I was considering adding a second GTX260 to my system but ran across what might be an issue with my MOBO. I see the PCIe 2.0 x16 has either one slot at x16 or two at x8. I didn't consider a board with two x16 at the time so short of buying a new MOBO could I add a second GTX 260 and have appreciable WU output improvement with two running at x8? I'm not entirely sure what the x16 and x8 signify so I was unsure how substantial their impact is on WU completion. Will both cards run at half speed with x8 slots? Thanks!

Those are both great points. After looking up some benchmarks it's painfully clear how outdated this 260 is. It feels like I just bought it; it's amazing what two years does in technological improvement. My RAC is so low because I had to reallocate the GPU to a project with lower requirements as WU's from GPUGRID were taking just too long (typically 3-4 days). That should have been my first hint it was time to update.

As far as upgrading goes what would make the 660 Ti a better purchase than say a GTX 580 or GTX 670 with comparable benchmark scores and within the same price range? Will I have any problems utilizing these boards with a two year old MOBO? I am currently running a P7P55D-E motherboard with first gen i-5 CPU although I plan on upgrading those as well in the near future.

gtx 580 will have a much higher power usage. GTX 670 will have practically the same output in GPUGrid but at a slightly higher purchase price.

they work fine in the first-gen i-series motherboards, but upgrade to sandy/ivy bridge ASAP as the power usage and heat output is so much lower! i would say make sure there is a minimum of pci-e 2.0 x8 slot available though. if pci-e 1.0 only, then need to have x16 slot IMO
____________
XtremeSystems.org - #1 Team in GPUGrid

Great information everyone, thanks. Looks like ill be picking up my new card this weekend. Cant wait to get crunching again.

Out of curiosity, you mentioned that all of the three GPU's would perform similarly for GPUGRID. Does this performance comparison between cards vary for different BOINC projects? I never quite found a way for measuring benchmark performance for GPU/CPU's against projects.

Sorry for my English, I'm from Russia.
I HAS GTI 660 TI from Gigabyte, and has test it on some projects on my Intel Z68 motherboard with two 16x8 PCIE slots. I bring in one GTX 560TI Gig, fnd one GTX 660 Gig. In GPUGRID, E@H and Milkiway@H GTH560TI was faster, then apponent.
Because GTX6XX's processors works at lower frequecy. I have tested it at gamer's banchmarks - its fine, but not for computing. I take it back, add small cash, and bye HD7900...

Hello: Milkiway is a Double Precision task, is the only think that at this time there in Boinc.

The GTX600 series are not good in DP in this case are better GTX500 and AMD / ATI. Greetings.

ATi HD7950 does it (MW@H) for 58 secs. 560 TI - about 5-6 minutes, Number of processors... But in Einshteih@H GTX560 works faster, that HD7900.