Gaming Performance Tested On 'Worn Out' RTX 2080 Ti Mining Card

[Kisai]

7 hours ago, leadeater said:

Really do think most aren't thinking these things through properly. What exactly do people actually think mining does that degrades or could degrade the card so much, especially compared to any other workload type.

 

Keeping track of what a card was used for is completely pointless and doesn't actually provide meaningful information to a second hand buyer, all that's going to happen is personal uneducated bias is going to come in to it and end up passing up a perfect good card for one that is potentially worse simply because it was only used for "gaming", as if that makes any difference.

 

I disagree. You have no idea what any second hand GPU was used for, but the guy with zero feedback selling 50 Geforce 3080's with "NO REFUND" is a major redflag. I'd argue miners treat their GPU's worse, have you seen the ramshackle rigs setup in rooms with sitting on carpets and a dozens of extension cords? Those rooms are probably over 40 degrees C like a typical server room would be without an air conditioner.

 

 

[Moonzy]

23 minutes ago, Kisai said:

Those rooms are probably over 40 degrees C like a typical server room would be without an air conditioner.

ambient temp doesn't really matter in this context, GPU component temps matters more

when mining, GPU draws less power compared to when gaming

my ambient rose from 24 to 31 degrees for where i keep my rigs, my GPU temps are still only ~68c for the hottest card's core temp

majority of them are around 55c

 

23 minutes ago, Kisai said:

have you seen the ramshackle rigs setup in rooms with sitting on carpets and a dozens of extension cords?

that's more of a fire hazard than component damage, i would think?

 

but of course there are dumb miners, there's always dumb people in every community

[Tellos]

On 2/26/2021 at 5:36 AM, LAwLz said:

Oh the horrors of only having a GTX 10 series graphics card. How terrible. It must be painful to have to switch settings down to medium to get decent framerates in the latest COD game or whatever the flavor of the month is these days... I really feel your and everyone else's pain.

 

 

They aren't more entitled to it.

The one who buys it the quickest and/or pays the most should get it if you ask me. That's why I don't get why people are mad at scalpers either.

Scalpers exists because demand outweighs supply. Some people will be without graphics cards because not enough exists, plain and simple. Scalpers take advantage of that situation and as a result the ones who pay the most are the ones who get the limited supply. If someone wants to buy a graphics card for 500 and someone is willing to pay 600 for it, then I think the one paying 600 deserves to get it. The problem is that right now the one only willing to pay 500 for it gets pissy and whines online how it is unfair because they for some reason deserve the card more, in their own eyes.

 

I don't think neither miners nor games are good/bad. In my eyes they are equal, but right now I am seeing a lot of bitching and moaning from gamers who think they are for some reason more entitled to get graphics cards than miners, even if miners are willing to pay more or buys them quicker. Boohoo.

That's why I might seem like I am "hating on gamers" right now. I would be replying exactly the same way if the forum was full of miners complaining about how "gamers are buying all of their mining cards". You don't deserve special privileges just because you think your use of a graphics card is more valid than someone else's.

 

 

Also, are you implying that you need a new graphics card in order to enjoy playing some games? You don't. Turn the settings down. If the game is good then it should be an enjoyable experience regardless. If you can't play some new, graphically demanding AAA game even with settings turned down then play something else instead. There are countless of great games that you can play. You don't always have to play the latest game.

Well my issue is my GTX 1080 WILL eventually break and once that occurs my PC is useless. Ryzen has no built in graphics so my rig will become unusable. Its gonna turn 4 years old this year. How many years can I run it before it breaks? Also if this is the attitude fine then I say never buy from miners deny them money. They bought the card why should I give them my money for a card they used? I am sorry my desire to buy a new card after having waited for so long offends people. 

 

Also why do Miners need the most efficient card by your logic why can't they be fine with less money made?

[suicidalfranco]

25 minutes ago, Tellos said:

Miners need the most efficient card

they don't need them either, they're simply more willing to spend than you are. Whether it's at MSRP or at an overly inflated price.

[Tellos]

59 minutes ago, suicidalfranco said:

they don't need them either, they're simply more willing to spend than you are. Whether it's at MSRP or at an overly inflated price.

Fine then I dont have to buy from them and get to encourage others not to. Not my job to give them ROI.

[Moonzy]

3 minutes ago, Tellos said:

Fine then I dont have to buy from them. Not my job to give them ROI.

ok, it's your money after all

 

4 minutes ago, Tellos said:

get to encourage others not to

sure, but i wish you wouldn't spread misinformation while doing so

[MageTank]

2 minutes ago, Tellos said:

Fine then I dont have to buy from them and get to encourage others not to. Not my job to give them ROI.

