Maximizing your GPU life?

[FaceToKeyboard]

I came across a post (Sorry, still trying to find it) about how AMD GPUs should be dropping in price soon thanks to some anticipated ASIC mining hardware, and there was a comment about how one should be careful buying GPUs used to mine because of the constant use and heat.

The next comment stated that it's actually a gamer's card that would be under more stress - the repeated cycles of heavy load (gaming), then instant idling (stopped gaming) cause expansion-contraction cycles with the solder on the card, reducing their life.

My question is
1) is this true?

and

2) In general, is there anything else I can do to minimize the stress to my graphics cards and maximize their life besides not gaming? Maybe underclock the GPU and decrease voltage levels unless I notice dropped frames?

 

[nurturedhate]

The best thing you can do is keep it clean and supply adequate airflow for it. A well made video card should last for far longer than most people would use it. I still have cards from the late 90s and early 00s that would work just fine if they were put into a system. I still have friends and family using cards that are at least 5 years old with no signs of issues. Thinking about it I know one friend who is still using my old ATI 9800pro in his kid's machine.

As for your questions; yes heating and cooling cycles affect a cards lifespan but usually (99% of the time) this is not an issue as the cards have been engineered to withstand this. This is where my suggestion to keep your card clean comes in as it minimizes the temp difference and allows the card to work as it should.

As for the recommendation of not buying mining cards I always looked at it more as an issue with the fans. They have been beat hard running 24/7 like that. Also, you have no guarantee as to how they were treated. We have no carfax for computer components.

In the end, keep your stuff clean and they should last you far far longer than they would be useful.

 

[FaceToKeyboard]

I always looked at it more as an issue with the fans. They have been beat hard running 24/7 like that.

That's a great point, I haven't even considered the fans.

Looks like I need to finally invest in some sort of convenient air blower or vacuum cleaner.

 

[PliotronX]

First rule of thumb, even if you stick with stock cooling, replace the thermal interface material with something decent and if the GPU silicon has a shim or heatspreader look into removing it with a razor blade. Then adding to nurturedhate's input, focusing constant airflow onto the voltage regulation is another aspect I'd look into.

 

[bononos]

That's a great point, I haven't even considered the fans.

Looks like I need to finally invest in some sort of convenient air blower or vacuum cleaner.

Get air filters on every input event. Saved me the hassle of frequent cleaning to a few wipes a year.
And I don't think its the expansion-contraction cycles which kills cards but a fault in the manufacturing process like the Nvidia solder bump problem in 2008-2010.

 

[PliotronX]

Get air filters on every input event. Saved me the hassle of frequent cleaning to a few wipes a year.
And I don't think its the expansion-contraction cycles which kills cards but a fault in the manufacturing process like the Nvidia solder bump problem in 2008-2010.

Another good point, dust kills! This is how I roll, available at most retail outlets

It was kinda funny the other day when I was replenishing my supply of pantyhose for use as case fan filtration, I also picked up a box of profilactics. It dawned on me the irony of the combination of pantyhose that could be used for facial obscuration and condoms for committing a rape
https://www.youtube.com/watch?v=GH1CsgADwBY

 

[IEC]

I wouldn't hesitate to buy a card used for mining provided its fans are still good. Unless a GPU is defective its going to outlive its usefulness anyways.

I've run things like F@H since they first came out on GPUs many years ago and *never* had a GPU die on me.

 

[BF4~GaM3r]

Keep it clean,proper airflow and reapply Dynex Silver Thermal Compound(or any good kind)six months to yearly -after cleaning the old off with gauze pad with lil' rubbing alcohol.
Votes-up all answers.

 

[Headfoot]

The only thing that will materially affect the life of your card is running it with lots of volts, or letting it hit really high temperatures or both. And even then, most cards can do that for a long time

 

[blackened23]

I honestly wouldn't worry about it. GPUs are pretty solid these days in terms of reliability, the big thing as mentioned above is over-voltage. Over voltage causes electromigration which can affect things, but it isn't overnight. Usually that process of electromigration causing problem depends on the extent of over-voltage and how long you do it. Once in a blue moon, probably not a big deal. 24/7 for two years? Who knows. I have had CPUs conk out after extreme over-voltage levels for 18 months with high overclocks. But that was on me for undershooting the cooling and over-volting way too high in 24/7 operated systems. To be fair, this was before offset voltage was a thing. It was manual over-voltage so that high voltage stuck even during idle speeds, which I was stupid to do in the first place. It was an i7-870 that I had conk out this way.

Just to re-iterate: you'd have to do extreme over-voltage 24/7 for an extended period of time to have issues. Then again, some people use over voltage and never have issues. The other counter argument being, since GPUs use offset voltage now (which wasn't the case in the past) that will help reliability even if you're over-volting. And that is also true. So: do not worry about it.

Now most people aren't over-volting. Back to your question. Seriously, don't worry about it. I'd say once every 8 months to 1 year, open up your GPU shroud and remove the dust. That's actually probably a generous estimate, you probably don't need to do it yearly. Unless you're like, chain smoking cigarettes right beside your PC or something.

GPUs are built pretty solidly these days, I believe.