IMO if you're running more than a single 580, they really need to be watercooled. Watercooling will mean less heat, less noise and higher overclock. Also, two watercooled and overclocked 580's will outperform 3 x 580's on stock cooling by a long way
No, they don't - two heavily overclocked 580s put up a good fight but three 580s still have a fair lead. Minimum FPS may be higher though, depending on the benchmark used.
But in certain mObo like the asus r3e you can put 2 gtx 580 a slot apart. Which reduces temps quiet abit I hear
Ha ha! Well, I did run a benchmark recently, and I think I remember Dead Beat's: Same settings (as far as I know). Mine (stock): ~24/60/135 (1500) Dead Beat: : ~35/50/110 (1200) So, Dead Beat wins out on smoothness IMHO (and he doesn't do badly in the benchmark either ) but I win out on sheer grunt. On the other hand, I bet he doesn't need to wear headphones to stop hearing the fans! I've been playing around with my overclock a little and the new mobo seems to let the cards go higher. I'll keep playing as in the past my 580s have been a bit temperamental.
I'm only running my fans at half power (6v) and they're basically silent. It would be interesting to see how Pete J's setup compaired to mine in real world gaming tests. I'm of the opinion that running two heavily overclocked cards will scale better and give better results in gaming than running three cards on stock cooling with less of an overclock. I'm not trying to have a dig at Pete at all. In benchmarks he maywell get a better score, but i'd be surprised if it was a better gaming setup. I expect both of our systems will perform pretty much the same at the end of the day.
I am thinking about getting another 580, to SLI them (once I get the updated UD7..) I have no intention of overclocking or watercooling them. Will they be OK, do you think?
I think it'll be very application specific TBH and I am with you on that benchmarks don't necessarily represent the in-game experience. Bah, doesn't matter: I'm sure we're both happy! Yup. Only if you're running at less than 1920x1200 and your CPU isn't overclocked.
I did this earlier today. GPUs were at 820MHz though, which boosted min framerates by ~0.4 and mean by ~2: CPU's actually at 4043MHz; my mobo permanently forces an additional multiplier .
Here's me. GPUs at stock, CPU @3.6, 580 SLI (and it runs at 9% CPU utilisation). http://www.adops.eu/dev/unigine_20110217_2137.html The only game I can't play at absolute top settings [STILL!] is the original Crysis. 2560x just doesn't work well ...
that's very interesting.. so 3 cards won't raise the min fps any- it's all on the top end? then again if pete watercooled those 3- could match the min with the same oc, and blow it out on the max still very good to know.. it looks like when the heat is thrown at the gpu's, dual is the same and tri
How on earth did you manage that with two cards ?! They must be overclocked to within an inch of their lives! What is interesting though is that we're getting the same minimums, whereas Dead Beat gets higher. The only thing I can say is that Sandybridge must boost SLI or something. And yeah, Crysis is still a b*st*rd! I've given up and run it at 1920x1200 4xAA with a high resolution pack and various tweaks that push it beyond 'very high'. I think the frame rate is directly related to resolution i.e. half the total pixels rendered, double the frame rate. Warhead, on the other hand, can run well enough at 2560x1600. Edit: I looked a bit into this and it seems my 580s are performing up to scratch - they match Maxishine's: Which brings me to my next point: when are they going to release a motherboard that can fit a sound card and four dual slot GPUs ?