Did I kill my graphics card?

F

fantazio

Limp Gawd
Joined
Dec 21, 2002
Messages
203
I have a brand new build with a Radeon 7850 in an Asrock H77M mobo. I had the radeon at a stable overlock at 1050/1250. I did NOT touch voltage or anything. Just core/memory OC. -- I stress tested for a couple hours yesterday with Furmark and had 66C max. During gaming I rarely got above 50-55c. Today while gaming my computer suddenly shuts down, and everytime I boot up Win 8, my monitor shuts off at the Welcome screen. This also happens in UEFI bios and if I press F8 to get into safe mode. I simply cannot boot up at all with the card now.

I have been able to boot up with IGP and remove ATI drivers with Drive Sweeper, and CCC is also gone. The Radeon still won't boot up.

Is the card fried? Anything else I can do? Is there a possibility that it's the motherboard? If so, should I try to find another gpu to test the slot?

Thanks in advance. I just got it so I can get Newegg to replace it I'm sure, but really sucks that this would happen. I read all over how this card overclocks well, and i really watched temps carefully. Not sure what I did wrong.
 
Last edited:
OCZ 600w

This is getting more bizarre. I tried to boot up with the Radeon plugged in, but the video output going through the IGP. It freezes but this time I see the BSOD error. First I get kernel security check failure. Then windows reboots and I get machine check exception.
 
I just tried another slot. Didn't work. I also borrowed another PCI-E card and put it in, and it works fine so it's not the slot. I guess it's the card. Could it be a software issue or is it definitely hardware?
 
Did you test the same card on a different mobo?... If that also does not work I am afraid it has to be the card.... You can try reloading the BIOS of the motherboard by shorting the right pins. Consult your motherboard manual for that... Do you have Virtu MVP support on your mobo?
 
Could just be a bad card, you need to remember that overclocking without increasing the voltage will cause the current to increase (P=IxV) so you may of just pushed the VREGs beyond their limit. Or had a bad card that would of died anyway :)

Seems you have your diagnosis anyway, RMA it and see what happens :)
 
I had no clue that overclocking could pose problems that would not be seen with high temperature. I thought I'd be ok since Furmark had it at 65/66 during burn in. :/ This card should handle 80c. Maybe it was just a bad card and would've failed anyway.

Do you guys know if my RMA could potentially be rejected?
 
its possible if you tell them it died because you were overclocking they won't accept the return....

Did you try the card on another power supply? That'd be what I'd check out first if possible, iirc those ocz psu's aren't too great, though I could be mixing ocz up with another brand...
 
Well overclocking does force the silicon to work at a higher frequency than it was designed for, not every "stress" on a GPU manifests as high temperature. Sometimes the fans will spin faster because of the increased heat output, but the temperature remains the same, for example. You can't gauge everything from the temperature alone.
 
I stress tested for a couple hours yesterday with Furmark
Click to expand...

There's you're problem right there.....

Next time, forget Furmark. The cooling solution is designed to run at 100% load peaks for short intervals, it is not designed to cool a GPU running 100% simulated load 100% of the time for hours at a time.
 
flu!d said:
There's you're problem right there.....

Next time, forget Furmark. The cooling solution is designed to run at 100% load peaks for short intervals, it is not designed to cool a GPU running 100% simulated load 100% of the time for hours at a time.
Click to expand...
Yes and the vrm temps could have been high. Some of the non referance coolers copl tge gpu good but not the vrms. I would keep a eye on vrm temps when running furmark.
 
455olds said:
Yes and the vrm temps could have been high. Some of the non referance coolers copl tge gpu good but not the vrms. I would keep a eye on vrm temps when running furmark.
Click to expand...
I wouldn't be surprised if the VRM's are the culprit here. Furmark and other programs like it are shit, plain and simple. Unfortunately, somewhere along the way misinformed/amateur reviewers worked it into their suites and we still have issues like this even today.

