Why did my GPU get fried

Gordon.C

Member
Hi,

I will probably never know the right answer so this is more of a thought experiment.

Why did my GPU got friend.

It was a Radeon HD 4870x2 old a little over 3 years. To ensure longer life span I kept it slightly underclocked for about 2 and 3/4 years but close to the end of its life I left it at its factory clock (1. reason it could have fried).

I rearranged my table in a way that the computer was situated closely next to a wall. The wall of the case where motherboard is held was more or less blocked from fresh air (2. reason it could have fried).

3. reason is simple coincidence because the GPU fried while I was playing GTA San Andreas (for the first time on this GPU).

Which do you think is the most likely?
 
without a lot more information about it we will never be able to tell with any more certainty than a "psychic" with no computer knowledge.

1. I do not think that running it at stock speeds would cause it to fry. It could in some situations be part of the reason. If you have a blocked up HSF on it, or removed the heatsink without properly reapplying the thermal paste, that could cause a fatal overheat from it at stock speeds. But assuming proper operation, and proper power, just setting it back to stock would not kill it.

2. Are you talking about the panel against which the motherboard is bolted, or the one that allows access to the equipment inside? Either way, should not cause much trouble as most cases take air in the front and blow out the back. You can generally put the computer in a place where both side panels are in contact with a desk, or wall, or pretty much any non electrically charged surface.

3. I am not familiar with GTA:SA, but that is most likely just a coincidence. Could have just been that the game charged both chips up to max and caused a failure.

Without knowing the exact mode of failure we can not say what could have caused it. Just take guesses at it.
 
To ensure longer life span I kept it slightly underclocked for about 2 and 3/4 years

I just wanted to point this out that this is not a good idea and will not in any way ensure a longer life span.

Underclocking processors can result in a lot of problems. They were designed to run at the factory clock settings, not any higher or lower. When you change it, including lowering it, you are running outside of what it was originally designed to do.

Believe me, I tried doing the same thing years ago to keep my old 6800 Ultra at a really low idle temperature while at the desktop and by lowering the clock about 50MHz it started to give me artifacts on the desktop.

The best thing to do if you want to ensure a longer life span for your processors is to leave them at their factory settings.
 
without a lot more information about it we will never be able to tell with any more certainty than a "psychic" with no computer knowledge.

1. I do not think that running it at stock speeds would cause it to fry. It could in some situations be part of the reason. If you have a blocked up HSF on it, or removed the heatsink without properly reapplying the thermal paste, that could cause a fatal overheat from it at stock speeds. But assuming proper operation, and proper power, just setting it back to stock would not kill it.

2. Are you talking about the panel against which the motherboard is bolted, or the one that allows access to the equipment inside? Either way, should not cause much trouble as most cases take air in the front and blow out the back. You can generally put the computer in a place where both side panels are in contact with a desk, or wall, or pretty much any non electrically charged surface.

3. I am not familiar with GTA:SA, but that is most likely just a coincidence. Could have just been that the game charged both chips up to max and caused a failure.

Without knowing the exact mode of failure we can not say what could have caused it. Just take guesses at it.

I underclocked the GPU because a long time ago I swapped the stock fan with a new one which was far more powerful while being silent.

I noticed through GPU-Z that the VDDCs on the GPU were getting over 100C so I experimented and underclocked it about 150MHz below stock clock (that was an optimal underclock which did not overheat the GPU and the games still ran very smoothly).

As I said the last quarter of a year of its life I simply put it back to its original clock and did not care if it fries or not. It did :D

I just wanted to point this out that this is not a good idea and will not in any way ensure a longer life span.

Underclocking processors can result in a lot of problems. They were designed to run at the factory clock settings, not any higher or lower. When you change it, including lowering it, you are running outside of what it was originally designed to do.

Believe me, I tried doing the same thing years ago to keep my old 6800 Ultra at a really low idle temperature while at the desktop and by lowering the clock about 50MHz it started to give me artifacts on the desktop.

The best thing to do if you want to ensure a longer life span for your processors is to leave them at their factory settings.