I don't believe anyone is telling you that you have to buy from them, nor are they encouraging you to tell others to do so. Unless I am missing where that occurred in this thread, I just don't see it. 

 

You're free to act on your own accord for anything you wish to do in life, as long as you don't intend to bash others for enjoying the same freedom. I personally wouldn't buy a card from a miner, mostly because I don't trust what some of them do with VBIOS mods. Voltage/thermally speaking, they are often operated at lower voltages/lower thermals than what most gamers do with their cards, but I wouldn't judge someone else if they decided to buy a card from a miner.

 

A friend of mine likes to flash his cards using different VBIOS's for modified power limits/fan curves and for some odd reason, it tends to result in one of the DP ports not working until it is flashed back. A consumer not knowing this might think they got a defective port, so I typically advise people avoid buying some models of used cards unless they know in advance what to look for and the tools required to remedy this in the event a card is sold in this condition.

 

I'd also say I wouldn't buy a card from overclockers such as myself because we get a little out of hand with the #2 pencil if you know what I mean, and you never know what kind of effect a modified power limit will have on a card in the long run after some aggressive benching sessions. In fact, I'd avoid buying anything from me over a miner given my track record with overclocking and how aggressive I can be with my hardware, lol.

[FakeNSA]

On 2/27/2021 at 5:01 AM, Lurick said:

I personally blame @Energycore because birbs are the root of all evil!

Squints menacingly

You aren't exactly wrong though.

[DildorTheDecent]

11 minutes ago, MageTank said:

A friend of mine likes to flash his cards using different VBIOS's for modified power limits/fan curves and for some odd reason, it tends to result in one of the DP ports not working until it is flashed back.

kek. Seems to be reported by users flashing the Strix XOC BIOS to their GPU. For whatever reason some of their Pascal and Turing GPUs 2x DP and 2x HDMI as opposed to 3x DP and 1x HDMI. So you end up losing a DP in the process. I don't think Ampere is affected.

[Blademaster91]

8 hours ago, Moonzy said:

At this point I just wanna cry tbh

Most of the hate on crypto is really unfounded and unreasonable, probably stems from jealousy.

 

Somehow miners who monitors temps and what not (most mining applications that I know reports card temps) is damaging the GPU more than gamers who games with dusty GPU that overheats and they wouldn't know until game fps drops to beyond playable

 

The application I use:

  Reveal hidden contents

 

The temp starts becoming yellow if it's above 70

Friggin 70, most gamers run their GPU above that

Some of my GPU runs cooler under load than your GPU when idling

 

Somehow pumping lower voltage is more damaging, somehow less thermal cycle is more damaging

 

Saying miners don't care about their hardware is stupidly false, we're in it to earn money, not to kill hardware, and hardware isn't free

That seems more like anecdotal claims than actual proof that every miner is going to care about temps or power limits.

But to your claim about gamers don't care about keeping their PC clean, or doesn't check temps that isn't true either. Also with how screwed the GPU market is I would guess anyone that uses their GPU for games or work cares a lot more for it because getting a GPU is unobtainium except for miners.

Most miners aren't going to want to stop their money machine to clean it, it doesn't really make sense as thats profit lost and most don't either because once the card hits ROI it goes on ebay.

2 hours ago, Moonzy said:

ambient temp doesn't really matter in this context, GPU component temps matters more

when mining, GPU draws less power compared to when gaming

I don't see how that matters either, a single gaming PC might raise room temps by 1-2c.

But VRAM temps mater a lot more with a RTX 30 series card, having GPUs sitting out in the open is worse than a single card being in a case with airflow.

2 hours ago, Moonzy said:

that's more of a fire hazard than component damage, i would think?

 

but of course there are dumb miners, there's always dumb people in every community

And static damage to the GPU, but i've seen those careless setups a lot so I'd definitely be hesitant to buy any used GPU.

[Moonzy]

1 hour ago, Blademaster91 said:

That seems more like anecdotal claims than actual proof that every miner is going to care about temps or power limits.

i can't say much about temps, but I can point out that almost all mining calculators have their GPU at below the default power limit (pretty much at optimal power limit as tested by users), that's why people tend to ask why their results differ so much from those calc

here's some mining review on the 30 series GPU, though (skipped to the numbers for you, hope it works)

 

Edit: here's a different source in case you say I only use 1 source

https://www.nicehash.com/blog/post/nvidia-rtx-3000-series-overclocking-guide-to-increase-mining-profits

 

temps correlates with power limit most of the time, so make out of that what you will

 

1 hour ago, Blademaster91 said:

But to your claim about gamers don't care about keeping their PC clean, or doesn't check temps that isn't true either.

majority of the gamer friends i know never checks temps, they dont even have any monitoring tool installed

they're not pc enthusiasts, I built their PC for them

and i do maintenance for them every so often when they come to me, and it's common to see dust bunnies in their PC

 

of course, my sample size may be limited, but this is my experience

perhaps someone working at retail can confirm or debunk my assumption.