PSA: NEVER USE FURMARK
 
You used furmark and failed to monitor your VRM temps. nice. furmark is the worst choice for stability testing. it is more a test of the quality of your cooling solution because it runs at 100% load consistently which is unrealistic. people who have water cooling don't need to bother but for people with air cooling the VRM temps can skyrocket to 120+c depending on the cooler especially with overclocking. anything above 100c for sustained periods is not advisable. anyway return your card for RMA and next time stay away from furmark or msi kombustor. Use heaven 3.0, 3d mark 11 for for stability testing. and confirm if those overclocks work with a demanding game like BF3, Witcher 2, Crysis 2.
 
raghu78 said:
You used furmark and failed to monitor your VRM temps. nice. furmark is the worst choice for stability testing. it is more a test of the quality of your cooling solution because it runs at 100% load consistently which is unrealistic. people who have water cooling don't need to bother but for people with air cooling the VRM temps can skyrocket to 120+c depending on the cooler especially with overclocking. anything above 100c for sustained periods is not advisable. anyway return your card for RMA and next time stay away from furmark or msi kombustor. Use heaven 3.0, 3d mark 11 for for stability testing. and confirm if those overclocks work with a demanding game like BF3, Witcher 2, Crysis 2.
Click to expand...

+1 to this. Save yourself the headache next time and just test it using the games that you play.:rolleyes:
 
raghu78 said:
You used furmark and failed to monitor your VRM temps. nice. furmark is the worst choice for stability testing. it is more a test of the quality of your cooling solution because it runs at 100% load consistently which is unrealistic. people who have water cooling don't need to bother but for people with air cooling the VRM temps can skyrocket to 120+c depending on the cooler especially with overclocking. anything above 100c for sustained periods is not advisable. anyway return your card for RMA and next time stay away from furmark or msi kombustor. Use heaven 3.0, 3d mark 11 for for stability testing. and confirm if those overclocks work with a demanding game like BF3, Witcher 2, Crysis 2.
Click to expand...

This is a problem with the card them. Cards shouldn't fry running specific software. If a new card can't hold out running Furmark for a bit, I don't want to know what happens to it 12 months down the road when the dust has built up a bit on the card and a new game comes out that works the VRM harder.
 
ocellaris said:
This is a problem with the card them. Cards shouldn't fry running specific cosftware. If a new card can't hold out running Furmark for a bit, I don't want to know what happens to it 12 months down the road when the dust has built up a bit on the card and a new game comes out that works the VRM harder.
Click to expand...
Nonsense. That's like saying a car's engine will blow up accelerating on to the highway in a year because it can't run at redline for a week straight. Graphics cards, like any other engineered product, have design specifications that are meant to be followed. Because someone can write a piece of software that circumvents that doesn't mean the product has a faulty design.
 
Mr. K6 said:
Nonsense. That's like saying a car's engine will blow up accelerating on to the highway in a year because it can't run at redline for a week straight. Graphics cards, like any other engineered product, have design specifications that are meant to be followed. Because someone can write a piece of software that circumvents that doesn't mean the product has a faulty design.
Click to expand...

Furmark doesn't circumvent the card's design specifications, it just works the card hard. The cards should be able to hand situations where the VRMs are maxed out. Future games or apps may work the VRMs just as hard as Furmark does, the card should operate correctly. Cards are designed poorly if Furmark kills them.

Car analogies are stupid in this situation since then you are dealing with mechanical wear as well. Plenty of PC components can go right from the box to 100% load for the duration of their life without a problem. If someone bought a new i5 and it failed IntelBurnTest right out of the box, people wouldn't go, "LOL don't worry about that, you will never use the CPU as much as InterlBurnTest does!"
 
ocellaris said:
Furmark doesn't circumvent the card's design specifications, it just works the card hard. The cards should be able to hand situations where the VRMs are maxed out. Future games or apps may work the VRMs just as hard as Furmark does, the card should operate correctly. Cards are designed poorly if Furmark kills them.

Car analogies are stupid in this situation since then you are dealing with mechanical wear as well. Plenty of PC components can go right from the box to 100% load for the duration of their life without a problem. If someone bought a new i5 and it failed IntelBurnTest right out of the box, people wouldn't go, "LOL don't worry about that, you will never use the CPU as much as InterlBurnTest does!"
Click to expand...
I think a stock video card should be able to handle furmark or there is something wrong with it. You should be able to run a stock car at wide open throttle on the highway also.
There are some exceptions,I would still keep a eye on things.

