Best way to test a used GPU

[boarder2k7]

I just picked up a trio of R9 290 cards and was wondering what the best way to test them is. I'd like to run a full function test to make sure everything is okay as soon as I can when they get here to make sure any issues can be sorted out with ebay.

Is there any particular one stop shop program that is best for this?

I have an ASUS M5A99X EVO R2.0 in a bench so I can check one at a time at 16x or crossfire at 8x.

[Tie Lightning]

I just picked up a trio of R9 290 cards and was wondering what the best way to test them is. I'd like to run a full function test to make sure everything is okay as soon as I can when they get here to make sure any issues can be sorted out with ebay.

Is there any particular one stop shop program that is best for this?

I have an ASUS M5A99X EVO R2.0 in a bench so I can check one at a time at 16x or crossfire at 8x.

Heaven Benchmark??

 

Edit: Furmark??

[MegaDave91]

Game with it

[boarder2k7]

Heaven Benchmark??

Edit: Furmark??

I'll check both these out when they show up on Friday.

Game with it

While a good real world answer, it wouldn't be useful in a potential ebay dispute since that's subjective. Something like failing a benchmark should have more leverage should there be a problem. (I really hope they just all behave though.)

[-rascal-]

Recommended to do all of the following:

• Unigine Valley and/or Unigine Heaven (Furmark is not 100% necessary, but is also an option)
• Cinebench
• Regular use / gaming

Test one card at a time; not all problems can be detected or arise when running GPU's in Crossfire (or SLi).

 

I am buying a used MSi GTX 760 for a friend (who lives like 2 hours drive from me), and I'll have to test it -- since the original owner no longer has a PC with a proper PSU to test it...

[koji]

Have it benchmarked repeatedly and consecutively. Because just single benchmark wont be comparable with your hours of gaming.

[thekeemo]

Just make sure they have warranty above all else

Then gaming + MSI kombustor

[Mew]

While a good real world answer, it wouldn't be useful in a potential ebay dispute since that's subjective. Something like failing a benchmark should have more leverage should there be a problem. (I really hope they just all behave though.)

 

I think if you're able to capture it artifacting or crashing while gaming, that's a pretty strong argument that it is indeed defective. 

[angrymoose]

Edit: Furmark??

 

I won't run Furmark on any of my GPU's, there's just no reason to.

 

Run Heaven/Valley and 3DMark.

[airdeano]

also look through GPU-Z for the asic value. some of these easy to get for the

price cards prolly were mining cards and tortured to less than 50% and make

for some short lived cards to use. another to look for is the VRM temperatures.

105° was usually a blower/reference max.

[boarder2k7]

I think if you're able to capture it artifacting or crashing while gaming, that's a pretty strong argument that it is indeed defective.

That's a good point.

also look through GPU-Z for the asic value. some of these easy to get for the

price cards prolly were mining cards and tortured to less than 50% and make

for some short lived cards to use. another to look for is the VRM temperatures.

105° was usually a blower/reference max.

Is mining that bad for a card always, or just if you let it run too hot?

[airdeano]

mining has a huge toll on cards.

[xAcid9]

also look through GPU-Z for the asic value. some of these easy to get for the

price cards prolly were mining cards and tortured to less than 50% and make

for some short lived cards to use. another to look for is the VRM temperatures.

105° was usually a blower/reference max.

I didn't know GPU-Z ASIC quality reading can change. It didn't change at all on my 7950 last time when i mined it 24/7 for almost 3 months.

[boarder2k7]

mining has a huge toll on cards.

Even if they never overheat/run too close to thermal limits?

I didn't know GPU-Z ASIC quality reading can change. It didn't change at all on my 7950 last time when i mined it 24/7 for almost 3 months.

That's more like what I'd expect if thermal issues were under control. I don't think of processing units as "wearing out" from use. I don't think Pixar throws out their rendering farm after each movie.

[Nena Trinity]

Recommended to do all of the following:

• Unigine Valley and/or Unigine Heaven (Furmark is not 100% necessary, but is also an option)
• Cinebench
• Regular use / gaming

Test one card at a time; not all problems can be detected or arise when running GPU's in Crossfire (or SLi).

 

I am buying a used MSi GTX 760 for a friend (who lives like 2 hours drive from me), and I'll have to test it -- since the original owner no longer has a PC with a proper PSU to test it...

 

This but my Nvidia card fails that test on Cinebench always!

[AMaleWhale]

Even if they never overheat/run too close to thermal limits?

That's more like what I'd expect if thermal issues were under control. I don't think of processing units as "wearing out" from use. I don't think Pixar throws out their rendering farm after each movie.

 

The thing with mining is that most miners don't really care for the cards themselves; usually they are just crammed as many as possible per CPU, running at absolute max and getting as close to overheating as possible without actually doing it, to get the maximum performance possible, there is little regard for the cards longevity. As long as it can calculate for them; it's fine. They don't care if the extended heat causes other errors, like artifacting or other errors that can occur onther parts than the GPU itself.

[boarder2k7]

The thing with mining is that most miners don't really care for the cards themselves; usually they are just crammed as many as possible per CPU, running at absolute max and getting as close to overheating as possible without actually doing it, to get the maximum performance possible, there is little regard for the cards longevity. As long as it can calculate for them; it's fine. They don't care if the extended heat causes other errors, like artifacting or other errors that can occur onther parts than the GPU itself.

Okay thanks, that makes more sense. I figured it would all come back to heat issues rather than the mining itself.