 

Edit: or you can visit the experience with non-techie thread in GD

 

1 hour ago, Blademaster91 said:

GPU is unobtainium except for miners.

it's also unobtanium for me now, since the price hiked to oblivion

i still draw the line somewhere, im not gonna buy a 3090 at $4000 in today's market

 

1 hour ago, Blademaster91 said:

Most miners aren't going to want to stop their money machine to clean it, it doesn't really make sense as thats profit lost and most don't either because once the card hits ROI it goes on ebay.

im not sure how mining operation usually goes, the ones i used to follow on youtube purchases RX480s occasionally when when there's Turing and RDNA GPU everywhere, so i don't think they sell the cards as long as they hit ROI, i would assume they sell them when it's no longer profitable (more electric than coins)

 

september 2019:

 

1 hour ago, Blademaster91 said:

I don't see how that matters either, a single gaming PC might raise room temps by 1-2c.

point being, the components temps doesn't necessarily correlate with ambient temp when other factors are at play too

of course given the same circumstances, higher ambient temp = higher GPU temp, but mining GPU runs at lower power limit.

 

1 hour ago, Blademaster91 said:

And static damage to the GPU, but i've seen those careless setups a lot so I'd definitely be hesitant to buy any used GPU.

fair, but computer should be grounded and the GPU dont lay on the carpet directly so i wouldn't worry too much about it

that said, i personally dont buy used hardwares so i can relate to the fear.

[BabaGanuche]

While I do not do mining I do folding at home which I would expect would be quite similar to mining. My folding cards have a way easier life than my gaming card. They do almost no thermal cycles and have their power limit set lower to be at the peak of the flops per watt curve so they run cooler. Versus my gaming card which has the power limit set to 130% and does tones of thermal cycles and runs hotter. Thermal cycles is what typically kills very large die GPUs. There is a lot of stress caused by the different thermal expansion rates of the die, substrate and PCB when warming up and cooling down. It is one the reasons baking them can fix them sometimes. It reflows the connections when they get cracked from thermal expansion. I would not be even sightly surprised if my gaming card dies first.

 

My only concern for general users would be if there was a non standard VBIOS flashed.

[leadeater]

6 hours ago, Kisai said:

Those rooms are probably over 40 degrees C

And a bad PC case isn't? Internal PC case temps are likely a lot hotter than you think. However ambient temps do not matter, like at all. That's why GPUs have thermal management and will not prematurely die, not even at Tjmax for it's entire running life. Even modern digital VRMs have thermal controls in them so they won't overhead and die either. Memory maybe, they do have some thermal controls but as long as the card doesn't have actually terrible memory cooling high ambient won't cause a problem either.

 

Australians love their GPUs too, as well as other very hot places and they do just fine.

 

6 hours ago, Kisai said:

have you seen the ramshackle rigs setup in rooms with sitting on carpets and a dozens of extension cords?

And? The rigs are still electrically insulated and grounded through the PSU, if not they won't function at all. Miners don't have systems just laying about, not one's with a lot of cards, no they use milk creates and wooden boxes etc which will not cause any problems at all. There is no difference between these and an open air test bench, as seen used by most reviewers.

 

Like I said you need to actually think through the potential issues more, a perceived issue is not an actual issue. Evaluate how what you think is a problem actually could be a problem.

 

Either buy a used GPU or do not, it's a risk no matter who where what or how. It's a risk I believe is worth it for many, so long as you have basic diagnostic capabilities. 

[leadeater]

Thank you GN for being so timely with this video that just came out now, directly relates to this topic.

 

 

 

Edit:

[Moonzy]

11 minutes ago, leadeater said:

Thank you GN for being so timely with this video that just came out now, directly relates to this topic.

 

 

 

I watched it, sad he didn't test performance numbers in fps

But only the 760 was bad enough iirc, 12c delta

 

I'm not sure how true this is, my intuition tells me it's the amount of heat that's being transferred instead of cycles

[leadeater]

3 minutes ago, Moonzy said:

I watched it, sad he didn't test performance numbers in fps

But only the 760 was bad enough iirc, 12c delta

Well tbh the performance wouldn't be different by any meaningful degree as the cards weren't reaching a thermal state where there would be significant clock reductions, some of those cards predate dynamic GPU boost too so that's slightly less relevant to GTX 10 series and newer. Either way it shows how easily a "not used for mining" GPU can actually get damn dirty lol.

