Testing results of new rig

Awesome review! Especially with the data provided.

Alex I posted pics in a previous post.  But here they are again.

The question I would like to see an answer to, that you may be able to answer is about the 4gb graphics cards.

Windows uses extra ram because it's there. Graphics cards, in relation to software are the same. The fact that you have 4gb means that smart game programming will use as much of this 4gb as it can, yet not necessarily benefit from it, similiar to how windows will use more ram if you have more free, yet past a certain point, the benefits of "using" this extra memory become so small as to not matter.

What I'm getting at is I have another PC in the house that has 2 GTX690 video cards, which have 2gb of vram per card. I run Skyrim with about 128 mods on it. On Ultra everything, with all mods enabled, I see about 1.8 to 2gb of VRAM usage, yet my FPS are higher than yours, and there is no slowdown whatsoever.

Also in my house is a computer with 2 GTX680s. Running Skyrim with the exact same 128 mods, ultra everything, the FPS is higher than yours but lower than my 690s, by a pretty good margin.

In conclusion, based on my testing with 2gb and 4gb cards, games use the extra VRAM just because it's there as unused VRAM is wasted VRAM, even when the difference it makes is nonexistent. I feel confident that if you had cards with 6gb VRAM, you would see even more average VRAM usage just because its there. I've never seen a benchmark and can't find one (after hours of looking) that show any form of a benefit of 4gb per card as opposed to 2gb.

The result across all benchmarks, is that the end result Frames-Per-Second, with identical graphics quality settings under the same conditions, in a 4gb vs. 2gb card, are not affected by VRAM. In other words, even though the 4gb cards have a higher average VRAM usage and more available VRAM, the 2gb cards produce the exact same FPS and performance under identical testing conditions. All tests I've seen are 2gb vs. 4gb, anything lower than 2gb is eventually going to affect performance, but so far, there have been no performance differences between 2gb and 4gb.

So what I'm getting at, is I can test the exact same game, under the exact same conditions using 2 different 2gb cards but I don't personally own a 4gb card, I've had to rely on a friends borrowed PC in a different temperature environment with a few of different variables to test the same game with a 4gb card. Do you have access to a 2gb card on another pc to do a direct comparison to your 4gb card with all other variables being as close as possible? This is a test that the online world would be heavily interested in, including myself, even though other tests have already been done. This is still a pretty new testing zone.

I took a PC with almost identical everything, except video cards and slight temperature differences, and based on running skyrim with the same 128 mods ultra everything the following results were recorded:

2 GTX 670 4GB SLI = 87 FPS average (~ 2.84gb VRAM used per card)
**This test used a PC with slightly different other hardware, but comparable. Same CPU**

2 GTX 680 2GB SLI = 104 FPS average (1.9 - 2gb VRAM used per card)

2 GTX 690 4GB (2GB per GPU) = 178 FPS Average (Not recorded - software had problems monitoring VRAM usage on dual GPU Cards)

Despite the higher available VRAM, it didn't equate to higher performance. For testing purposes, all were done on a 1920 x 1080 monitor and all involved the same route across central Skyrim, through some of the most demanding environments for shadows, etc.

I would just be curious to see how your results compared to mine.

Thanks for your review.

Edited by Err0xx - 05 Jul 2012 at 6:25pm

Thanks for posting the info! 

Another great job.     Time to sit back and Enjoy.   Thanks.