For this article, we used an overclocked Q9650 (4.2Ghz) with 4GB RAM, an Nvidia GTX690, Windows 7-64Bit and the latest version of the GeForce ForceWare drivers. Nvidia has already included an SLI profile for this game, however we were CPU limited and we could not determine whether SLI scaling was ideal or not. An option to find out whether SLI is behaving as it should be would be to further increase the game’s graphics settings, however in this case we were also bottlenecked by our VRAM. Each GPU core of our GTX 690 features 2GB of VRAM. As a result of that, we experienced major stuttering side-effects the moment we enabled AA or Ultra settings for Shadows and Reflections. Hell, even enabling HBAO+ High resulted in a stuttering mess.
Been waiting for ages for DSO's take on this.
What I found interesting is how they encountered the same minimal performance difference between 3-4 cores as I did. But doesn't this game engine scale to 8 cores? How well does this game use those cores?
Once again, the cores are maxing out, alongside high ram usage, the performance difference appears to be mild from 3 to four cores, it's not as massive as the jump from 2 to 3 surprisingly, however 4 cores may get higher max fps than 3 cores, I'll look at doing a fraps benchmark between all of them on or before Tuesday If I have time.
Here's what they found.
Edited by TopicClocker - 5/30/14 at 5:23am