[Rym]

Following a new video released by Gamer's Nexus where the temperatures were tested on GPUs before and after re-paste we have noted the fact that unlike mining, normal use of GPUs require up to between 8 and 10 years of usage for their temperatures to increase by an average of 10 degrees. In comparison, mining has placed such an immense stress on GPUs that their temperature increase is up to double that of normal use-case scenario after a decade of use, that is, 20 degrees. In addition to further clarify, the GPUs for mining have only been used for 1,5 years.

 

In total, on average the mined GPUs have been worn out 5.3 times faster or 530% faster than the worst tested GPU in normal use case scenario which had a temperature of 80.3 degrees in full load after never being re-pasted for nearly 8 years, and a temperature of 68.0 degrees after it was re-pasted.

 

 

In addition the GTX 970 was also tested with a before and after re-paste. However, the GTX 970 saw a difference of only 0.3 degrees.

 

The effects of mining on the GPU are clear, the cooling solution is worn out on average 5 times faster than any other normal use-case scenario and this is without mentioning the fact that the GPU itself is permanently worn down by the action of continuous mining. This reveals the reality of the situation and demonstrates that the lifespan of the GPU is greatly diminished even when conservative mining operations are being done. 

 

This is simply for a buyer's perspective however, that they should not expect their mined GPU hardware to live past 1 year of normal use, and after 2 years of use sudden catastrophic failure should be an expected outcome rather than a surprise.

 

However, it is also true that older GPUs may use less powerful cooling solutions when compared to newer models which use more advanced coolers that may provide better results, as for example while the GTX 970 has a normal operational gaming temperature of between 70 to 80 degrees a normal RTX 2080 Ti operates between 50 to 60 degrees, so the wearing down aspect of mining may even be up to 1000% greater than that of normal use case scenarios.

 

 

This post is not meant to defame or attack miners but to show the reality of mining and how it affects the GPU from the point of view of a buyer. A mined GPU is worthwhile to purchase if the price is in accordance to how worn down and close to dying the GPU is, as for example a 1 year lifespan for a 300$ RTX 3080 is still a good deal nonetheless.

[leadeater]

1 hour ago, Rym said:

In comparison, mining has placed such an immense stress on GPUs that their temperature increase is up to double that of normal use-case scenario after a decade of use, that is, 20 degrees. In addition to further clarify, the GPUs for mining have only been used for 1,5 years.

Or wherever this single sample of GPU was used was very dirty, which wouldn't be much different from a computer case on carpet with bottom intake either.

 

And no the video, and others, have clearly stated thermal cycles is the only issue in regards to thermal paste and pads. A typical mining GPU would under go below 10 cycles in it's entire life where as a gaming GPU would be thousands.

 

In furtherance to that you just used some very bad logic here, you used a temperature delta on a vastly lower TDP GPU comparative to one that is much higher, the same amount of dirt buildup on a high power GPU would result in a higher temperature delta.

 

But the long and short of it is, blow the dust out without even taking the cooler off will restore almost all thermal performance. Coolers do not "wear out", the fans on them sure.

 

1 hour ago, Rym said:

The effects of mining on the GPU are clear, the cooling solution is worn out on average 5 times faster than any other normal use-case scenario and this is without mentioning the fact that the GPU itself is permanently worn down by the action of continuous mining.

No it isn't clear, you're just using a single sample and very poor logic here.

 

Mining is not the root cause of what ever test happened for this source article, the root problem is where that singular GPU was used was dirty. That's it, mining workload is irrelevant.

 

For all we know that GPU was setup mining in a sawmill. Or this was a specific targeted attempt to drum up anti-mining rhetoric and the test was a farce to start with and designed to show this.

[Moonzy]

32 minutes ago, leadeater said:

the video, and others, have clearly stated thermal cycles is the only issue in regards to thermal paste and pads

im actually curious, why would thermal cycle affect the longevity instead of the total energy transfered?

 

edit: Micro cracks from material fatigue?

[Kisai]

45 minutes ago, Moonzy said:

im actually curious, why would thermal cycle affect the longevity instead of the total energy transfered?

 

edit: Micro cracks from material fatigue?

Probably the thermal pads/paste cracking, resulting in uneven thermal transfer.

 

Like in laptops and desktops, removing the cpu/gpu heatsink should come straight off. In the oldest GPU in the GN video, it came off with a lound "clink" , which meant the thermal paste had shrunk and cemented itself, which means it doesn't transfer very well. The 760 also had a similar "cemented" heatsink from the thermal paste, and that's the only one that actually had a large temperature delta after the thermal paste was replaced.

