at least you caught that now as oppose to later; this helps you make decisions though. If you are looking for a big upgrade for the time being I would suggest going a really nice single GPU then when you make a motherboard and chipset upgrade you could always buy another and go multicards then. The GPU to upgrade to today though will generally vary on your current PSU and how much you want to spend.
If you have 550W or greater system power supply (with a minimum 12V current rating of 38A) and 2 x 6 Pin power connectors I would suggest something like the EVGA GTX 570 as it will be in a similar price/performance ballpark of the SLI setup you were looking at with the 2x460, will play all the games you listed at max settings with 60+ fps, and when you upgrade board and chipset way down the road you could always upgrade to a second one in SLI or even x3 SLI, which would be beyond overkill for gaming.
As always though research the compatibility of GPU with your current setup. Hope this helps.
physX allows one GPU of your multicard setup [can even be a low end GPU different from your master GPU(s)] to run all physics operations of a game/application. While really cool the problem is there are only so many physX compatible games so it depends which games you want to play.
I would suggest again looking at the gtx 570 as it would be the choice I would go with if I were in your situation. The option (if you do have crossfire support) to run multiple 6850s is there and is not a bad option however I would not recommend it because crossfire is not going to give you 2x the performance of a single 6850, and if you are buying a card now you will want it to last for your future builds. Here is where the 570 will have the advantage as it will outperform almost all other options you currently have, has room to be overclocked, gives you future options of +2x SLI, and will not cost much more than the 460 SLI setup you were looking at.
Last edited by ziekatron; 2011-06-14 at 10:53 PM.
I wouldn't consider a Nvidia card for PhysX at all. There's 0 reason to do so at this point, as today's GPUs are powerful enough to handle games without PhysX just fine.
SC2 runs fine on AMd cards. So does Vindictus. La Noire, Mafia II. It's just another dumb gimmick.
Now read his previous posts people.
He can't SLI.
He's planning on Xfire 6850's, now stop replying.
agreed no one should choose a card based on PhysX, I was explaining because he was curious about it. His decision now that he can't SLI is go with a crossfire of the cards he was looking at (6850) or maybe looking at another option purposed which is a more powerful single GPU. This third option may be worth thinking about for him as it might end up being the most performance for price and have the higher longevity; giving him more options in future gaming, overclocking, and upgrades. Just trying to help :-)
Last edited by ziekatron; 2011-06-14 at 11:06 PM.
Nvidia has slighly more optimized titles than ATI. Their drivers are also a little better imo. Hence I'd opt for the 460. But in any case it's no big difference.
The 6850 is hands down the better card, even though it's a little more expensive now than the cheapest GTX 460. However, claims of Nvidia being the better choice for a "proper" system is laughable. Both companies produce stellar performing cards, with either, you're winning. Also, 6800 Barts GPUs have far superior Crossfire scaling than GTX 400 series cards have in SLI.
That said, AMD video cards also produce a higher quality image in video (such as movies, AVIs, WMVs, etc.) than Nvidia cards (http://www.tomshardware.com/reviews/...orce,2844.html).
However, once you get into the 6900 Cayman line and the GTX500 line, then things get hairier and Nvidia starts to come out ahead.
To answer your question, Omnitron, CUDA cores are essentialy Nvidia's proprietary GPU cores, that's about all.
Last edited by haxartus; 2011-06-15 at 06:15 PM.
Cuda is is a "technology" in most nvidia cards that allows your Graphics card to process very long taking processes like video encoding and rendering and stuff. Since overall your graphics card is much more powerfull than your cpu.
So letting with CUDA, letting your GPU render a video is much faster than having your CPU do it.