Sign in with
Sign up | Sign in
Your question

Mixed brand GTX660ti SLI issues

Last response: in Graphics & Displays
Share
October 3, 2012 4:31:13 AM

Hey guys, sad to say my first post has to be one made in distress, but after lurking and appreciating all the the 3rd-person experience and help, I figured I couldn't do much better for help.

So right after the GTX660ti's came out I did some research, and figured the best card for me to buy was a Zotac AMP! model, as it preformed well in benchmarks. More recently, I have been wanting more power so I bought a used PNY stock GTX660ti for a good price. Being familiar with SLI, I knew that the higher end card *SHOULD* underclock to meet the stock card for SLI, and then together could overclock at matching levels.

After installing the second card and enabling SLI, I noticed no FPS gain on BF3. MSI afterburner showed that with SLI enabled, the load was equally distributed as it should @ 40-60% each card, but no FPS gain was had in Identical environments with V-synch off. Then, with SLI-disabled, Load was put on a single card at 99% usage

Made sure to do the obligitory: Safemode>driversweeper>re-install. After that made no difference, I made a complete new install of windows- met with the same results.

Looking at GPU-z each card is retaining it's core clocks, The zotac@1006mhz boost and the PNY@980mhz boost. I have never seen this before with mixed brand cards.

I did several tests to Prove the following:

*That the used PNY card works 100% on it's own.
*That the core clocks were indeed different during usage
*That Both x16bit PCI lanes work and SLI is fully funtional
*SLI has no Effect on game Performance with mixed brands.

First I removed the functional Zotac card, and tested the PNY card on both PCI lanes. The card pushed through Furmark and Farcry2 bench just as the zotac had,with marginal differences.

I ran the PNY card on Furmark@ stock for 29 fps avg. then with a minor overclock for 31 fps avg.
Satisfied with this, I rebooted and enabled SLI and got a whopping 58 avg. FPS first run, and then 61avg. when applying the same minor OC.
Here is a pic of those results http://i.imgur.com/HZKigh.png

My celebrations were shortlived however when I ran Farcry2 bench again, and noticed no difference in the scene.
I ran 3 tests both with SLI-enabled and SLI-disabled, and saw the avg FPS lay between 85-95 in both cases.
Here is a Pic showing performance changes http://i.imgur.com/Dp9bDh.png
And here a pic showing the Differant core clocks during the test. http://i.imgur.com/HJofw.png

To push it home and show practical application, Here is a screen shot of afterburner showing Clear load distribution between both GPU's, and the heavier load when Only using the single GPU. In both instances I got 50-110 fps settling around 70 avg. http://i.imgur.com/6Aesyh.png


This is really bothering me and I'm at my wits end. Does anyone have any advice or theories?

a c 109 U Graphics card
October 3, 2012 4:49:18 AM

Drivers?
m
0
l
October 3, 2012 4:54:41 AM

Like i said in the post, I made sure to go through the 3 available drivers for the 660ti, all in safe mode with the aid of driver sweeper.
m
0
l
Related resources
a c 109 U Graphics card
October 3, 2012 5:03:20 AM

What about the BETA ones?
m
0
l
October 3, 2012 5:07:41 AM

Yea, unforutunatly with them being so new, there is only 3 nvidia available right now: WHQL 305.68, BETA 306.02, and WHQL 306.23.
I notice no difference between the three. :S Thanks for looking though
m
0
l
!