Through the ATI Catalyst I saw that the clock of the GPU varied hugely between 250MHz (when Idle) and 900Mhz (when in full screen). It is probably something like Intel has the Turbo Boost feature isnt it.

The GPU has worked perfectly while underclocked for about 2 years. Though you are right that I would not recommend underclocking a GPU that does not have underclocking and overclocking feature as a native tool.
 
I noticed through GPU-Z that the VDDCs on the GPU were getting over 100C so I experimented and underclocked it about 150MHz below stock clock (that was an optimal underclock which did not overheat the GPU and the games still ran very smoothly).

If your card was overheating (I would consider over 100c pretty hot considering my G80 only ran at about 85c max and I believe about 120c is the threshold before most cards fry) then underclocking it should not be the answer. If its overheating at stock speeds, there is another problem.

But to answer your original question, sounds like it just overheated.
 
But to answer your original question, sounds like it just overheated.

Most probably yes... I just hope the problem wasnt lack of fresh air, because at the moment the PC could have had less air flowing in since once side (the side to which motherboard is bolted) is covered by table.
 
Graphics cards can be very tricky. I actually bured one two weeks after i stressed it. I played Call of Juares . It lagged a bit, but i managed to finish the game. Two weeks later, it burned out of the blue (no big stress since call of juares. )..
 
Most probably yes... I just hope the problem wasnt lack of fresh air, because at the moment the PC could have had less air flowing in since once side (the side to which motherboard is bolted) is covered by table.

The panel that the motherboard is bolted to is not usually ventilated so I doubt that played any significant role in over heating. Even if both panels were covered it shouldnt have caused any significant heat issues as long as there are fans in the front and back to pull air in from the front and blow it out that back. That is how my case is designed at least.
 
What power supply are you using?

I have been using since beginning a 850W power supply. Never experienced any issues concerning power.

The panel that the motherboard is bolted to is not usually ventilated so I doubt that played any significant role in over heating. Even if both panels were covered it shouldnt have caused any significant heat issues as long as there are fans in the front and back to pull air in from the front and blow it out that back. That is how my case is designed at least.

I see... Well lets rule this option out then because the case was well ventilated from the front, back and one side of the case. I believe the GPU was covered from each side by 3 ventilators :-) just to be safe... It worked amazingly for over 3 years after all, time to move on and find a new beauty

Graphics cards can be very tricky. I actually bured one two weeks after i stressed it. I played Call of Juares . It lagged a bit, but i managed to finish the game. Two weeks later, it burned out of the blue (no big stress since call of juares. )..

Tell me about it... As I mentioned I burnt mine during playing GTA SA for the first time on this PC setup. The GPU was caput after about 10 mins of gameplay, how wicked is that :D
 
What make and model of 850w power supply?

Sorry I dont remember. It has been however running very smoothly delivering power to Radeon HD4870x2 and Core i7 920 for over 3 years so I believe the quality is fine of the power supply.

I definitely has 3 more years of life ahead of it.
 
Eunning parts at lower frequency will make them run cooler and has no adverse effect beyond performance. Keeping dust out of the HSF and cooler air in the case are the other things you can do to maximize life. All that said, video cards like the 4870X2 run pretty hot, and have the disadvantage of the air passing over one hot GPU before it gets to the next.
 
Eunning parts at lower frequency will make them run cooler and has no adverse effect beyond performance. Keeping dust out of the HSF and cooler air in the case are the other things you can do to maximize life. All that said, video cards like the 4870X2 run pretty hot, and have the disadvantage of the air passing over one hot GPU before it gets to the next.

I used to have the reference fan for about a year and I though I was going to go deaf from it.

After about a year I installed a 3 fan cooler which made it run very quiet however for some reason the VDDC parts of the GPU kept overheating, going above 100C. I had to underclock the GPU to keep all temps below 80C at full load.

Close to the end of the GPUs life I clocked it back to its original clock (did not care what was gonna happen to it) and it lived for about 5 months before it fried.

When I took the GPU appart, there was no apparent damage really. The GPU as well as the cooler seemed completely intact.
 
Back
Top