 

The newer cards, were still "soft" on the thermal paste, which is why there wasn't much of a difference. Thermal paste dries out after 2-3 years, and it's caused by the high thermals, not the thermal cycle of the chip heating up and cooling down. So as it hardens it goes from being able to fill in the gaps when it expands, to leaving gaps/cracks in the hard paste that isn't able to expand anymore. That's just simple cooling shock. Also the same reason why you don't use compressed/canned air on a system that isn't cold, the canned air will lower the temperature quickly and crack the the die.

 

So my expectation is that a computer that is never turned off, will need to replace the thermal paste sooner, and one that is run full out would need to replace the thermal paste after 2 years at the soonest. The cheaper the GPU, the more frequently it will have to be done due to the cheaper designs that allow more dust into it. Since the high end GPU and CPU coolers are fairly involved to dismantle it's not something most people will do.

 

But as pointed out in the GN video, even if you didn't replace the cooling paste, the cards were all still operating within their expected thermal parameters, so at best you'd probably lower the noise from the cooling fans rather than improve the performance.

 

[leadeater]

3 hours ago, Moonzy said:

edit: Micro cracks from material fatigue?

This, the movement of the two surfaces unsettles the thermal paste. What many don't realize is that thermal paste/grease is actually supposed to dry out and cure a bit because the wetness has little to do with the thermal conductivity of it. It's actually conductive materials suspended in a solution, think sand in water, so once it goes dry the contact between the two surfaces can be broken by movement of these, cracking the paste dislodging the contact. Once this has happened enough you'll get thermal issues.

 

So if you take off a cooler and find the paste is dry it's not actually correct to assume it was bad, only thermal testing like Steve did could tell you that. You can see the paste on the 970 was getting a bit dry but could still be spread, it is however getting to the point where movement from thermal cycles would start to affect the contact between the surfaces.

[Moonzy]

3 hours ago, leadeater said:

This, the movement of the two surfaces unsettles the thermal paste. What many don't realize is that thermal paste/grease is actually supposed to dry out and cure a bit because the wetness has little to do with the thermal conductivity of it. It's actually conductive materials suspended in a solution, think sand in water, so once it goes dry the contact between the two surfaces can be broken by movement of these, cracking the paste dislodging the contact. Once this has happened enough you'll get thermal issues.

 

So if you take off a cooler and find the paste is dry it's not actually correct to assume it was bad, only thermal testing like Steve did could tell you that. You can see the paste on the 970 was getting a bit dry but could still be spread, it is however getting to the point where movement from thermal cycles would start to affect the contact between the surfaces.

Ah, I thought the wetness is what allows it to conform to the shape of the gaps, so it can fill in the gaps even if the heat load changes, and once it dries it becomes unable to do that and is ineffective, guess I overestimated the thermal strain those dies experience

 

But yea, I never had to repasted a single one of my GPU yet, thermal pad too

[leadeater]

49 minutes ago, Moonzy said:

I thought the wetness is what allows it to conform to the shape of the gaps

Yes this is true but a mining card for example wouldn't need this as much due to the constant heat load and temperature so it curing in place isn't as much of a problem, that's why Steve said thermal cycles is the primary factor for interface degradation. If you aren't thermal cycling then you don't need the paste to be compliant and malleable anymore, problem is if it cracks when taken out of the system and sold the thermal performance will be bad but it's not going to happen in 1.5 years, no way.

 

Thermal pads this is much less of a factor, those are much harder to dry out, much harder. That's why more industrial and rugged equipment favor pads over paste. 

[Mihle]

Mining on my RX 5700 XT, GPU temp is actually lower than while gaming or F@H, but memory temperature is higher. No power limit or anything was adjusted.

[StDragon]

7 hours ago, leadeater said:

Thermal pads this is much less of a factor, those are much harder to dry out, much harder. That's why more industrial and rugged equipment favor pads over paste. 

It depends on the quality and if the thickness was specced out properly. I've seen some pad shrink due to dry-out to the point of surface separation from the chip (gap). I'm sure the formulation is much better nowadays that still keeps its spongy like pressure on contact.

 

The only reason to really use pads is due to conformity on contact, but whenever possible a paste should be used. IMHO I prefer graphite thermal pads. Superior thermal conductivity for being solid-state like (as in never have to re-apply regardless of thermal cycling). The only downside to graphite pads are that they're really thin, so not good for ICs like VRMs and VRAM in terms of conforming to differing gaps and tolerances in manufacturing variances.