Chances of a graphics card dying if always running?

[ayqazi]

Hello,

I have jumped onto the bitcoin mining craze, but only with the view of earning back how much I spent on my graphics card within a few months.

However, I'm concerned that, with the card keeping at around about 75-80C in temperature at all times, it and the system may suffer a lose of lifespan.

I'm quite interested in overclocking it, was wondering what the thoughts of the forum were with regards to the chance of it dying? Overclocking voids the warranty, which I would rather not do (and I don't intend to lie to the manufacturers if it does die while overclocked to get an RMA anyway).

Thanks. Regards,
Asfand

[peterb]

If you are overclocking andor over-volting any component, you are adding addtional stress which could shorten the lifespan of the component in question.

As for voiding the warranty... The warranty states that it will operate to the published specification for a given period of time. Obviously if you try to exceed that specificatio (but it wont operate at the speed) you can't claim that it doesn't work at that hgher speed because it wasn't guranteed.

As fr the life - the warranty probably states that it is void if it is subject to misuse - and it could be argued that overcocking/overvolting is misuse, so premature failure under those conditions might not be covered.

Is the increased risk of prmature failure worth the additional speculative return bitmining? Only you can judge that. The chances are that it won't significantly reduce the life - but you might be unlucky...

[Fraz]

Surely if it breaks under warranty when all you've done is bitcoin mining with a card at stock, then you just get it replaced under warranty?

It annoys me how many people seem to be taking up bitcoin mining. It's just burning precious resources for a small monetary gain. I see it as no different from buying some coal at £1 and being told that if I burn that coal someone will give me £2. Still, when did the environment ever get in the way of people's greed.

[shaithis]

Hardware prefers to be on or off. You will find that most hardware will die as the machine boots up from cold.

Of course there is a MTBF on all hardware but it's pretty hard to hit those numbers before the hardware is massively outdated anyway.

Overclocking will bring the MTBF figure down, that's a certainty. Depending on an absurd amount of factors, that MTBF could drop by a bit or completely plummet.......

[Agent]

It's just burning precious resources for a small monetary gain. I see it as no different from buying some coal at £1 and being told that if I burn that coal someone will give me £2. Still, when did the environment ever get in the way of people's greed.

It might surprise you to know that many people do Bitcoin at a loss.

I know it's been covered in other threads, and you make your dislike for Bitcoin well known ....but the purpose of Bitcoin as a project isn't to allow people to 'make money for nothing'. It's to establish an alternative currency that is controlled by the people involved, not the banks or governments.
It's also almost impossible to create coins fraudulently to, which is a huge advantage over current 'real world' currency.

Bitcoin, as a idea is actually very very good. Granted, the way of ensuring the integrity of the currency uses up a bit of computing power, but it's a necessary evil in it's current form. I'm saying that assuming that you realise Bitcoin isn't 'calculations for nothing' as so many people seem to think

[watercooled]

Much as shaithis says, hardware prefers to be on or off, or more specifically stay at a constant temperature to prevent thermal stress. Cycling between hot and cold many times can cause cracks in materials, for example this is one of the main causes for the 360 RRoD, the die solder joints would crack away from the substrate, and this also explains why the 'x-clamp' often worked to fix or prevent it, it applied more pressure between the surfaces to keep them together.

However, this is provided it's kept below the specified temperature range obviously. And electrolytic capacitors (VRM components) can degrade over time and faster at higher load/temperature, however most graphics cards on the marked use polymer capacitors which have a much longer lifespan so it shouldn't be a concern.

As for overclocking, I wouldn't do it on a stock card - the VRM is designed to power that GPU at TDP and will even throttle down the GPU if load gets too high on newer cards (HD 5000 series and over). Look at what Furmark did to some VRMs on previous generations of cards without this protection. This probably isn't covered by warranty. While there's still a chance it will cause problems, it should be much safer on a card designed for overclocking (better cooling/beefier VRM/overclocking controls).

[ayqazi]

If you are overclocking andor over-volting any component, you are adding addtional stress which could shorten the lifespan of the component in question.

As for voiding the warranty... The warranty states that it will operate to the published specification for a given period of time. Obviously if you try to exceed that specificatio (but it wont operate at the speed) you can't claim that it doesn't work at that hgher speed because it wasn't guranteed.

As fr the life - the warranty probably states that it is void if it is subject to misuse - and it could be argued that overcocking/overvolting is misuse, so premature failure under those conditions might not be covered.

Is the increased risk of prmature failure worth the additional speculative return bitmining? Only you can judge that. The chances are that it won't significantly reduce the life - but you might be unlucky...

Nice points on warranties there, thanks

Hardware prefers to be on or off. You will find that most hardware will die as the machine boots up from cold.

Of course there is a MTBF on all hardware but it's pretty hard to hit those numbers before the hardware is massively outdated anyway.

Overclocking will bring the MTBF figure down, that's a certainty. Depending on an absurd amount of factors, that MTBF could drop by a bit or completely plummet.......

Much as shaithis says, hardware prefers to be on or off, or more specifically stay at a constant temperature to prevent thermal stress. Cycling between hot and cold many times can cause cracks in materials, for example this is one of the main causes for the 360 RRoD, the die solder joints would crack away from the substrate, and this also explains why the 'x-clamp' often worked to fix or prevent it, it applied more pressure between the surfaces to keep them together.

However, this is provided it's kept below the specified temperature range obviously. And electrolytic capacitors (VRM components) can degrade over time and faster at higher load/temperature, however most graphics cards on the marked use polymer capacitors which have a much longer lifespan so it shouldn't be a concern.

That's very nice.... It's already been at a steady 75ish degrees C for a few weeks now at stock (well, with underclocked memory except when I play games).

As for overclocking, I wouldn't do it on a stock card - the VRM is designed to power that GPU at TDP and will even throttle down the GPU if load gets too high on newer cards (HD 5000 series and over). Look at what Furmark did to some VRMs on previous generations of cards without this protection. This probably isn't covered by warranty. While there's still a chance it will cause problems, it should be much safer on a card designed for overclocking (better cooling/beefier VRM/overclocking controls).

I have one of those silly custom-cooled dual-fan noisy things, so I think I'm covered on the cooling front. I just have to play with the fan speed to not make it so noisy so I can sleep

[c.ruel]

I have one HD 5850 which runs 24/7 around the 70-80 mark and another that I only mine with at night, (it's my main card during the day). Both so fair touch wood have been fine. I run no overclock btw.

I also have an old HD 4850 on the go as well with a non stock cooler that also seems to be fine.

[=assassin=]

I always thought restarting alot was more dangerous as more heat cycles (cooling down fully then heating up fully) can damage the solder or something. So keeping it on all the time at a set temperature, as long as it isn't above specification, shouldn't be too bad. I guess the fan is more likely to go on it though.

[fuddam]

I remember reading (on hexus?) some time ago that letting a laptop sleep is more destructive on the GPU longevity since it shuts off any cooling fans, but the GPU isn't properly asleep - this true / ring any bells?

Have had one laptop GPU die 3 times due to solder issues, and now on a replacement. Hoping the same won't happen, since is old gear and hard/expensive to replace, while laptop is still pretty good power-wise tc.

[watercooled]

I don't think it's a rule, rather poor design of specific laptops. But I think there was a problem with some mobile Nvidia chips failing, can't remember the exact details.

[1nsan1ty]

If you're worried about the increased temps over a long period of time then get an aftermarket cooler for your card. It'll also reduce the noise level making mining much more bearable