[SOLVED] Gaming's thermal cycling vs Mining's components degradation? Which is worse to GPU's lifespan?

[Secrutinize]

What title says.

I see a lot of miners saying thermal cycling is worse than mining's components degradation.
However mining gpus overclock memory forcing it to run at about 90c 24/7 no pause for months or even years.
That can't be good for components lifespan, such as capacitor and memory controller.

In your opinion which one is riskier, a gamer gpu that played some high end game several hrs a day or a mining gpu that ran at 90c 24/7, both were used for 1 or 2 years
If you were to choose one which one would you pick, and why?

 

[Phaaze88]

Aww geez...
There's some variables to this debate that don't make this a simple yes/no.

Taking care of yo 💩 , is one, and perhaps the most important. There seems to be more of a lazy streak with non-mining PCs; miners are more likely to keep their investments clean, but it still depends on the individual.
Sometimes I see pictures of PCs, where the user hasn't cleaned it in a year or more... yikes.
Dust is a silent killer. Like, come on, if you're going to be like that, maybe stick with a prebuilt; those companies at least expect you to buy a brand new one after some time.
I clean my PC about every 3-4 months, but I think I'm one of the more extreme examples.

Miners are more likely to undervolt and underclock(gpu core). That's a little less stress on the gpu's components.
Plus, not doing it is wasting power and slowing down their profits.
Some gamers go in the opposite direction via overclocking.

Excessive voltage, heat, and duration of, are what kill electronics - besides just breaking down and dying when their time is up; they can't last forever.
Heck, you can even apply some of the following to gpus - it's all silicon anyway:
[Taken from, https://www.tomshardware.com/features/inside-intels-secret-overclocking-lab/7]
"... the amount of time a processor stays in elevated temperature and voltage states has the biggest impact on lifespan.
...
"As an overclocker, if you manage these two [voltage and temperature], but especially think about 'time in state' or 'time at high voltage,' you can make your part last quite a while if you just think about that."
Probably best to read the whole thing, but I wanted to avoid posting a wall of text.]

So, I'm not sure there's really an argument for thermal cycling, except against people who are lazy with their PCs, which there are likely many...
Sure, underclocking and undervolting the card reduce stress across the card - it also indirectly reduces Vram temperatures, since their voltage can't be controlled.
Memory OC'ing does not appear to be as forgiving as core clock OC'ing. Once those telltale signs of failing memory start to show, there's not much to be done about it but try to downclock the memory. You're still on borrowed time though.
With mining gpus, the fans are the first to go, likely followed by memory. The gpu core and VRM would be good for a while yet.

FYI: 90C is still within spec for GDDR6 and R6X. The critical limit isn't until 105C.
https://www.micron.com/products/ultra-bandwidth-solutions/gddr6x

GDDR6 memory

Micron GDDR6 RAM meets the needs of increasing memory bandwidth requirements for high-powered computing, gaming and networking applications.

www.micron.com

GDDR5 and R5X are 95C:
https://www.micron.com/products/ultra-bandwidth-solutions/gddr5



To answer the question though: the mining card(s) is probably the safer bet, as they're more likely to have taken care of it.

 

[Secrutinize]

Thanks for the input.

So basically dust buildup is the most harmful to the cards because it raises both core and memory temperature as fans get old and dusty, thus degrading components much quicker...
I see your point but can't it be still applied to all miners as well?

I know a lot of miners (presumably) take care of their cards because longevity is what they're after.
But at the same time they're also after profits, not the longevity of the cards itself.
They simply need to pump up voltages of the memory quite a bit in order to gain optimal hash rates.

Yes miners may clean the dust or reapply pads/paste and whatnot, but the real problems are 'temperature' and 'duration' that components get exposed, as well as the voltages.

