Originally Posted by shadow001
Don't get me wrong,i know that microstuttering does happen in a few games here and there,and some do notice it,while others don't,and the same goes for the variable scaling issues that plague all Multi GPU setups,at least up to now,and i don't deny that when comparing single GPU versus single GPU,the GTX280 rocks despite it's price.
But when other factors are considered,such as the agressive price shifts that Nvidia were forced to do once the HD4850 was released and Nvidia has been marketing SLI for a few years now as the ultimate in gaming experience/performance and with now very good cards on the market for 200$ a piece,and depending on individual users current motherboard(SLI capable or not),the situation is a lot more murky and not as clear cut as Nvidia would like i'm sure.
If that wasn't enough,we now have the HD4870 wich if the rumors are true and it really does perform at about the same level as a GTX260 but costs 100$ less,and the GTX 280 only outruns an HD4870 by only 15% best case,but costs 350$ more,it's just another big factor that can rain on the GTX280's parade,an i'm not even talking about the HD4870 X2 card that can be quite a bit faster than the GTX280,based on the HD4870 numbers and using the usual 60~70% higher scaling from 1 to 2 GPU's in terms of overall performance.
What ever the case, I would prefer a single card solution to negate multi GPU problems and I think that is what pizza is on about nevermind the peak FPS etc. Never also mind the min fps if it is high but, what we should get is the data like we did from FEAR's inbuilt demo, the percentage breakdown of the fps, times below 30 fps, below 45 fps and above 45fps etc :0 That would prove decisive info IMHO.
Watching this space with NV's and ATIs new offerings with regards to this