Help Me Fix My GTX 285(s)!

[DanishDevil]

Background info: Bought them used (registered, willing to RMA, fairly well established member of a forum). They are BFG GTX 285 OC cards, but were shipped with the OC2 BIOS loaded. That's how I kept them. However, through all furmark testing, I also ran OC clocks. They have been folding for the weekend, and put out around 18,000 PPD.

Initially, I was having problems with both in SLI. I was running the latest WHQL Drivers (191.07) with an SLI bridge. While playing Borderlands, on occasion, the game would freeze, and I would have to open Task Manager and then "Switch To" the game.

After further investigation, I noticed that after this happened, my primary 285 ran at 3D clocks, and my secondary 285 ran at 2D clocks.

I then tested them individually, GTX285 #2 first. I hit up a Furmark stability test, and within the initial heat-up, the screen froze for a second, then continued. This was also paired with a drop in temperature. This was the Display Driver crashing, no?

So then, I tested GTX 285 #1, all went well. Then, I tried putting my Geforce 210 as secondary to fold/physx, and it had a fit starting up. I finally reseated my good 285 a number of times and took out the 210 and booted up. I then cleaned out all the nVidia drivers, and loaded up the latest Beta drivers (195.39).

I was just playing Borderlands on it, and within about a half hour, I noticed a quick drop in FPS then a rise, but it never got as fluid as before. I checked GPU-Z, and it was running at 2D clocks.

What is going on with these cards? I'm already planning on RMAing #2 because I know it's crashing on a regular basis. This is the first time I've had a problem with card #1.

PS: Are cards supposed to clock up to 3D clocks when folding? Because after that last crash, this one isn't.

 

[Jstn7477]

Yes, cards run at 3D clocks for folding.

Did you try each card in Slot 1? It sounds like whenever you have a second card in there it throws a fit. Also, when the display driver crashes, the card will run at 2D clocks until the computer is rebooted.

I have a G210 and a 9800GT coexisting in my system, both work great.

EDIT: Oh wait, both were tested individually?

If you load up GPU-Z, do both cards show the same clock frequencies?

 

[angelkiller]

You mentioned some Furmark testing, but you didn't say what your temps were. How are they? Is the card(s) overheating?

Your issue sounds like some kind of protection mechanism kicking in for whatever reason. Maybe go back to stock BIOS and see if the issue is still there?

 

[DanishDevil]

Card #1 maxes at 76C and the other maxes at 86C. The 86C one is the one I'm planning to RMA, but now I think I need to RMA both!

I may try flashing them. Can somebody link me to a flashing guide for nVidia cards? And maybe a BIOS (BFG OC). I haven't flashed a GPU in a long time.

 

[qamulek]

Try using evga-precision/msi-afterburner/rivatuner to lower the clocks on both your gtx285's to see if they stay a bit more stable. If that helps to increase stability(not necessarily be 100% stable, just be more stable then previously) then you probably can blame the cards and RMA them.

Another cause for random display driver crashes is too much volts or too high of a clock on the system ram, or just plain faulty system ram. I believe bad system ram should appear just as unstable regardless of which gpu you use though.

 

[DanishDevil]

I did that (as noted in the OP).

Everything else in the system is at stock or lower right now.

 

[qamulek]

I did that (as noted in the OP).

Everything else in the system is at stock or lower right now.

OP states OC settings rather then OC2 settings :0 Have you tried lower then OC settings? It might help.

 

[DanishDevil]

OC settings are barely above stock. I can try bone stock or underclocking if all else fails.

BREAKTHROUGH: I just talked to my seller, and he said that the buyer of his third card is having the same problem as well. He said that if he manually sets his fan speed to 60%, the problem goes away. The seller ran them at 80% the entire time, so he didn't have a problem. I'm going to give that a go tonight if I get the chance.

I'm thinking I still want to RMA card #2, though. There's no reason two identical 285s on stock TIM application should be running 10C apart.

 

[xBruce88x]

yea the other card switching to 2d clocks def. sounds like some kinda overtemp protection kicking in, since the 2nd card is hotter. have you tried swapping them? the faulty one as the first card and the good card as the 2nd card? see if the good card in this config does what the faulty one was doing... if so then you have a heat issue. hence why boosting the fan speed helped the other guy. try to add another fan near the cards to draw in cool air for them and see if that helps. 86c is pretty warm, almost 200degF

also what powersupply do you have? an overpowered supply (a weaker one trying to run too much) can actually cause parts to heat up too (from what i've heard). oh i see you have a 950w, i guess that would be enough but you do have 2 high end cards along with a 3rd card running. I dunno what the amps on the rails and such are.

you may also want to run a mem test program for the heck of it.

edit: the temp difference is due to the cooler design and how airflow gets affected when multiple cards are near each other

 

[DanishDevil]

Good one is card #1.

950W Enermax Revolution.

Memtest passed when I bought the RAM.

Temps are cards running individually.

I'm 99% sure it's not overtemp. protection (on the second card at least). It would have kicked in at different times during Furmark than it did. It crashed at 65C and 70C, then went up to 86C and stayed there for a few minutes without crashing again.