Sign in with
Sign up | Sign in
Your question

Graphic card problem. clueless! Need help

Last response: in Graphics & Displays
Share
April 30, 2012 5:44:45 PM

Hi I am new here. But I need some help guys.

I originally had a radeon 5570 card. (not the greatest) I had a blast on pc with no problems. But I upgraded to a XFX 6850 1gb and this is where it goes wrong. I had mysterious random shutdowns at random intervals. Mostly during intensive things such as gaming. I immediately jumped to overheating, but no abnormal temps. The gpu stayed around 70 degrees Celsius on kombustor. And about 40 or lower idle. And the highest I have ever seen the CPU get to is 62 degrees and that was after hours of gaming then rendering at 1080p. It stays around, also, 40 idle. I thought it might be the psu failing. I was sure of it, so I purchased another psu. A 550W OCZ. And guess what? It turned off again. I decided to switch back to my 5570 and no problems. So it's definitely related to the gpu.

So next I went round a mates house and put my gpu into his computer. We played games, kombustor and stress tested. No problems. We put it back in mine, and after 3 mins of sniper elite 2 demo (good game btw) it turned off. Now this is where it gets even more interesting. Lowering the graphical settings and taking the case off, increases the time I can play a game, before it turns off. I can play games for about an hour and it turns off. After that it turns off after 3 mins of gaming, which makes me think it's overheating again. But no abnormal readings.

We tested with a multimeter and it seemed normal. 12.28 we were getting on the 12v.

Here are my specs :
Amd phenom ii x4 840 @3.2ghz
XFX radeon 6850
OCZ 550w psu
4gb RAM
500gb sea gate barracuda
M4N68T-M LE V2

The psu is within the recommended standards. The old psu I used it on was a CX500w corsair.
So any ideas guys? Works with 5570 and not with 6850.

Oh and yes of course graphical drivers all up to date.

My mates dad who has worked with computers for computers for 10-15 years is clueless. We are going to put his 850w psu in my pc and see if it changes anything. We didn't have enough time this weekend. But a 500w psu is fine so that CAN'T be the problem?

Thanks guys. I need help here
a b U Graphics card
April 30, 2012 5:57:52 PM

you should try the 6850 with your friends power supply in your system. i think it has something to do with not enough voltages going to the card
April 30, 2012 6:14:49 PM

alvine said:
you should try the 6850 with your friends power supply in your system. i think it has something to do with not enough voltages going to the card


This is what I don't get. It's above the recommended from XFX (by a bit) I have seen people go below the recommended and are fine. It uses 38A on the 12v rail and is a trusted brand. Is it really a possibility that the 6850 is too powerful for both my tested psu's? Like I said I will be trying a 800w one soon.
Related resources
a b U Graphics card
April 30, 2012 6:34:40 PM

If you rule out the PSU (which seems like the most likely culprit, but you've already swapped once, so who knows) :

Check the PCI-E slot on your mobo, it's possible something fell in during the swap, if it needs cleaning, try compressed air or find instructions for how to clean online.

Try lugging your PC to your friends house, play with his peripherals for a while on your case, and see if it dies. (If it works here, it maybe the power in your home isn't very good, or you don't have good grounding, or a flunky monitor or something)

Look around for bulging capacitors on your video card and on your mobo.

Try to run a single ram stick in single channel mode, try all you RAM modules to see if the problem persists.

Since you've got a multimeter, with the power off, and case unplugged, do a continuity (resistance measurement, the omega symbol on the meter) between the ground pin of your mobo connector and the case near your GPU, you should see a very low number (I would guess under 50, but definitely less than 1K). Next, power up the PC, carefully check the voltage between the same spot on the case and the ground pin, which should be 0, or very very small (< .1V).


Alot of these things are likely fruitless, but you've got a real headscratcher on your hands, so I thought I'd throw them out there.
April 30, 2012 6:43:22 PM

djscribbles said:
If you rule out the PSU (which seems like the most likely culprit, but you've already swapped once, so who knows) :

Check the PCI-E slot on your mobo, it's possible something fell in during the swap, if it needs cleaning, try compressed air or find instructions for how to clean online.

Try lugging your PC to your friends house, play with his peripherals for a while on your case, and see if it dies. (If it works here, it maybe the power in your home isn't very good, or you don't have good grounding, or a flunky monitor or something)

Look around for bulging capacitors on your video card and on your mobo.

Try to run a single ram stick in single channel mode, try all you RAM modules to see if the problem persists.

Since you've got a multimeter, with the power off, and case unplugged, do a continuity (resistance measurement, the omega symbol on the meter) between the ground pin of your mobo connector and the case near your GPU, you should see a very low number (I would guess under 50, but definitely less than 1K). Next, power up the PC, carefully check the voltage between the same spot on the case and the ground pin, which should be 0, or very very small (< .1V).


Alot of these things are likely fruitless, but you've got a real headscratcher on your hands, so I thought I'd throw them out there.

Thank you very much for the reply. I might not have mentioned it, but it does turn off at his house as well as mine. So it's not my house, and we dont use a power strip either. We see no bulging capacitors. And I only have one ram stick anyway. I really should get another. And the multimeter is only available at his house. So I will try that when I go the next. I suspect that putting his psu in my pc may fix the problem. But why?