Usually gamers don't prefer core temps go skyrocketing(90c or above).
Same as miners since they also don't prefer memory temps going above 100c.
But gamers don't play games all day long and it's not like they actually play games every single day.
Whereas mining rigs run at 90c memory temp, possibly above that, for 24/7 without a rest every single day for months and years.
Some gamers are ignorant about the gpu temp but so do some miners as they don't care much about memory temp as long as it stays under 106c(throttle point).

And memory indeed is designed to withstand high heat(90c or even 100c depending on the types of memory), but it is not designed to operate at such high temp 24hrs for months and years.
Mining exhausts components MTBF in a much much shorter time period compared to gaming.
Which is why most companies 'TRY' to refuse RMA on gpus used for mining.

So in terms of components life expectancy gamers cards clearly have an upper hand imo, as components get exposed to less heat(except core), and most importantly for much less duration, even though gamers tend to be lazyass when it comes to maintenance.

What I'm not getting is, whether thermal cycling actually causes gpu failure more often than components degradation done with mining gpus.

 

[Phaaze88]

I see your point but can't it be still applied to all miners as well?

No, it can't, just like it can't be applied to all non-miners either. You have to take individuals into account - throwing everyone into two baskets, when you know every egg in their group won't operate the same, is an over-generalization.

They simply need to pump up voltages of the memory quite a bit in order to gain optimal hash rates.

What I'm saying is, yes miners may clean the dust or reapply pads/paste and whatnot, but the real problems are 'temperature' and 'duration' that components get exposed, as well as the voltages.

But they can't, as gpu memory voltage is locked in from vbios.
They'd do well to underclock and undervolt the gpu core so it's not wasting power and needlessly warming up the memory - well, not as much as it could be.

If it's all still within spec of the hardware - even the 90C memory temperature you're proposing, we're still talking years here... It's going to last at least until they cover ROI and then some, opening up a window to buy more.
The large scale mining ops are going to run 'em till they die. It's the small scale that have to think about selling off, as they may not be able to afford to ride out some dips.

Usually gamers don't prefer core temps go skyrocketing(90c or above).
Same as miners since they also don't prefer memory temps not going above 100c.

Tell that to those who don't even use monitoring software to check the condition of their stuff from time to time. Heck, some aren't even using the right software - looking at you, AMD FX owners/users, because we still have folks out there on them.

But gamers don't play the game all that long and it's not like they actually play games every single day.

Don't give people the benefit of the doubt. I'm sure some Twitch/Youtube streamers out there would make you sigh in disbelief.
Some have made a job out of this.

Whereas mining rigs run at 90c memory temp, possibly above that, for 24/7 without a rest every single day for months and years.
Some gamers are ignorant about the gpu temp but so do some miners as they don't care much about memory temp as long as it stays under 106c(throttle point).

And memory indeed is designed to withstand high heat(90c or even 100c depending on the types of memory), but it is not designed to operate at such high temp 24hrs for months and years.
Mining exhausts components MTBF in a much much shorter time period compared to gaming.
Which is why most companies 'TRY' to refuse RMA on gpus used for mining.

So in terms of components life expectancy gamers cards clearly have an upper hand imo, as components get exposed to less heat(except core), and most importantly for much less duration, even though gamers tend to be lazyass when it comes to maintenance.

It'd be nice if there wasn't this stigma that all cards are being run like that, because they're not.

Miners are perhaps more akin to PC enthusiasts than gamers are - actually, gamers may be too broad of a term. A miner is easy to define, but a gamer isn't.
I play games from on my PC every week, but I don't ID as a gamer.

What I'm not getting is, whether thermal cycling actually causes gpu failure more often than components degradation done with mining gpus.

There are ~SO many variables to that, that I would wish you good luck, and then ask: If it lasts several years either way, does it really matter?

 

[USAFRet]

Also, fail statistics can only be applied to fleetwide numbers.

The singular component in your PC may last years, it may die tomorrow.
No matter how it was used originally, or even if it is new off the shelf.

