I have been having this issue since January when I purchased the card, I will run through everything I have tried then sum up.

Removed previous GFX card (EVGA 470 GTX), uninstalled nvidia drivers,restarted in safe mode, ran driversweeper to be sure.

Installed a new Corsair HX 650 PSU.
Installed the MSI R6970 (dusty on the back, fan shroud seemd slightly loose but otherwise looked fine).
Installed drivers from MSI website for the card.
First issues started that day, running two EVE clients on two displays (both DVI-D connectors).

Downloaded the latest drivers from AMD's site (12-1 at the time i think) and the corresponding CAP. Removed previous drivers, ran driversweper in safemode again, installed new driver, issue persisted.

Wiped SSD, fresh install of Win 7 x64 Ultimate, standard slog through windows updates, installed drivers, CAP and steam. Driver crashes and recovers again.

Decided to test memory, ran memtest in various configurations (4x 2GB Corsair XMS3 1600mhz), individually, in pairs and every configuration I could try. All checked out 100% healthy running at stock voltages.

After the issue continued to persist I RMA'ed the card with OCUK (who i got it from naturally) , got it back today, it checked out fine with them (£30+ to find out it was healthy so not arguing). Installed it in to my machine and the issue remains.

I'm out of ideas. I have tried an array of drivers from 9.1 to 12.4 (both from MSI and AMD directly). My RAM is fine, everything else seems healthy. My chip (AMD Phenom x6 @2.4Ghz) has given me no issues. Running the updated BIOS version for my MoBo (no updates released in the last year), No major updates to windows 7 that i can think of, just the standard updates they push via the autoupdate. I really am at my wits end and now have no more money for a replacement card, or any other parts really.

Does anyone have any ideas or advice at this stage?

Sys hardware specs

I used to get this with my 7970, only when running two screens.
Have you tried with only one screen attached to see if it still continues to crash?

Cant really help solve the problem as mine still does it (found an old 5670 to run my tv from - sorted)

Currently running on just one screen, continues to crash out. I'm thinking that the only component i have not yet tested/replaced is the MOBO. It seems to function ok, no visible signs of ware or damage. Everything else has been tested and/or replaced (PSU, RAM, HDD's).

Aside from the MOBO, would OCUK have tested the GFX card with two displays when I RMA'd it? It was something I had wondered since i sent it off, should i point out I run a dual screen setup. It struck me as not all that relevant at the time, but now i'm feeling pretty stupid for not mentioning it. I have never come across screens as being the source/cause of an issue.

Try one stick of mem.

This is what i had to sort out for a client.

OCZ Gold 4GB (2x2GB) DDR3 PC3-10666C9 (OCZ3G1333LV4GK) (1333MHz)
Asus M4A79XTD Evo.
Passed memtest 12 hours, Passed Prime-95 8hours passed, Furmark passed, 3Dmark06 passed.
Changed the gfx card.
Drivers 10.4/ 10.5 hotfix/10.6.

Fired up WOW which is the only game the client plays at the moment & instant multiple driver has stopped responding or sometimes hours or more would pass.

Put in some G.Skill RipJaw 4GB (2x2GB) DDR3 PC3-10666C8 & not had a single issue.

Update:

Well the 12.4 Catalyst update sucked. Having continued to run the system normally, enduring the usual driver crashes I have turned my attention to heat and temperatures. This is a new one as i have not looked in to this as a cause in great detail.

Exhibit A:


Now i'm no expert, but GPU Temp #2 & #3 both look damn high given the fan profile i have set. This was taken while running the built in test provided by GPU-Z to ramp the card up to max load.

There is a slight over clock on the card (880 to 910) however this strikes me as way to small of a change (given a default voltage) to cause temps like this.
Just to provide some additional information:

CPU fan faces UP toward the top of the case, two fans on the top and one on the back all vent OUT of the case, two fans on the front draw IN to the case and one large fan draws IN to the case from the side.

The card is in its original condition in regards to thermal paste and heat sink seating, but i'm tempted to open it up, clean and apply some Arctic Silver.

Anyone have any advice or suggestions at this point?

try lifting the gpu up so its not hanging/sagging.

used to have the same card as you, noticed temps we getting progressively worse so i put a screwdriver to support the card and temps were superb then.

give it a go but id RMA it.

I have been reading more and more that this card has an issue with such a heavy cooling setup causing it to be seated in a bad way. The whole fan/heatsink assemble moves around independently of the PCB when I remove/install it. Its this movement that makes me think cleaning off and reapplying the thermal paste and fan assembly would be a good move.

As for RMA, I returned it to OCUK for testing and got it back a few days later with a letter stating they tested it and found it to function perfectly well. Now i'm not sure what OCUK's testing regime is, but its either that which is flawed or the card. I'm really not sure. Can I contact MSI directly and do an RMA through them? Would such a move be worth the effort and cost?

Finally, would cleaning and reseting the heatsink/fan assemble void any warantys and mean i would not be able to get any form of RMA/support? I simply dont have the money to replace this card, in fact I'm not sure I have enough to ship it anywhere (Final year uni student, about to graduate and no cash or overdraft left).

As always, thanks for your time and input.

OCUK most probably tested it on a flat testbed. the card wouldn't sag in that instance.

have you tried lifting card up while under load to see if temps improve?

Screv said:

I have been reading more and more that this card has an issue with such a heavy cooling setup causing it to be seated in a bad way. The whole fan/heatsink assemble moves around independently of the PCB when I remove/install it. Its this movement that makes me think cleaning off and reapplying the thermal paste and fan assembly would be a good move.

As for RMA, I returned it to OCUK for testing and got it back a few days later with a letter stating they tested it and found it to function perfectly well. Now i'm not sure what OCUK's testing regime is, but its either that which is flawed or the card. I'm really not sure. Can I contact MSI directly and do an RMA through them? Would such a move be worth the effort and cost?

Finally, would cleaning and reseting the heatsink/fan assemble void any warantys and mean i would not be able to get any form of RMA/support? I simply dont have the money to replace this card, in fact I'm not sure I have enough to ship it anywhere (Final year uni student, about to graduate and no cash or overdraft left).

Click to expand...

Hello Screv,

The temperatures aren't wonderful but they are far from terrible and nowhere near the levels that would cause data errors. I would expect these temperatures if the case temperature was a little high.

As for the reseating the cooler, feel free, as long as you are careful and don't damage the PCB you won't get any warranty issues from MSI. I just recommend that you don't use a metallic, electrically conductive TIM, stick to the silicon ones.

We'd usually recommend sticking to your reseller for RMA support as we have agreements in place to keep the turnaround time to a minimum. You could RMA directly with MSI but before you did that you'd want to be sure that you've exhausted your other options as you'd have to ship it to our service centre in the Netherlands.

What youdsy is saying is true, a lot of modern cards are heavy and if they aren't firmly supported there can be issues with coolers, cases & PCI slots. I have the same issue with my GTX 570 which I have to occasionally reseat to prevent errors.

Cheers for the reply Scotti, much appreciated.

I'm tempted to go down the paste and reposition the cooler route first. Getting the temps down would give me something to check of the list and probably extend the life of the card.

Do you have any recommendations in regards to pastes to use? I have read about various concerns with Arctic Silver (High capacitance it seems) so i'm tempted to stay away from that. Also should i take in to account thermal pads on any of the components (are they adhesive? can I expect them to be damaged/disturbed by removing the heatsink?).

Like you, I'm doubting the card being the cause of the driver crashes now. My current suspects are the MoBo and CPU, but I don't have the funds to do anything about either for the time being.
Given the recent announcements from AMD about the new chips due out in the next few months I'm going to hold off getting a new chip but i may invest in an AM3+ MoBo to replace my current one. Still this wont occur for a good few months, unless anyone fancys donating one to the cause

Thermal pads are not adhesive & Arctic Silver is very bad for VGA cards due to the exposure components around the core.

As I said, stick to the non-conductive ones and you'll be fine, which shouldnt be difficult as most of them are Silicon based (so much so that I suspect a number of them are the exact same product rebadged). Gelid, Akasa, Arctic Cooling, Xigmatek, Coolermaster, they'd all be fine

Just an update for anyone interested, got my order of Arctic Cooling MX-4 and Arctic Clean delivered about two hours-ago.

Cue me pulling apart my machine, getting some practice in on my CPU, cleaned off and applied and the my GPU. A quick screen shot just to show you the impact.

Exhibit B


Needless to say a vast improvement in temperatures when under load. No driver crashes as yet, however given the system has only been running for 45 min, its to early to tell if they have been cured also. Regardless, a cooler machine is certainly a massive benefit to me.