What aspects of overclocking can damage a gpu?

[BeCoolBro]

When overclocking anything,I understand there is a chance that you may cause damage to the part you're trying to oc.When ocing temp rise due to the parts working at higher than stock frequencies and if you up the voltages that increases them too.So a very high voltage or too high temps can damage a graphics card.But are there other issues that could appear and cause damage to a gpu except temps and voltages?I'm asking because I want to oc my 5850 and I want to understand whether the gains are worth the risks;p.

 

[sportsfanboy]

Excessive heat and voltage are really the only things that can significantly lower the life span of an overclocked computer part.

 

[BeCoolBro]

Excessive heat and voltage are really the only things that can significantly lower the life span of an overclocked computer part.

So as long as I keep the temps in a low level and don't raise volts at all or by a bit no damage can occur by overclocking except a lower lifespan?

 

[sportsfanboy]

The rule of thumb is, keep the voltage below factory rated max value. And run the card at or below "safe temperature", and no significant damage will ensue. You'll have to look up what the max rated voltage is, as well as how hot you can run your card. Usually video cards can safely run in the 80's or low 90's Celsius will out any problems.

 

[4Ryan6]

So as long as I keep the temps in a low level and don't raise volts at all or by a bit no damage can occur by overclocking except a lower lifespan?

That's not necessarily true, even if you don't touch the voltage at all and push the core clock or memory clock too high, you can damage the card.

Been there, Done that!

 

[BeCoolBro]

That's not necessarily true, even if you don't touch the voltage at all and push the core clock or memory clock too high, you can damage the card.

Been there, Done that!

I'll try to keep a low profile then.

Spoiler

(I'd love to have a spare useless rig to experiment on this though)

 

[sportsfanboy]

You damaged your card by not upping the voltage and keeping temperatures cool? How is that possible?

 

[arthurh]

You damaged your card by not upping the voltage and keeping temperatures cool? How is that possible?

If you do more work then you make more heat, even if you did not have to raise the voltage to achieve that OC.

We have built-in sensors on our graphics cards that allow us to monitor the GPU temperature but "not" the memory chips and not everyone has the capabilities to monitor the memory chip temperatures.

So we know when we have OC'd our GPU too much because it will artifact in a bench. Memory chips on the other hand might be running at a OC that does not artifact during a bench but would in a particular spot during game play.

If not recoqnized for what it is and left to just run that way it is causing instability in the GPU that "could" lead to graphics card failure. Some of us ole timers have had that experience.

All of the older and some newer graphics cards I have seen with a HSF on them are designed to make excellent (sometimes just adequate) contact with the gpu. However "Thermal Pads" are used to make contact with the memory chips as they are designed with a slight gap between them and the memory chips so as to not interfear with the GPU contact. My assumption on that last remark as my 480 has a seperate GPU heat sink from the memory chips. Thou they still use what appears to be a thinner thermal pad with sticky on both sides I imagine to composate for the minute differences in height between all the memory chips. I do not OC this 480 'cause it plays all my games fine for me and my needs.

Newer graphic cards seem to have a better designed HSF unit than say the g92 8800 - 250 series grahic cards, i.e. I have 2 in SLI and had to replace them both serveral times (now 250s). I would get them in and the frist thing I did was check to make sure the HSF was on correctly as they tended not to make good contact with the memory chips.

I have a 9800 Pro that will only run at stock speeds because I OC'd it and kept it that way for a long time. I started getting artifacts and was forced to lower my OC on that card, eventually getting to stock and then it was not reliable. I removed the stock fan on its heat sink and super glued a 80mm fan over the opening where the stock fan was and that allowed me to run it at stock settings only. Also had seperate memory chip heat sinks on that particular card. Also had dedicated fans blowing on the back side of that card. All that just to keep from replacing that card which I have any way.

Guess I kinda got on a soap box there but the short story is "yes" it is possible to ruin a new graphics card with out a BIOS flash or using MSI Afterburner to OC with out raising the voltages. Probably not as easily as it used to be but nothing is failure proof if OCed.

 

[sportsfanboy]

I'm saying if you don't mess with the voltage, and you keep the temperatures in check, you should see no I'll effects. I've bee overclocking my cards since software has been available to do so, and have never had any problems.

 

[arthurh]

I'm saying if you don't mess with the voltage, and you keep the temperatures in check, you should see no I'll effects. I've bee overclocking my cards since software has been available to do so, and have never had any problems.

