Crypto Miners Start Dumping GPUs, RTX 3080s Listed for $523

bigdragon said:

The RTX 30-series also has anemic amounts of VRAM.

8GB of VRAM is enough for the vast majority of games. Worst case, you may have to settle for high-resolution textures instead of ultra and that is about it. No major VRAM problem there.

Also keep in mind that people who have been holding off from buying a new GPU for this long and may be eyeing heavily discounted retired crypto-mining GPUs are likely very much willing to compromise on some details to make their games run smoothly on 8GB if that means saving another good chunk of change.

bigdragon said:

The RTX 30-series also has anemic amounts of VRAM.

I've yet to find a game where my RTX 2070 Super struggles because of VRAM issues, and that's assuming the game even chews up all of the VRAM to begin with. Plus things like DLSS makes VRAM pressure less of a problem.

-Fran- said:

That is only for the GPU, but not the memory. The memory is tuned for max performance and it was abused for a long time and hard, 24/7 even, in those mining farms. This is well known.

So, in short, no. Buying these cards is almost guaranteed to not last a long time if they don't just die immediately afterwards. That's not even taking into account hacked firmware and other "customization" done to it.

Even low scale farmers, I'd say, are not free of these points/issues. I would advise people to stay away from confirmed mining cards if possible and, as Makaveli mentioned above, there's also a stance to take: do not help these rats recoup their money.

Regards.

I don't think that there's any evidence that running memory at max performance 24/7 reduces it's lifespan. Semiconductors like CPUs and RAM (when run within spec) can essentially run forever - and these cards have thermal protections to essentially force them to throttle/shutdown rather than run at a too-high temperatures.
Capacitors fail faster when run hot, flash memory has limited writes, and moving parts wear out... but the GPU and GDDR themselves are the least of your worries.

"Commercial" mining farms want to waste cycles on fake math at maximum efficiency, and GPUs are less efficient as temperatures increase. I would really only be only worried about dust, normal thermal paste maintenance, and the fans, which are probably still fine and usually not even that hard to replace. Of course, all bets are off it it was run on brownout power in some hacked-together rig that an amateur hot boxed in his garage.
You can still try to avoid mining cards but, realistically, 99% of the used current gen cards you'll find have been used for mining, regardless of what the seller says.

Remember kids, you may be desperate for a new GPU. But, whatever you do, DO NOT buy a 2nd hand clapped out mining card. You will be sorry.

Giroro said:

I don't think that there's any evidence that running memory at max performance 24/7 reduces it's lifespan. Semiconductors like CPUs and RAM (when run within spec) can essentially run forever - and these cards have thermal protections to essentially force them to throttle/shutdown rather than run at a too-high temperatures.
Capacitors fail faster when run hot, flash memory has limited writes, and moving parts wear out... but the GPU and GDDR themselves are the least of your worries.

"Commercial" mining farms want to waste cycles on fake math at maximum efficiency, and GPUs are less efficient as temperatures increase. I would really only be only worried about dust, normal thermal paste maintenance, and the fans, which are probably still fine and usually not even that hard to replace. Of course, all bets are off it it was run on brownout power in some hacked-together rig that an amateur hot boxed in his garage.
You can still try to avoid mining cards but, realistically, 99% of the used current gen cards you'll find have been used for mining, regardless of what the seller says.

kdzsBDenww4:497View: https://www.youtube.com/watch?v=kdzsBDenww4&t=497s

Regards.

Giroro said:

I don't think that there's any evidence that running memory at max performance 24/7 reduces it's lifespan. Semiconductors like CPUs and RAM (when run within spec) can essentially run forever - and these cards have thermal protections to essentially force them to throttle/shutdown rather than run at a too-high temperatures.
Capacitors fail faster when run hot, flash memory has limited writes, and moving parts wear out... but the GPU and GDDR themselves are the least of your worries.

"Commercial" mining farms want to waste cycles on fake math at maximum efficiency, and GPUs are less efficient as temperatures increase. I would really only be only worried about dust, normal thermal paste maintenance, and the fans, which are probably still fine and usually not even that hard to replace. Of course, all bets are off it it was run on brownout power in some hacked-together rig that an amateur hot boxed in his garage.
You can still try to avoid mining cards but, realistically, 99% of the used current gen cards you'll find have been used for mining, regardless of what the seller says.

You could also add that thermal stress is limited on a mining card if they've been working basically at 100% most of the time. This is unlike a consumer card that experiences swings of thermal stress at the very least. Also consumers tend to leave their cards at default settings, or at the very least, push the card hard for that 5% extra performance.

Basically, I equate this to someone who drove a car for 300,000 miles but mostly on the freeway versus someone who drove only 20,000 miles and it was a track car.