[SOLVED] will overclocking a GPU fry your motherboard, CPU, PSU, explode the computer in smoke? void warranty of the graphics card or any other component?

antonis mark

Prominent
Jan 21, 2021
108
9
595
never overclocked before, so what warranties will be voided? will my CPU or motherboard die from this? planned to overclock my GT 1030 from time to time since its IMPOSSIBLE for the time being to purchase a new GPU! (the GTX 1050 TI HAS DOUBLED ITS PRICED AND EVEN HIGHER along with other graphics card) .will the GPU die and not covered by warranty?
 
Solution
never overclocked before, so what warranties will be voided? will my CPU or motherboard die from this? planned to overclock my GT 1030 from time to time since its IMPOSSIBLE for the time being to purchase a new GPU! (the GTX 1050 TI HAS DOUBLED ITS PRICED AND EVEN HIGHER along with other graphics card) .will the GPU die and not covered by warranty?
Look what you started LOL almost a cat fight.

Their really no reason to overclock a 1030 video card if you got 10% and that is a if you could go from 50 to 55 FPS just as a example. Not really enough to stress over.

Their also a chance you can go to the point of no return and brick the card.

Eximo

Titan
Ambassador
Not entirely risk free. Putting additional voltage and power through components can shorten their lifespan.

Really only void the warranty through physical damage or tampering (that is itself debatable, and is the core of the right to repair movement) Warranty void if removed stickers are effectively suggestions, though no one is willing to spend the money to take large companies to court over such relatively small sums when they do reject warranty claims.

As for exploding, no, pretty much impossible. Late model GPUs especially are already limited at the factory, they basically can't exceed power/temperature limits. CPUs are a little more easy to break, but you have to go to extremes with core voltage. The CPU will still thermal and power throttle, but you can overheat them rapidly with too much voltage, and that isn't good for them.
 

OtapIc

Commendable
Jan 19, 2021
63
7
1,545
CPU or Motherboard explosion? No
Damaging your GPU? Most likely will happen
As far as I know the GTX 1030 models only come with passive cooling, and the temps is what I'm more worried about. Unless you got some beefy cooling in your case, the 1030's passive cooling won't be enough to "safely" overclock
 
never overclocked before, so what warranties will be voided? will my CPU or motherboard die from this? planned to overclock my GT 1030 from time to time since its IMPOSSIBLE for the time being to purchase a new GPU! (the GTX 1050 TI HAS DOUBLED ITS PRICED AND EVEN HIGHER along with other graphics card) .will the GPU die and not covered by warranty?
No warranty I know of remains in effect when overclocking. Not even when using utilities the manufacturer provides to do it with, as AMD warns us in both Ryzenmaster (for CPU) and Radeon Adrenaline (for GPU).

You do it at your own risk. Bear in mind many GPU manufacturers sell 'overclocked' cards. They bear the warranty responsibility though, but you have to know they don't do it if the risk is too great.
 

Eximo

Titan
Ambassador
CPU or Motherboard explosion? No
Damaging your GPU? Most likely will happen
As far as I know the GTX 1030 models only come with passive cooling, and the temps is what I'm more worried about. Unless you got some beefy cooling in your case, the 1030's passive cooling won't be enough to "safely" overclock

Not all GT 1030 are passive, many have fans. Though, yes, typically quite small heatsinks. However, this is not a high power part, and its power limits are very low. It will not be damaged using the typical MSI Afterburner or EVGA Precision tools. It will run a little hotter, and maybe crash, but that is just fine tuning as you do in any overclock.
 

Eximo

Titan
Ambassador
No warranty I know of remains in effect when overclocking. Not even when using utilities the manufacturer provides to do it with, as AMD warns us in both Ryzenmaster (for CPU) and Radeon Adrenaline (for GPU).

You do it at your own risk. Bear in mind many GPU manufacturers sell 'overclocked' cards. They bear the warranty responsibility though, but you have to know they don't do it if the risk is too great.

They have no way of knowing you did it. That is simply their policy, it is not a guarantee of denial of warranty replacement or repair. Most typically if you request an RMA and the card/component has no signs of physical damage, and you are within the warranty period, they will honor it. Not like the card stores the boost clocks or power targets it has been running at.