I agree with you but you wanted to know how it was possible to ruin a graphics card with out raising the voltage (which used to mean a BIOS flash) and 4ryan6 and I both have experienced this in the past as I'm sure many others here have as well.

I'm sure you'll agree that the same hardware OC'd by 2 different people does not mean identical results due to mfg tolerances amongst other variables as well.

The OP was concerned about this and you responded according to your experience.

We wanted to respond according to our experiences as well and have done so. All of us who OC have our limits as to how far we are willing to go. Some are happy with a 5% OC while some one else may not be happy til they have a 20% OC.

My response was not meant to dispute you but rather add information that hopefully helps the OP and your query.

 

[4Ryan6]

You damaged your card by not upping the voltage and keeping temperatures cool? How is that possible?

Actually I've damaged 3 cards in the past, the best way for me to describe my statement is simply it probably did have something to do with temperature but not the GPU temperature that was reporting everything was OK. All those cards had no way of reporting what the temperature of the memory was getting to, and increasing the memory clock was increasing the memory temperature. I thought all was well from my GPU temperature as it was in the green, but that was a false confidence as running the memory at higher clocks was increasing the memory temperatures, which didn't show up as a problem until the artifacting while gaming began.

I'm sharing my own personal experience here and just saying it is possible, to give the OP another point of view, I think it's great you've had no misfortune in any of your GPU overclocking, but if and when you do, you may end up being cautious too. My statement is not meant to discount any advice you've given, it's meant to add to your knowledge that damage can occur and has occurred to others.

Caution should be the rule of thumb when it comes to any overclocking, unless a person has so much money to blow, it doesn't matter if they risk hurting their new toy.

@arthurh, Great explanations there Art!

@sportsfanboy, Graphics card manufacturers are mounting much better heatsinks today that cover the memory chips better than in the past, the past best results I ever got was going with an aftermarket cooling solution that would also direct airflow downward to cool the memory and voltage regulators on the card.

My best GPU memory cooling results were from making my own memory heatsinks and Thermal Adhesive bonding them directly to the memory chips, [Definitely Voids Card Warranty], but very effective.

 

[BeCoolBro]

Got my 5850 to 850/1135 stable at stock volts.I probably could push it a lot more but it is more than enoughfor my needs now

 

[BeCoolBro]

Sorry for doubleposting but I see no need to start another thread to ask a little question about ocing.I have seen that if I increase the clocks beyond overdive limits powerplay gets disabled so I have more heat and power consumption.To counter this I have made 2 profiles in msi afterburner,one with stock clocks and one with the overclocked ones.I switch between the two depending on what I do.So would th constant overclocking and underclocking cause any damage or as long as I keep the temps in check othing will be affected?

 

[jackfairman98]

So as long as I keep the temps in a low level and don't raise volts at all or by a bit no damage can occur by overclocking except a lower lifespan?

That's not necessarily true, even if you don't touch the voltage at all and push the core clock or memory clock too high, you can damage the card.

Been there, Done that!

I was just messing around with my gtx 1060 3g in overclocking, got a bit bored/trigger happy and slammed it to +1000mhz core clock, the screen went a solid colour and stayed that way until i shut it down (like 5 secs)
had to uninstall and reinstall msi to get rid of that clock, now it seems fine again. i should also mention that was +100mv (the max for my card) with max watts too +108

 

[jackfairman98]

Sorry for doubleposting but I see no need to start another thread to ask a little question about ocing.I have seen that if I increase the clocks beyond overdive limits powerplay gets disabled so I have more heat and power consumption.To counter this I have made 2 profiles in msi afterburner,one with stock clocks and one with the overclocked ones.I switch between the two depending on what I do.So would th constant overclocking and underclocking cause any damage or as long as I keep the temps in check othing will be affected?

Depending on your card you can push harder with newer gens, especially the pascals, the pascals are amazing at OC'ing.
No overclocking and underclocking constantly will not damage it. The extra power/whatever from overclocking just slightly reduces lifespan for low to medium overclocks (anything that doesnt raise core VOLTAGE and doesn't crash or cause artifcating etc in test benches or under high load, you can also get away with low core voltage overclock (id say anything under 20-30 % or mv, the safe zone is around under 15mv/15% extra however)).

high overclocking (pushing the card to a point where it artifacts or crashes and/or with a medium to high overvoltage will cause a great reduction in the cards lifespan, if done propperly the card will last around 2-3 years minimum)