AMD 6970 Black/Grey Bars, Crashing

[ipullstuffapart]

Hi there, I'm new here, just been having a bit of a pickle with my 6970

A while back I started noticing during some games (GTA IV in particular) that my system was crashing to Black/White/Grey vertical bars on my screen, leading to a full system lockup, requiring forced reboot.
After a few times that had happened, the problem seemed to go away, never saw it again until just now.
I have purchased three 1080P IPS panels and hooked them up in Eyefinity, this induced more crashes in the same style, anywhere from 1 minute into a game, to 1 hour into a game.

I'm running it on a 650w Coolermaster power supply (system never pulls over 400w from the wall anyway) and it was running on a 500w Silverstone previously, this changed nothing of the result of the crashing, still does it regardless.

The problem still happens with the sidepanel open in an airconditioned room with a pedistal fan blowing directly over the card.

The card seems warm/somewhat hot when running even still, but the temps in HWMonitor are still in a perfectly fine operating zone.

My case is adequately cooled, although it may not look it in the photo attached, but the card gets plenty of air.

Sapphire AMD 6970
AMD Phenom II X4 955BE
Gigabyte 880GM-UD2H
Zalman CNPS 9700NT
Coolermaster 650w power supply (was Silverstone ST50F-ES, same problem occurred)

8GB Generic 1333Mhz RAM
 

Any ideas on what is going wrong? Any help would be greatly appreciated.

Cheers!

Edited September 11, 2013 by ipullstuffapart

[WarWeeny]

it could be a number of things.

 

1- the card is dying (highly doubtful, but it could be possible)

2- the drivers you have currently installed gives problems (would be my first guess)

3- the card is running to hot even tho you might think the temps are fine (would be my second guess)

4- the overclock you have (if any) is to high and the card malfunctions because of that

 

 

the first thing you should do is downloading a previous gpu driver when you didnt had this problem and see if it persist.

if it still persist, try re-seating the gpu, this can do miracles, clean the pc/videocard while you are at it

check temps with light stress testing programs like heaven or msi afterburner, if it is over 80c, it can be to hot for the card to handle and you should up the fanspeed or downclock the card for lower temps.

Edited September 11, 2013 by WarWeeny

[ipullstuffapart]

1) I doubt it, it's pretty new. One of my friends had used it in his system for only a couple of months before giving it to me, in that system it had so many fans it could probably be classed as a hovercraft.
2) I did try some older drivers and newer beta drivers, which made no difference
3) I'll see, when I get a chance I'll run Furmark on it with HWMonitor to see what happens
4) No OC.... Yet

I reseated it and changed the power supply over with no difference.
I have however seen a topic somewhere saying that sometimes the thermal compound layer on the GPU can burn/isn't very robust. In this case I would have to pull off the cooler. Do you know if by any chance I would need to replace the thermal pads on the RAM? I'm hoping I don't have to, but is it really necessary?

[l1il]

Another simple thought, as I faced it when I went from a 6950 to a XFX 7970 GHz Edition, are you always running some kind of monitoring software while playing?

Whether GPUTweak (Asus), TechpowerUP's GPU-Z, HWMonitor, I had issues running them while playing GTA IV.

A shot in the dark, is your GTA IV modded (Using ENBSeries or another graphic enhancing mod) ? They are increasing risks to be unstable.

Another game I know to be one of the most sensitive to stability is Shift 2 (I like it 'cause I'm a bit of a graphics whore).

 

I agree with what Warweeny said, those can be good clues.

Another suggestion aside the GPU overclock... since GTA IV does fully use quad core CPUs, is your CPU/RAM overclocked?

Those bars as described tend to show a GPU problem, but I noticed your Black Edition CPU. That would be my last guess.

 

Wish you to find your problem.

[l1il]

*EDIT

Double post

Edited September 11, 2013 by StefenHeif

[ipullstuffapart]

After running Furmark, I managed to get it to crash after about 50 seconds. The card went from an idle temp of 50c to hovering around 90c (at that point the fan on the card set to auto hit about 50% and it stabilized), within 20 seconds of that, it crashed to black/grey/white bars. Curious and curiouser.
 