I wouldn't expect or trust a overclocked card running furmark.
 
It was the combination of overclocking and Furmark that killed the card, so it had nothing to do with the design. This was a fault due to lack of knowledge coming from the user. I would only use such a 'burn-in' application at stock speeds, or a factory OC in which case the video card manufacturer has already upgraded the cooling solution for said card.

In such use of this analogy, he boosted the engine and didn't think about the cooling aspect. Stock cooling components can only handle so much heat.
 
ocellaris said:
Furmark doesn't circumvent the card's design specifications, it just works the card hard. The cards should be able to hand situations where the VRMs are maxed out. Future games or apps may work the VRMs just as hard as Furmark does, the card should operate correctly. Cards are designed poorly if Furmark kills them.
Click to expand...
This is completely wrong, and unfortunately gets perpetuated by amateur/ignorant posters on boards, which leads to problems like we've seen in this thread. Furmark and programs like it are power viruses and are designed to push cards beyond their operating specifications. Both AMD and nvidia actually have hardware and software protections against such programs, so even when you think you''re running them "100%" it actually isn't even close.

The reason behind this is it's stupidly expensive and a waste of money to overengineer cards to meet unnatural or realistic operating loads and environments. Since this card did burn out but obviously the design passed internal testing, my guess is somewhere in overclocking these protections were circumvented or completely turned off (which some programs do), which is a shame.
ocellaris said:
Car analogies are stupid in this situation since then you are dealing with mechanical wear as well. Plenty of PC components can go right from the box to 100% load for the duration of their life without a problem. If someone bought a new i5 and it failed IntelBurnTest right out of the box, people wouldn't go, "LOL don't worry about that, you will never use the CPU as much as InterlBurnTest does!"
Click to expand...
Actually that can quite easily happen on cheaper motherboards with poor or cheaper power delivery systems. Also, CPU's obviously aren't nearly as delicate as GPU's. The car analogy illustrated the problem and stupidity in running something in an unrealistic or improper load and then being upset when it fails.
455olds said:
I think a stock video card should be able to handle furmark or there is something wrong with it. You should be able to run a stock car at wide open throttle on the highway also.
Click to expand...
Again, that's not the whole story. Also, it's the engine speed that's the problem, not that car's speed on the highway or the driveway. I'll make a note not to use car analogies on a forum where people aren't knowledgeable about such things.
 
455olds said:
I think a stock video card should be able to handle furmark or there is something wrong with it. You should be able to run a stock car at wide open throttle on the highway also.
There are some exceptions,I would still keep a eye on things.

I wouldn't expect or trust a overclocked card running furmark.
Click to expand...

You do know that you can turn off any reference 4870, 4890, or 4870x2 with OCCT, right?

There is no good reason to run any of these programs for two hours. They won't tell you anything in that time that it won't tell you in a few minutes. Honestly I prefer Crysis for stress testing my cards anyways.
 
Mr. K6 said:
I wouldn't be surprised if the VRM's are the culprit here. Furmark and other programs like it are shit, plain and simple. Unfortunately, somewhere along the way misinformed/amateur reviewers worked it into their suites and we still have issues like this even today.

PSA: NEVER USE FURMARK
Click to expand...

Or use furmark to kill your graphic card. ;)
 
Hi everyone. OP here. I received a replacement today. Should I just not overclock it at all and leave it at stock settings? This is the DD model, so it has dual fans and is supposed to overclock well, but due to my previous experience I am very wary of OCing it.
 
Last edited:
fantazio said:
Hi everyone. OP here. I received a replacement today. Should I just not overclock it at all and leave it at stock settings?
Click to expand...

No, you shouldn't overclock anything unless its directly covered under the warranty.
 
Yes, overclock it if you see the need for more performance. No idea why you wouldnt.
 
Yes? If overclocking were unsafe, it would not be so common. I've never had a part die from overclocking... because none of my hardware has ever failed.

He got a bad card and thats that.
 
45454asasa said:
Yes? If overclocking were unsafe, it would not be so common. I've never had a part die from overclocking... because none of my hardware has ever failed.

