Why are almost all benchmarks done at the absolute max settings? Anybody can figure out that a 760 can't run crysis 3 at max. In my opinion it's more interesting to know what settings would be optimal.
I would much rather play Crysis at 720p medium settings at 60 frames than 1080p ultra at 30 frames.
When somebody says "A gtx 760 only gets 30 fps in Crysis 3 at max settings " I dont learn anything. However, if they where to say "To get the most out of your 760 i would recommend playing at these settings....).
Please note that this isn't me tellign Logan and the crew how to do their job. I enjoy almost everything they put out, and i often use their benchmarks as guidelines. I'm just curious as to why you wouldn't put the cards in a real world scenario.
The idea of benchmark is that its the same standard stress for all the competitors. It brings out the difference between different cards. If you start optimizing the game for every card you test you don't really get the same result. It's all about the standard.
And you cant really do all benchmark on medium setting because then cards like Titan can be bottlenecked by other components just because some component can't handle that high FPS.
And if you want easy information on what settings you should use Nvidia has that experience thing were you can just use what other people have discovered what is good for your system.
In short, it provides a better comparison with other hardware. Price/performance is probably the most important comparison that can be made, and that requires some standardised test.
In addition to that, most people can work out the appropriate settings based on the ultra setting performance values.
e.g 25 FPS at ultra settings = turn down to high settings for a playable framerate