I agree 100% this is a headscratcher. My first pc as well, haha
a b U Graphics card
April 30, 2012 6:49:32 PM

Well, if it does work with the new PSU, it would likely be that your card either simply operates out of spec, drawing more current than intended from the 12V rail, or that you have some other devices, such as HDDs or optical drives, that are sucking power at the wrong times. Or it could be the heat of the PSU building up and reducing it's efficiency to a point where it doesn't supply enough power (check your PSU ventilation and make sure it's adequate, try adding a case fan or two maybe?)
April 30, 2012 6:52:04 PM

Are the capacitors on the video card OK? There might be one or two failing capacitors, which you can tell by ballooning or busted cap head.
April 30, 2012 7:00:11 PM

Thank you. Speaking of this, while playing games we kept checking for heat. We found. Considerable amount of heat coming form the back of the power supply. As far as I know it's meant to blow out cool air, this could be a sign the psu is straining for the 6850. And overheating. It's hard to tell as there no program to measure psu temperature.
April 30, 2012 7:04:34 PM

On your computer, are you able to go into safe mode? If you are go into devices and hit show hiddendevices. If you see amdkmdag dis-able it. If you can install an earlier set of cayalyst control do it. If you can run your card/computer in safe mode it is possibly that version of catilyst control. If you cant find amdkmdag that way just search your c drive for it and delete it.
April 30, 2012 7:57:58 PM

Hey, I'm the friend that he talks about from the original post. We just downclocked the graphics card by 10%, and it survived playing Sniper Elite v2 (where the problems occurred last time, and it seemed to survive practically the entirety of the demo instead of the first few minutes at most. Good sign?

Just playing the waiting/hoping game now. If it were to not turn off now, what would that entail? That there isn't enough amps to support the card full pelt? Or if it does turn off, what would you guys reckon that will entail?
a c 90 U Graphics card
April 30, 2012 8:03:20 PM

try the card without any oc so psu could be handling it well
a b U Graphics card
April 30, 2012 8:10:02 PM

I would say that it likely has something to do with the power draw. Out of curiousity, the card isn't being overclocked at all (nor is the CPU?), is it? As pushing the hardware beyond it's specifications will often increase power consumption more than performance (so a 10% performance bump may consume 20% more power, I'm not a big overclocker, so that's just an example and not likely to be accurate).

It may be, as you said the PSU being unable to supply the current required, or it could be the PSU simply cannot sustain the output required over time, and it's going into a thermal protection mode. IIRC, the PSU will get less efficient as it heats up, making it harder to output the required power and generating more heat, so it's a self feeding cycle.

April 30, 2012 8:25:12 PM

djscribbles said:
I would say that it likely has something to do with the power draw. Out of curiousity, the card isn't being overclocked at all (nor is the CPU?), is it? As pushing the hardware beyond it's specifications will often increase power consumption more than performance (so a 10% performance bump may consume 20% more power, I'm not a big overclocker, so that's just an example and not likely to be accurate).

It may be, as you said the PSU being unable to supply the current required, or it could be the PSU simply cannot sustain the output required over time, and it's going into a thermal protection mode. IIRC, the PSU will get less efficient as it heats up, making it harder to output the required power and generating more heat, so it's a self feeding cycle.


No, everything was at stock before today. The CPU appears to consume more watts than its TDP (95w, it uses 105-110w in-game) and we just underclocked the GPU by 10% appox. It seemed to survive the Sniper Elite v2 demo, whereas it wouldn't before.
a c 90 U Graphics card
April 30, 2012 8:31:47 PM

then you need another psu to play safe on everything
a b U Graphics card
April 30, 2012 9:57:29 PM

PSU problem I'd wager. The cheaper OCZ PSU's are notorius for being bad. Just check the customer reviews on newegg.... I would do as someone else suggested and try your buddies PSU in your computer. What type and wattage of PSU does your buddy have BTW? While your at it, could you give us the exact make and model of PSU you are using so we can check out some reviews on it from various sites to see if there were any issues with the unit during testing. Some PSU's are rated higher wattage wise than they can actually provide under a large load...

PS: AMD recommends minimum 500w PSU for the 6850 so you barely have any room to move power/wattage wise with a 550w PSU.(you can't OC anything) So if the power supply draws any more power at all for whatever reason(eg. Overclocking) it could cause a random shutdown, or if the PSU is getting pushed too hard it will overheat and shutdown. Either way it looks like the PSU to me....
a b U Graphics card
April 30, 2012 10:09:05 PM

Juggernog said:
No, everything was at stock before today. The CPU appears to consume more watts than its TDP (95w, it uses 105-110w in-game) and we just underclocked the GPU by 10% appox. It seemed to survive the Sniper Elite v2 demo, whereas it wouldn't before.


I'm not sure if your cpu has the turbo core feature, but that may be why your seeing the spike in power usage from your cpu. Turbo core=auto overclocking when needed.(eg. games)
April 30, 2012 10:11:27 PM

500w minimum recommended. My system is not overclocked in anyway so its nt drawing any unnecessary power. My friends psu is a 850w. More than enough.

I do believe there is a lot of strain on the psu. Its only 50w above the recommended. But it's my second psu. It did it on my corsair 500w psu. I purchased a new one because I thought the pv turnoffs was a dodgy psu. Clearly not...
a b U Graphics card
April 30, 2012 10:19:34 PM

kermit1515 said:
500w minimum recommended. My system is not overclocked in anyway so its nt drawing any unnecessary power. My friends psu is a 850w. More than enough.

I do believe there is a lot of strain on the psu. Its only 50w above the recommended. But it's my second psu. It did it on my corsair 500w psu. I purchased a new one because I thought the pv turnoffs was a dodgy psu. Clearly not...



Once again it was a PSU that was barely enough....Actually I wouldn't be surprised if under load the Corsair was able to provide more power than the OCZ....Try to get the info I asked you for in one of my previous posts, thanks.
!