In many countries, they basically have to honor the warranty with 6 months or 1 year as part of consumer protection laws.

EVGA has a nice policy as well, they allow you to dismantle the card. As long as it is returned reassembled, with no signs of physical damage, they cover it. Handy for those of us that water cool.
 
They have no way of knowing you did it. That is simply their policy, it is not a guarantee of denial of warranty replacement or repair....
True enough, although not always. But it IS a guarantee they have the right to deny warranty. All they really need is their judgement that it was overclocked. In the case of a GPU certain damage to VRM sections in particular could be considered evidence of overclocking and...DENIED.

But here's the question: in a supply constrained market does even the manufacturer have an unlimited supply of parts sitting around to throw out to all the over-zealous overclockers? So, are you certain you'll get a warranty claim approved when you overclocked? It's the risk you take.

A funny thing about CPU's...Ryzen's in particular and probably all, even Intel's. They don't really 'pop', or fail completely, all that often when overclocked too far. They degrade, requiring ever more voltage or lower clock to remain stable and useable. I suspect you can keep that up for 3 years at least; overclocking too far and reducing clocks and increasing voltage bit by bit until it won't even function at stock clocks reliably. By then the warranty's expired.

EVGA's 'nice' policy is cute and...irrelevant. We have the 'right to repair' (which obviously entails dissassembly) anything we own in the USA now. But it comes with the important caveat: ham-fisted attempts to 'repair' that actually end up damaging it will void warranty. A card returned disassembled, or poorly re-assembled, obviously isn't going to cut it.

So we have full rights to take heatsinks off of GPU's and re-apply thermal paste. But be careful because a chipped die or scratched card trace will void warranty. They still put tell-tales, on screws among others, so they will know you disassembled the card.
 
Last edited:

Eximo

Titan
Ambassador
I'm not sure you can tell a failed VRM vs a VRM that failed due to overclocking. That would indicate a weak link was somewhere to begin with. The VRMs should self regulate and protect themselves from damage as well. (To be honest, overclocking isn't as fun as it used to be unless you start modifying the card or vBIOS) In most cases you are just playing within the pre-defined limits AMD or Nvidia has set.

It is very difficult to break a modern piece of silicon, that is the point I am driving at. They have built in protections. CPUs are the easiest to damage since you still have control of the core and other voltages. But as soon as you break a temperature or power threshold they throttle within milliseconds.

The telltales themselves are in dubious territory. EVGA is one of the few that stopped putting them on at all. Warranty doesn't imply that they will necessarily give you the same equipment. Often times they will run out of the hardware in question. In which case you can receive better or newer models. They are basically committing to doing 'something'. If that is repair, they will repair it. If that is replacement, they will replace it with what is on hand. Including refurbished/scratch and dent stock. Typically they do not go backwards if an exact equivalent isn't available. But I have seen people post about getting back 'close enough' like getting an ACX 3.0 instead of a Classified after warranty, but they did get a working GPU. These companies all offer relatively the same thing with a fairly narrow profit margin, so they really do trade on their reputation to a large extent.

Right to repair itself is variable depending on where you are and how the hardware/software is licensed. It is slowly getting better as more and more politicians are getting pressure from their constituents and the lobbyist's arguments make less and less sense. (Mostly brand image and safety concerns)

Companies like Apple are just fighting the supply chain and making it impossible to buy replacement parts and making third party components incompatible. A lot of that they claim under the security and privacy of devices, but really it is just to keep their customers coming back to them.
 
I'm not sure you can tell a failed VRM vs a VRM that failed due to overclocking. That would indicate a weak link was somewhere to begin with. The VRMs should self regulate and protect themselves from damage as well. (To be honest, overclocking isn't as fun as it used to be unless you start modifying the card or vBIOS) In most cases you are just playing within the pre-defined limits AMD or Nvidia has set.

It is very difficult to break a modern piece of silicon, that is the point I am driving at. They have built in protections. CPUs are the easiest to damage since you still have control of the core and other voltages. But as soon as you break a temperature or power threshold they throttle within milliseconds.

