GPU memory failing? RMA or wait? Can I test myself?

[dr.stevil]

So I got a Zotac GTX970 for my workstation/server a few months ago. For the most part, the card has worked just fine... until about a week ago. I came into the computer room to do a few things and noticed that when the display woke up, most of the screen was covered in pink and green colored squares. At this point, the computer had been running for about a week (give or take... its a server so it's always on)..

I tried checking the connections to the display, all of which were good, so I rebooted the machine and everything was fine again for about another week. Yesterday, again, I entered the computer room to get a few things done and noticed the display covered in colored squares.... same problem as before.

So I went through the same process of rebooting and everything was fine again. Putting the GPU under load doesn't seem to cause it (like other GPU's that I've seen go bad) nor can anything else than I can think of. Just letting it idle for a week or so seems to do the trick. At this point, because it seem to be such an intermittent problem, I wondering if I'd be wasting time/money on an RMA... since they'd likely be unable to reproduce the problem.

Is there a utility that exists like memtest that will test GPU memory/etc for errors?

 

[arestavo]

This might help: 5 Ways to Check the Memory of Your Video Card For Errors • Raymond.CC

But I'd try using DDU (free driver cleaner) to remove your old drivers and try the latest, or an older version.

 

[dr.stevil]

Hey thanks for the reply man! Unfortunately (or fortunately, depending on how you look at it) the GPU up and died on me yesterday, so I guess that means I don't have to worry about testing the memory haha. Thanks again.

 

[arestavo]

Bummer. Hopefully you are still under warranty!

 

[dr.stevil]

Yeah it is... its only a few months old. I submitted an RMA request, but I got the card on JET.com. I'm hoping that Zotac honors their warranty