Is a Graphics Card Worthless After 4 Years of Mining?
Every GPU manufacturer warns us from cards used for mining crypto with statements like “Mining cards lose 10% performance every year”, this encourages us, consumers, to forget about used cards and buy a brand new one. If you think about it, this is their main revenue stream. So it is obvious that they will not tell us to get a used card.
Buying a graphics card is usually painful for the wallet. Since the RTX 3k series release, it was crazy expensive with scalpers buying every single card available. The prices spiked along with cryptocurrencies. We can compare those two graphs where one of them represents the average price for the RTX 3060 ti and the second one represents the Bitcoin price over the same period.
Those charts are almost identical with two spikes, and we can finally see that it is getting better since it is less and less profitable to mine cryptocurrencies. Right now we can find many graphics cards on the market which were used for mining, what should you expect from such cards? Is it worth the risk?
The Test Subject
For testing purposes, I have bought two GTX 1070 which have been used for mining constantly for almost 4 years! The reason I got two cards from the same rig is to compare if they have degraded differently. Each card was cleaned before the tests including the exchange of the thermal paste. The goal is to test the performance, temperatures, and stability of those cards.
Testing Platform
CPU AMD Ryzen 7 3700X
GPU Gainward GeForce GTX 1070 8GB GDDR5
MEM G.Skill Trident Z Neo, DDR4, 32 GB, 3600MHz
Motherboard MSI MPG X570 GAMING EDGE
Hard drive ADATA 512GB M.2 PCIe NVMe XPG SX6000 Pro
Power Supply SeaSonic FOCUS GX-650
Full specs of the graphics card are available here.
FurMark
This test is a simple burn test to check the maximum card performance along with the temperatures, additionally, we will check for any stability issues. The test duration was 30 minutes with a resolution set to 1920x1080.
Results
The usage was a constant 100%. Both cards behaved almost identical holding an average of 103 fps and keeping stable 75 degrees Celcius on the processor which is not ideal… but acceptable. The minimum fps, the cards hit, were 99 and 96, during the test no stability issues occurred.
Unigine Heaven 4.0
The next test's purpose is performance and stability. The Unigine Heaven 4.0 is a great tool for testing your graphics card. This software provides us with a benchmark scenario allowing the user to choose a level of details during the test which translates to how heavy it will be for our GPU. For this test, we will go with the High details and tessellation disabled.
Results
Card number one
Card number two
Here it gets a bit more interesting. Both cards scored on average pretty good with scores similar to the ones achieved from cards that have not been used for mining. But there was a small difference between those two cards, the Min FPS. The first card had some problems during the test and running the test again did not fix the issue. Even with this small stability issue, both cards received a very similar score.
Witcher 3
What would a graphics card test be without some gaming? I have tested the GPUs on the Witcher 3 since it was used in multiple benchmarks back in the day and it will be perfect for a comparison between the used and new cards. For our test, we will play the game on Ultra preset in 1920x1080 resolution.
Results
Again, both cards scored almost identical, but we can notice a clear difference in comparison with the average score on UserBenchmark.
The degradation in performance was around 11%, of course, this result could vary from the game location since some of them are more GPU heavy. Nonetheless, the temperatures were set to constant 75 degrees Celcius just after a few minutes of gameplay. During the gameplay, the minimum FPS that I managed to hit was 64.
Conclusions
It is incredibly hard to advise someone on a used GPU card especially if there is a risk the card was used for mining crypto. During my tests, I have observed that the degradation of such cards is not as severe as some companies suggest, but it is real and we should keep that in mind.
Both tested GPU temperatures did not exceed 75 degrees, but on the other hand, such temperatures were achieved very quickly which could lead to potential thermal throttling (reducing the performance due to overheating) in the future.
If you consider purchasing such a card, it should be fairly tested with benchmarking software to check if the clock frequencies are close to the stock ones, or if the card does not overheat.
To sum everything up, it might be worth buying a used card, but only if we understand the risk.