 

[Secrutinize]

No, it can't, just like it can't be applied to all non-miners either. You have to take individuals into account - throwing everyone into two baskets, when you know every egg in their group won't operate the same, is an over-generalization.

I agree. But you just generalized that a lot of gamers tend to overlook gpu temp while there are a lot of gamers who actually worry about gpu temp going up high, myself included.
And then, now you're stating it depends on each individual when it comes to applying your logic to average miners.
Call me nuts but I feel like you're very biased...

That aside, if we don't assume some level of generalization on any topic, there's not much to talk about on any subject.

But they can't, as gpu memory voltage is locked in from vbios.
They'd do well to underclock and undervolt the gpu core so it's not wasting power and needlessly warming up the memory - well, not as much as it could be.

Okay. I confused overclocking memory with pumping up voltages. This is rather embarrassing.
Pumping up voltages was a thing before eth, not after.

If it's all still within spec of the hardware - even the 90C memory temperature you're proposing, we're still talking years here...

Doubtful.
For gaming, indeed it would take literally years.
However, mining eth even for a year nearly reaches components MTBF, depending on which temp and how long mining rigs operated.
Basically, the higher the temp, the longer the duration, components get degraded that much quicker.
Given components like capacitors and vram are first things going out due to heat, I don't think running cards at even 80c 24/7 for years is too safe from failure from gpu failure due to components degradation.

Don't give people the benefit of the doubt. I'm sure some Twitch/Youtube streamers out there would make you sigh in disbelief.

"Overall, Americans spent an average 37.2 minutes per day on gaming and leisure computer use."
Not only that, depending on how demanding the actual game is, the degradation done on gamers cards could be even far much less insignificant than you and I think.

It'd be nice if there wasn't this stigma that all cards are being run like that, because they're not.

First of all, I never said ALL cards run over 90c.
Second, with something like gddr6x temp easily reaches 100c. Even for gddr6 there are A LOT of miners on forums such as reddit saying 90c is 'fine' for eth mining, meaning they're mostly okay with running the cards on that temp. So it's just an educated guess A LOT of miners could be running the cards at such high temps.
Third, you just can't know how many miners are fine with running cards on 90c and above. Which means you can't confirm that "they are not" part. The only thing you and I can do is speculating, like what you did with about average individual's gaming hours per day which was proven wrong.

Miners are perhaps more akin to PC enthusiasts than gamers are

True. Miners should know how gpus operate, very basic concepts of undervolting and underclocking overclocking, replacing fans and whatnot.
But does that prove miners cards are in a better condition in terms of lifespan compared to gamers cards?
No, because gamers don't run their cards 24/7 for years without a rest.

There are ~SO many variables to that, that I would wish you good luck, and then ask: If it lasts several years either way, does it really matter?

Indeed, there are a lot of variables for gpu failure due to thermal cycling. Someone could be really stupid and just OC'ed without a second thought about gpu temp. Maybe ambient temperature is either too high or low. Or maybe it could be just some random solder joints defects.
But you presume as if mining cards will surely last several years after all that degradation done on components. I'm not so convinced they would based on components MTBF.

Obviously some used mining gpus will survive, but so will some used gaming gpus.
But which cards have a higher chance to survive "in general"?
Because The Merge is coming this year and soon enough the market will be flooded with used mining gpus.
Consumers at least need to be aware of the potential risks about which one is more dangerous, used mining cards vs used gaming cards.
Of course if the price is right consumers will buy anything either way.
Having said that, if one side is worse than the other, it's only natural to have price adjusted to compensate for seemingly worse value.

Anyways, I think I've done enough 'speculating' how mining cards would have a much less lifespan left compared to gamers cards, "in general".
Yes yes it varies, but everyone knows that...

I just want to know how often does gpu failure due to thermal cycling actually occurs.

 

[Phaaze88]

Gonna call it quits, 2 steps too late.
I replied to 'which would you pick and why', and shouldn't have taken another step.

Best of luck finding a concrete response to your query.