I'm sitting here on my 3-year old E3-based workstation, and looking at the dual socket X5600-series workstation that's just been retired from it's job as a fileserver.
Both systems have between ludicrous (24GB) and Insane (48GB) of RAM in them for a desktop, so I don't have any real problems running GC3 massive-sized maps. And I've pretty good graphics cards in each (GTX650 / 750 Ti).
What I'm wondering is this:
Has anyone bothered to do actual benchmarking for GC3, and determined where bottlenecks are and at what point upgrades of RAM/CPU/Video make sense?
Things like:
- After you get to a certain point, does improving the Video card seem pointless? i.e. at 2500x1600, my GTX650 seems fine, so I don't see any real point in using the GTX750, but that's just "feel", not a metric.
- Does turn speed improve drastically with better single-core speeds, or better with more cores, or is there some inflection point?
- How big a map can you really play on with 8/12/16/24/32/... Gigs of RAM?
For me, I'm trying to figure out if the 12-core/24 thread X5660 is better than the 4-core/8-thread E3-1240 for running GC3, given that the latter's single-core performance is roughly 30% better but the latter has 3x the multiprocesssing capability. I don't want to bother swapping everything over to the X5660 workstation if it doesn't look like I'll get much (if any) gains.
The closest thing I can find is here:
http://www.game-debate.com/games/index.php?g_id=23701&compareGPU=Galactic%20Civilizations%20III:%20Mercenaries
But their site really isn't set up to evaluate turn-based systems, and the benchmarking makes little sense for 4X games.