UPDATE: Quick question to everyone. What do you think about the 4800x900 resolution gaming tests? How many people use that? Comparatively, how many people game at 2560x1440?
I'm still at 1080p and single monitor for gaming, although using two monitors in general. (It definitely helps with productivity.) Any higher is an investment that is outside of my budget scope. (While QD and 4K monitors are coming down in price, they still aren't really all that close to 1080p pricing... AND my current 1080p 24" primary monitor is still going strong.)
THAT being said, I'd say stick to common resolutions and configs. Just because I am still a single monitor @ 1080p gamer (and dual monitor developer) doesn't mean that there are not enough "Eyefinity" (AMD's term) setups out there not to test for.
As to price/performance ratios, I think this is still a super metric to measure. We all know Intel holds the performance crown. (Will Zen threaten it? WAY too early to speculate.) Its at the lower price points that combinations are made to determine what has the best ratio. Is the pricing structure of AMD's lower (in comparison to Intel's top end parts) performance parts still enough to keep them truly competitive in various build targets? These SBMs can help answer these questions as simple reviews of parts usually are focused on performance and not ranking them in any other way.
I like the idea of "upgrade" competitions regardless if they are integrated as SBMs or not. I like competitions that have a target in mind and those that don't. Sometimes I do get the feeling sometimes the builds do get targeted for what will work for the budget at the moment, and wouldn't mind seeing targets that are chosen with average lifespan for their target in mind as determined by the usual user-base for the target. We all know, for example, that overclocking competitions focus on max stable over-clocks and not so much on longevity beyond a few months... office computers and HTPCs usually for as long as possible, and gaming PCs 2-3 years, 5 at the very most.