He got a bad card and thats that.
Click to expand...

Unless overclocking is specifically covered by a warranty, it's not very kosher to overclock a card and then send it back for a warranty replacement later.
 
ocellaris said:
Furmark doesn't circumvent the card's design specifications, it just works the card hard. The cards should be able to hand situations where the VRMs are maxed out.
Click to expand...

Such situations only exist during synthetic tests designed specifically to do that. Real world use never even gets close to that, ever.

Cooling solutions are not designed to handle that kind of load, period. No card is. Plain and simple.

Especially when overclocked. Redlining an overclocked card for hours using a synthetic test specifically designed for max stress is a recipe for a dead card.
 
I didn't actually run Furmark for hours. I ran it in 5 minute segments over a couple of days as I fiddled with settings. I did not leave it on consistently for hours, which is why i am skeptical that the card failed due to running Furmark. Every review I read of this card has said that it overclocks very well, and the card never exceeded 60c even in Furmark, which means actual gaming had lower temps. Now I am thinking it was just a bad card. I read around a lot and haven't found too many people complaining about dead cards from overclocking without touching voltage. Just makes no sense to me. Also, the card did not fail while I was running Furmark but the day after while doing light gaming. Could it just be that it was a bad card that would have died anyway? is there a way to actually monitor VRM temp? GPU-Z only shows one GPU temperature
 
Ok so I think these are just bad cards, which makes sense why it's $150 after rebate on newegg..

My replacement card just died. I just got it today. I only bumped the core from 860 to 950 and left memory as is. No Furmark this time. Just Far Cry 3 and Guild Wars 2. My machine shut down and will not boot up (monitor loses signal after 5 seconds).

I really can't believe the 2nd card would fail again after such a minor OC. I have a very cool system, and the temperature did not increase after the OC. What the hell? I'm starting to think it's something with my system because what are the chances of receiving two bad cards? I've had a Geforce 550ti in this system for a week while waiting for replacement and have had no issues. Ugh. Newegg is giving me a full refund (thank the lord), and I guess I'll go with another company. I found a Visiontek 7850 2GB for $200. Not bad!
 
Last edited:
From your first post it sounded like you left furmark on for 2 hrs. If your card has sensors on vrm you can see the temps on gpuz. It will be below your gpu temp if you scroll down.
Its sounding like you may have a bad power supply or something. I would try another power supply or try card in another computer.
 
455olds said:
From your first post it sounded like you left furmark on for 2 hrs. If your card has sensors on vrm you can see the temps on gpuz. It will be below your gpu temp if you scroll down.
Its sounding like you may have a bad power supply or something. I would try another power supply or try card in another computer.
Click to expand...

I used a GTX 550 Ti in this computer for a week and had no issues. It's a brand new PSU (OCZ). Well, I ordered an MSI 7850 so if that one dies, then I know my computer does not like Radeons.

The card actually only had core temp in GPU-Z. And yes i tried it in another computer.
 
"I overclocked the card, my machine shut down"
"They sent me a replacement card, and I immediately overclocked it. My machine shut down."

And the conclusion is..... "these are bad cards."
 
Hmm second card failed. thats very odd, some word of advice, before I OC anything, I tend to leave everything at stock for a week to make sure the everything is running smooth at stock. If I have issues with anything I can rule out the OC because we are at stock. Whats the difference between power draw from the 550ti to 7850?
 
kumquat said:
"I overclocked the card, my machine shut down"
"They sent me a replacement card, and I immediately overclocked it. My machine shut down."

And the conclusion is..... "these are bad cards."
Click to expand...

It's normal to get 30% OC on these cards. They're actually advertised for their overclockability. I highly doubt it was the minimal OC that did it, but who knows. I should've retrospectively taken martinje's advice and let it run stock for a week
 
kumquat said:
"I overclocked the card, my machine shut down"
"They sent me a replacement card, and I immediately overclocked it. My machine shut down."

And the conclusion is..... "these are bad cards."
Click to expand...

Tons of people buy 7XXX series GPUs for overclockability
A minute amount report issues

And the conclusion is.... "the overclocking did it."
 
You must log in or register to reply here.
Back
Top