Date:
Monday , August 23, 2010
Author:
Brent Justice
Editor:
Kyle Bennett
Google +1

GTX 480 SLI PCIe Bandwidth Perf. - x16/x16 vs. x8/x8

In our continuing coverage of Multi-GPU configurations with varying PCIe Bandwidth, we put x16/x16 and x8/x8 PCIe to the test. Does having less PCIe bandwidth make difference in gaming? We know that a "lesser" motherboard can save you money. Using GeForce GTX 480 SLI we show you the real-world differences.

continued...

GTX 480 SLI - 5760x1200

On the previous page of tests at 2560x1600 with 4X AA we saw no differences in performance. There was clearly no disadvantages to running at x8/x8 at 2560x1600. On this pages set of graphs we are running the games now at 5760x1200 across three displays in NV Surround mode. Since this resolution is more graphically intense than 2560x1200 we cannot run at the same high settings we did at that lower resolution. We had to lower some AA settings in order to actually play the games and get through our run-throughs.

Aliens vs. Predator

Article Image

Unlike the previous page's testing at 2560x1600, we see a definite difference in average frame rate gaming at 5760x1200 resolution in AvP. We have AA disabled because even with 2X AA enabled this game experiences some texture thrashing with GTX 480 SLI at this resolution. Still, even without AA, we are seeing conclusive performance differences in the minimum, maximum and average framerates. They however are not large differences.

Since we are already experiencing great performance, with average framerates above 60 FPS, a slight reduction in framerate of 2-3 FPS makes no difference in our real world gaming scenario. There is no change in the gameplay experience and in no way could we tell it was "slower" while we were gaming.

Battlefield: Bad Company 2

Article Image

In BC2, we see what I think everyone has really been looking for, an obvious difference that is noticeable on a graph at least. You can clearly see that x8/x8 is consistently slower than x16/x16, from start to finish in our gaming experience. The minimum frame rate is lower, the maximum lower and the average frame rate shows an almost 7% difference in performance, favoring x16/x16.

While the difference is very visible in a graph, we honestly could not feel this difference while we were gaming. We had no idea it was actually slower until we looked at it on a graph. The one thing we can see now though, while looking at the graph, is that x8/x8 dipped down more to the 30 FPS line than x16/x16. As any gamer knows, a drop down to framerates like that in a fast first person shooter can be detrimental to their gaming ability.

Metro 2033

Article Image

In Metro 2033 we experienced no change in performance or the gameplay experience at this resolution. This game is not truly "playable" at these settings, so we are pushing it as hard as we can at this resolution, moving up to 4X AA caused the game to be a slide show. Even pushing it as hard as we could, we found there to be no difference in performance. This game is different than the other two games tested here, this game is more shader intensive than texture/bandwidth intensive.