RTX after mining and rendering. Is there a difference?

Question

Video cards

Hello.
I'm looking at the RTX 3080 video card. For experiments in 3D and rendering. I read how cards are raped in cryptocurrency mining and that after mining they have problems with both electronics and the cooling system (such as “thermal pads leaked”). And I thought that 3D rendering also loads these cards wildly, when scenes can be rendered for hours and days. Will there not be the same problems in this case as with mining?

Answer 1

I don't quite understand, are you looking for a new one or a used one?
If it's new, just look for a new one with LHR and with miners' reviews. There are certain cards that don't care at all, there are very hot ones, in general, you can find an option for you.
But if you want to buy used after mining, etc., you definitely have to test it.
As for leaks and other things, only certain cards from certain companies suffer from this. For example, 1080 gigabytes, but unlike AMD, there are no seals on the screws, you can easily disassemble and smear it yourself even on a new video card, okay.
You also need to understand that miners are not stupid either, as a rule they adjust the voltage and so on without damaging the video card. The most important thing I want to convey to you is that video cards in mining have a constant load, unlike simple home use, where constant jumps in frequencies and temperatures prevail. Let's just say that games will harm the video card more than mining.

Answer 2

Yes it loads. But usually overloading is not done, as is often done in mining
. But my personal opinion is that the card should be serviced in any case, if it passes all stress tests, I don’t see why not to take it

Answer 3

The root of all problems with video cards is the cooling system. If the maximum and constant temperature does not rise above 70 degrees, then there will most likely be no problems.
There is not much iron, which cools normally out of the box. The best solutions are dropsy fullcover, when the radiator covers all the chips, and not only the memory, but also the power subsystem. Unfortunately, the replacement of the cooling system is the official reason for the denial of warranty service (it can be used but not necessary, everything is at the discretion of the service).
Why does mining rape iron to death? because miners are trying to squeeze out dozens of additional percent due to extreme overclocking. The RAM of the video card is used to the maximum here (there are cryptocurrencies where this is not the case, they even lower its frequencies to reduce power consumption), no other task loads the hardware so much (neural networks are still questionable, but there the load profile is highly dependent on the tasks) It can be seen by what the memory chips turn into after mining.
Rendering does not load the hardware so much, the load profile in rendering is shifted towards the processor

Answer 4

So let's be clear, because I'm mining and rendering on my 3080 and 1080ti. At the same time, we will refute or confirm a couple of myths in the answers.
Base: mining is in most cases ether, ETH, which works according to the Dagger Hashimoto algorithm . It heats up the video card memory (VRAM) very much, because it works by moving data in it. Therefore, the memory on the same 3080 easily warms up to 100 degrees with the declared maximum of 110 degrees.
https://www.hardwaretimes.com/mining-ethereum-rais...
There is no memory temperature sensor on the reference 1080ti, but there is a hotspot - it is 94 degrees.
The GPU itself does not get very hot at the same time - about 10-15 degrees colder than 100% load in the render.
At the same time, I have an argument for you in favor of the card from under the miner, if we are talking about Ampere (3080 and others like them). The buyers of these cards not so long ago faced the fact that instead of thermal pads they put real garbage in their video cards: the memory in the video cards warmed up to 100+ degrees and throttled even in games.
https://www.tomshardware.com/news/gpu-thermal-pads...
If such a card got to the miner, the thermal pads were replaced with high-quality ones on the same day. Because memory overheating = hash rate drop = profit drop.
In addition, miner cards usually operate at 100% fan speed and are often (in a farm, for example) cooled additionally, if necessary, by cool fans.
Thus, it turns out that, in general, a map from a competent miner, and even better - from a farm - is better than a map from a rendering amateur and sometimes even a gamer. The literacy of the miner will have to be figured out on your own, though. Well, it's not entirely clear why they would sell normal, working cards: they still pay off very well.
Mining "rapes" the memory, but not the chip. Memory, if desired and skill, can be changed, if anything, stupidly solder, if some module gets sick. With a GPU chip that has fallen off the renderer (for example, I don’t think that this still happens in desktop cards), such a trick will not work.

And I thought that 3D rendering also loads these cards wildly, when scenes can be rendered for hours and days. Will there not be the same problems in this case as with mining?

Will. But in such loads, you need to monitor the temperature of the GPU, not VRAM.
In general, constant monitoring of the temperatures of everything , in the background, with notifications of strange values, has not hurt anyone yet. Through hwinfo, for example.