I've read a fair chunk of it.
I tried all of the updates for BF3 and all of the driver updates from Nvidia that were supposed to improve performance.
Yet, even though FRAPS showed a decent FPS in the top corner the game was laggy and stuttered.
It almost felt like vsync issues, yet I had it enabled and I could actually see it capping at 60 FPS in the few instances it went that high.
I was just as eager to disprove it as some others in this thread. I would be ! I had three bloody expensive graphics cards that could not run it properly.
To make sure all variables were ruled out I ran the game on all of the cards in two completely different systems. One was a I7 920 @ 3.6ghz, the other was my I7 950 at both stock and 4ghz.
And each time the results were exactly the same.
My GTX 470 was by no means a slouch. Infact, it was pretty much dead even with the brand new 560 TI 448. However, it could not run Battlefield 3 smoothly with everything on ultra AND 4SFSAA. And there is a reason for that.
How do I know for sure? easy. At the time I first started playing BF3 I had 9gb ram. 3x1gb 1600mhz and 3x6gb 1600mhz. So 9gb total, running Windows Ultimate X64.
Again at that time I was using a 60gb Patriot SSD to boot from, so I had disabled my paging file. Up until that point like many others who have brought it up on this forum I had never had any issues with having it disabled. However, as soon as I started playing BF3 I was getting paging errors.
So what was trying to use the paging file? BF3. It was clearly running out of vram and trying to cache textures to my paging file on the hard drive. Yet, because I had disabled it the game simply crashed to desktop with paging file errors.
Now in hindsight what I should have done was just, you know? uninstalled Battlefield 3 as it's a bit crap. Instead I fell for the hype and ended up spending a fortune on a new GPU.
Now there was a very good reason I did that. I don't upgrade my computer unless I deem it worthwhile and completely necessary.
So I maintain. The FPS scores you see in benchmarks are smokescreens. They do not depict tearing or stutter or even microstutter if it doesn't last for more than a second. However, *I* notice it. What my 295s did on paper and what they did in reality differed hugely. When I sold the second system with the 295s in I simply ran Vantage when the guy came. He was happy, PC sold.
Last thing I would have done was run BF3. He'd have ran a mile.
So maybe it comes down to how acute your vision is at picking out anomalies? I know for a fact that I have an incredible set of eyes.
Some good points about what benchmarks dont show.