The telltales themselves are in dubious territory. EVGA is one of the few that stopped putting them on at all. Warranty doesn't imply that they will necessarily give you the same equipment. Often times they will run out of the hardware in question. In which case you can receive better or newer models. They are basically committing to doing 'something'. If that is repair, they will repair it. If that is replacement, they will replace it with what is on hand. Including refurbished/scratch and dent stock. Typically they do not go backwards if an exact equivalent isn't available. But I have seen people post about getting back 'close enough' like getting an ACX 3.0 instead of a Classified after warranty, but they did get a working GPU. These companies all offer relatively the same thing with a fairly narrow profit margin, so they really do trade on their reputation to a large extent.

Right to repair itself is variable depending on where you are and how the hardware/software is licensed. It is slowly getting better as more and more politicians are getting pressure from their constituents and the lobbyist's arguments make less and less sense. (Mostly brand image and safety concerns)

Companies like Apple are just fighting the supply chain and making it impossible to buy replacement parts and making third party components incompatible. A lot of that they claim under the security and privacy of devices, but really it is just to keep their customers coming back to them.

In the US right to repair is everywhere. Many other jurisdictions in the world it's variable; in some opening an electric appliance can be flat out illegal (in Australia for instance). I'm sure it depends on the appliance of course...point is I'm aware it varies.

And yes, you can make a judgment on VRM damage. FET's, for instance, are extremely robust and don't generally burn up for no reason. When they do there's going to be a trend line established that will be revealed in RMA data and confirmed with lab testing. In the absence of such trends, an isolated case can be considered extraordinary and quite likely blamed on excessive overclocking. You'll also know by the state of the underlying PWB: you know what temp is required to bring on glassification, for instance, so if that's started you'll know the FET's have been running a long time at temps way higher than stock settings could induce.

That they don't deny so often is just good business relations because it IS so rare. But again...it is a risk so advising people 'they NEVER deny' is not smart IMO.

The tell-tales aren't always as obvious as a sticker on a screw, and I'm not saying that a broken tell-tale of itself gives cause to deny. It just gives them credence when they say YOU opened it up so YOU caused the internal damage we see. And yes, some mfr's may not use tell-tales but that won't stop them from simply saying "Our QA process is too good to let that scratched trace go out the door. You must have done it so DENIED." And again, if they don't deny coverage that often doesn't mean they lack the option or won't exercise it in YOUR individual case.

Apple makes it really easy to break things you try to dissassemble. But they are also a poster child for customer-hostile tactics and arrogance, along with Intel. I never buy either Apple or Intel for that reason alone.
 

Eximo

Titan
Ambassador
I mentioned that they do reject claims in my first post. I have not stated that they won't reject claims. Just saying that 'overclocking' using common methods will not result in a warranty denial, unless you outright say "I was overclocking and it broke"

Overclocking using MSI Afterburner is not going to leave any visible signs, you are within their pre-set limits on power and temperature. It may crash, and you may need to reduce or undo your overclock. Chances of outright failure are minimal.

vBIOS modding/overwriting, shunt modding, soldering on your own power delivery, exposing the card to liquid nitrogen, lapping, and other physical modifications will void the warranty. On that first one, not sure if the card retains number of flashes, but it might. And they may also keep track of the vBIOS it left that factory with (though you could put the old one back on)
 
....
Overclocking using MSI Afterburner is not going to leave any visible signs, you are within their pre-set limits on power and temperature. It may crash, and you may need to reduce or undo your overclock. Chances of outright failure are minimal.
...
Really?? I though AB had the ability to adjust power limits too...in addition to opening voltage limiters. Except on AMD cards, I suppose, which need additional tools in many cases (Power Play Table tool).

I use Radeon Settings so don't know these things about AB.

And yah, I've turned my 5700 into a 5700XT with vBIOS. It works great, but more than once it's entered my mind that if it pops there's no way to change the BIOS back to the original. Even used, this is a $650 dollar card on eBay right now.

I don't like giving glib assurances about overclocking, and not just concerning results. Any one doing it assumes the risk so if not prepared to deal with consequences don't do it. It's that simple.
 