 

Another simple thought, as I faced it when I went from a 6950 to a XFX 7970 GHz Edition, are you always running some kind of monitoring software while playing?

Whether GPUTweak (Asus), TechpowerUP's GPU-Z, HWMonitor, I had issues running them while playing GTA IV.

A shot in the dark, is your GTA IV modded (Using ENBSeries or another graphic enhancing mod) ? They are increasing risks to be unstable.

Another game I know to be one of the most sensitive to stability is Shift 2 (I like it 'cause I'm a bit of a graphics whore).

 

I agree with what Warweeny said, those can be good clues.

Another suggestion aside the GPU overclock... since GTA IV does fully use quad core CPUs, is your CPU/RAM overclocked?

Those bars as described tend to show a GPU problem, but I noticed your Black Edition CPU. That would be my last guess.

 

Wish you to find your problem.

No, I never have any monitoring software going, I used to run coretemp so I could see my temps on my G510 keyboard but they removed the support so I don't use it anymore.
My GTA IV is heavily modded, lots of high polygon cars and a few mods of my own. But this isn't causing it and they're stable, as the crash occurs on barebones GTA IV 1.0.7.0
System crashes on all other games (GRID, Shift/Shift 2, COD4, COD BO etc) after any given amount of time. This time is shorter when running Eyefinity.
I'm not overclocked at the moment, I run stock as my computer runs day and night typically so I like it to be nice and quiet without too much heat to turn my room into an oven.

So as far as I can see it is a temperature problem. I'll try it with the sidepanel off, fan facing at it, etc. and post up the temps.
My question is; why is this happening? I have plenty of cooling.

[ipullstuffapart]

Furmark running 1280x1024 for a max of 6 minutes. 2xMSAA

Temps:
 

• Idle: 65C70C
• Sidepanel on, standard operation and setup -  91C, fan at 46%-54% | Crash after ~ 20 seconds at 91c
• Sidepanel off, standard operation and setup - 91C, fan at 46% | Crash after 8 seconds at 91c
• Sidepanel off, small desk fan pointing at card intake - 91C (slower climb), fan at 54% | Crash after 25 seconds at 91c
• Sidepanel off, small desk fan pointing at card intake, GPU fan set to 100% in MSI afterburner | Stable at 69c, although brought on the need to ratchet strap the case to the floor in fear of a small helicopter coming to life. No crash

It seems worringly hot, especially at idle. Any ideas how to fix?
Don't say watercooling, ain't nobody got money fo dat.

[SpikeSoprano]

Exactly what do you mean by plenty of cooling, from the pic you posted your case looks kind of small, and it would appear your card is over heating.

[ipullstuffapart]

Now to do a more real-world test with Crysis benchmarking tool. See how that goes.

Tex quality: High
AA: 4x
Max of 20 runs on bechmark_gpu
Sidepanel on

Very slow climb to 90c, took 6 loops to get up to temp
Crashed after 9 and a half loops.

[ipullstuffapart]

Exactly what do you mean by plenty of cooling, from the pic you posted your case looks kind of small, and it would appear your card is over heating.

120mm Fan at front running full speed 12v

92mm fan at rear running full speed 12v

PSU fan directly above GPU exhausting

CPU fan runs CPU at 60c and under.

 

Also, ambient case temps are around 35c - 40c.

Edited September 11, 2013 by ipullstuffapart

[ipullstuffapart]

Just installed an Arctic Cooling Accelero Twin Turbo Pro that I had lying around. Aside from the fact that the memory/VRM heatsinks don't stick very well, it fitted perfectly. I'm doing some burn-ins and it seems to be lowering the temps slightly. No crashes yet but it still manages to get to about 90 degrees running completely open air in an air conditioned room. Hopefully once the thermal paste burns in a bit it will be better.

[WarWeeny]

yeah what i thought, the temps are to high for that card, most ati/amd cards go down at around 80/90c and this is true in your case.

ramp the gpu fans up to 100% if you are going to game, 50% isnt going to cut it if the temps are getting 90c.