InversionYT Posted August 22, 2019 Share Posted August 22, 2019 Okay so I know that we haven't seen any sort of minimum specs for KSP2 yet but I'd like to ask anyway. Will KSP2 be any more capable of utilising multiple cored and threads in games since there has been a massive increase in the number of cores and threads on consumer CPUs in the last 2 years (Thanks AMD, love you guys!) Back when KSP was released, Intel still sold the i5-2500K which was a 4 core part (that sold insanely well). Now for the same tier of CPU, you can have a Ryzen 5 3600X with 6 cores and 12 threads (I have one of its predecessors, the 1600). It would be nice to see KSP2 be more effectively multithreaded as KSP1 does not deal particularly well with large structures and it can grind to a halt on even the most powerful computers because it's only capable of running on a couple of cores and bogs down a lot. Link to comment Share on other sites More sharing options...
shdwlrd Posted August 22, 2019 Share Posted August 22, 2019 We'll have to wait and see. Star theory hasn't released any recommended or minimum specs yet. It would be good thing if KSP2 was true multicore capable game, but we just don't know at this point. Link to comment Share on other sites More sharing options...
Citizen247 Posted August 23, 2019 Share Posted August 23, 2019 The problem as I see it is that you can't really parrallise physics very well if at all if all the rigid bodies are always interdependent. Which they are in a ship built of discreet parts. The forces acting on individual parts are dependent on all the other parts. You can't put the physics for a fuel tank in a separate thread to the engine if it depends on the thrust forces from the engine for instance. The fuel tank depends on the engine thrust so must be calculated after those forces are calculated. What happens if, when parrallised, the fuel tank has physics applied before the engine thread gets done? The fuel tank doesn't have thrust applied to it for that tick, and next tick everything explodes because the engine has thrusted forward into the fuel tank which hasn't had that acceleration applied. The only way I can see to simplify that would be to treat the vessel as a single part unless it's experiencing enough forces to break part connections. Link to comment Share on other sites More sharing options...
Kerenatus Posted August 23, 2019 Share Posted August 23, 2019 (edited) 12 minutes ago, Citizen247 said: The problem as I see it is that you can't really parrallise physics very well if at all if all the rigid bodies are always interdependent. Which they are in a ship built of discreet parts. The forces acting on individual parts are dependent on all the other parts. You can't put the physics for a fuel tank in a separate thread to the engine if it depends on the thrust forces from the engine for instance. The fuel tank depends on the engine thrust so must be calculated after those forces are calculated. What happens if, when parrallised, the fuel tank has physics applied before the engine thread gets done? The fuel tank doesn't have thrust applied to it for that tick, and next tick everything explodes because the engine has thrusted forward into the fuel tank which hasn't had that acceleration applied. The only way I can see to simplify that would be to treat the vessel as a single part unless it's experiencing enough forces to break part connections. Yeah that's a tough one. i remember reading a Star Citizen development article saying that they used a "grid" technique which procedually treats certain parts of a larger vessel as a whole, calculating the interactions between those collection of parts, then calculating the parts within each collection seperately. This way the calculation of a big vessel is much more distributed. Though theses kind of things may be too much to ask for in KSP2 Edited August 23, 2019 by Kerenatus Link to comment Share on other sites More sharing options...
Recommended Posts