Last edited:

Eximo

Titan
Ambassador
Yes, you can adjust the power limits in Afterburner, but you can't just keep increasing it to point of failure. On Nvidia cards particularly you are usually limited to something like 115% on the top end. And that depends on the base number the vBIOS has in it, so sometimes you will see 108%, but that is because the card has a higher preset floor. The ceiling is defined by Nvidia at the factory. You need modified vBIOS to exceed these values. or do shunt mods.

Pretty much since Maxwell you have been unable to modify voltage beyond some token increases, and even then, it may only have been visual and not actually changed the card. On newer cards, they don't even show you voltage, only power.

Typical Nvidia overclocking today:

Max power limit
Max temperature limit
Max fans (or what you consider tolerable)
Add boost until it becomes unstable, note
Add memory boost until it becomes unstable, note
Try your two maximums together, if it crashes, reduce one or the other. But one at a time. Compare some benchmark results online, while a card may appear stable, recoverable memory errors can occur and reduce performance. (GDDR6 and up also have ECC built in, so they don't tend to crash, just go really slow)
Once you have confirmed the GPU is stable you have reached maximum performance.

As always, monitor temperatures and keep it below 90C or so. late model Nvidia likes to be as cool as possible though and the default throttle point on many cards is 83C, though you will also see 73C on some cards. These usually have beefy coolers and they were aiming for silence.

Just remember you are only really messing with a boost target, the actual results will generally not be that. So it will bounce around your target, and probably never reach it.
 

Karadjgne

Titan
Ambassador
I oc'd everything back in the day, had a 1GHz OC on my 3770k and 124/128% OC on my 660ti and 970.

And then upgraded everything and realized there wasn't much point. Ryzens boost themselves, Intel high core/high speed boosts, gpu boost 3 does a respectable job, and I can't blame Dell/HP attitudes of if you wanted better performance, buy the upgraded model. So if you want max fps out of a 3070, don't buy the Evga Black, buy the Evga FTW3 instead and get the performance you were seeking in the first place, without the need for pushing OC, and still being fully covered under all applicable warranties.
 

Zerk2012

Titan
Ambassador
never overclocked before, so what warranties will be voided? will my CPU or motherboard die from this? planned to overclock my GT 1030 from time to time since its IMPOSSIBLE for the time being to purchase a new GPU! (the GTX 1050 TI HAS DOUBLED ITS PRICED AND EVEN HIGHER along with other graphics card) .will the GPU die and not covered by warranty?
Look what you started LOL almost a cat fight.

Their really no reason to overclock a 1030 video card if you got 10% and that is a if you could go from 50 to 55 FPS just as a example. Not really enough to stress over.

Their also a chance you can go to the point of no return and brick the card.
 
Solution
I oc'd everything back in the day, had a 1GHz OC on my 3770k and 124/128% OC on my 660ti and 970.

And then upgraded everything and realized there wasn't much point. Ryzens boost themselves, Intel high core/high speed boosts, gpu boost 3 does a respectable job, and I can't blame Dell/HP attitudes of if you wanted better performance, buy the upgraded model. So if you want max fps out of a 3070, don't buy the Evga Black, buy the Evga FTW3 instead and get the performance you were seeking in the first place, without the need for pushing OC, and still being fully covered under all applicable warranties.

Intel overclock is still worthy though, i find the turbo boost too limited, plus not on all cores. There's still a hefty performance to gain, about 10% or so, but still there, though its true that older models benefit a bit more... i still remember oc my pentium 3 from 733 to 1066, thats almost a whopping 50%, mostly that is.
 

antonis mark

Prominent
Jan 21, 2021
108
9
595
Intel overclock is still worthy though, i find the turbo boost too limited, plus not on all cores. There's still a hefty performance to gain, about 10% or so, but still there, though its true that older models benefit a bit more... i still remember oc my pentium 3 from 733 to 1066, thats almost a whopping 50%, mostly that is.
I use i5 9400f for the new PC and i5 2400 for my old one, none of them are k and have locked multipliers meaning I can't overclock them
 
Last edited: