[FIXED - Skyrim Issue] 1090T Bottlenecking 2x GTX 460's?

The problem is resolved. I am using driver version 290.36. For further information please visit this post:

http://www.overclock.net/t/1184541/1090t-bottlenecking-2x-gtx-460s/20#post_15950079

*********************************************************************


So, I recently got a second GTX 460 for SLI. I didn't think it would bottleneck at all, but I think I am wrong.

Now, I was playing Skyrim today with ALL settings on ultra, 8x AA 16x AF @ 1920x1080p resolution. In most areas I will have a constant 55-60 FPS. However.. I noticed in certain areas where there was extreme lighting, fog, or water effects the FPS would dip down into the low 30's.

I was like?.. "Are you F'ing kidding me? This better not be a bottleneck?" My 1090T is currently clocked at 4GHz/3GHz NB (I like it at 4GHz because I get to use 1.39v). I can put it up to 4.2GHz and be completely stable at 1.5v but I doubt that would even make a difference in this bottleneck.

Well.. I opened up MSI Afterburner and AMD Overdrive to get a feel of what my usages were.. The results are below..



You can see where both graphics cards are utilizing near 100% usage, but then you can see the drop to nearly 65-70% for both cards(this was when I was in one of those extreme lighting/fog/water areas). Then, I went to a different area that wasn't so extreme and then the graphics cards were back at 100% usage. Bottleneck? I think so..


Here is the CPU graph:



Being a 6 core it's kinda hard to tell how the cores are being utilized towards the game.. but if you look at core 5 you can see how when the graphics cards were running at 65-70%, core 5 was running at near 100%. Then, when I changed areas to a less extreme one, core 5 jumped down to about 75% usage and the graphics cards were at 100% usage once again..

So what do you think? Is my 1090T bottlenecking 2 GTX 460's? At least in certain scenarios?


EDIT********More Info

Someone suggested to take the FPS from one area and compare it to 4GHz and 3GHz to test for a bottleneck. The graphs from above are done at 4GHz. The graphs look exactly the same for the tests below except GPU usage was increased/decreased depending on clocks. Results on certain clocks below:

Tests were done at same saved game location(extreme lighting/fog/water area) & 3GHz NB speed:

3GHz - 29 FPS - GPU Usage on both cards: Peaked at 59%

4GHz - 35 FPS - GPU Usage on both cards: Peaked at 67%

4.2GHz - 39 FPS - GPU Usage on both cards: Peaked at 72%

4.3GHz - 41 FPS - GPU Usage on both cards: Peaked at 74%

4.4GHz - 43 FPS - GPU Usage on both cards: Peaked at 77%
Edited by iinversion - 12/16/11 at 6:53pm

I just went ahead and tried this. Saved the game at a certain location and everything. @ 4GHz and I was getting 35 fps. @ 3GHz the fps dropped down to 28 fps. Graphs looked the same as above except even lower GPU usage.


They drop down to 0 because I tabbed out of the game to look at the graphs. They go from 100 to 65-70 % ish because the test at the end of the graph was done in two different locations. One using all GPU the other using only 65-70%.

Damn AMD.

I'm just about to side grade to an i5-2500K. I was already planning on getting Ivy Bridge when it's released.. and they are going to use the same socket so I could just re-sell the i5-2500K and get my Ivy Bridge. Sounds like a good plan to me.

This could very well be the case.. but effects being horribly optimized or not, I don't like not having enough CPU power to push those horribly optimized effects to the max.