?

[Aital]

Is it possible to do something more unusual with multicore to get better performance. Like running the simulation process redundantly on multiple cores simultaneously and then drawing the data from all of them at once to try to make up for problems. Is there a hypothetical number of cores to guarantee it runs like one more enhanced core? Or split up the core into multiple chunks, run one core as the main core, then run smaller parts of the game redundantly like how Static ram runs with 7 simultaneous versions for redundancy? then the most important aspects of the game could run as a chance to get data access to smooth our performance even when performance is technically maxed? Then you are still being a simulation as it all feeds the main core. but you have multiple chance to get/que the logic to smooth out problems.

If you have multiple cores running the entire simulation can't you make a main thread then call on them individually to get near xcore value of performance and still be a pure simulation? I would think Sram is a good example of how it could work.This type of game is all about performance gains as far as practical needs go. Isn't simply duplicating the same thread somewhat easy? Wouldn't you just need a que system to maintain performance overall?

[chew-ie]

Sure, everybody has his own methods though. Some even let one core do a bit of physics, sound and only one or two sectors. Depends on the CPU.

/s

On a more serious note: the devs are constantly working on improving their universe simulation - that's nothing you can do from the outside.