Bad graphics card or just bad coincidence?

Xavier1216

Reading about all these great PC games and communities on the forums lately made me want to get back into PC gaming, so I finally decided to spend part of last weekend upgrading the tower that I built in 2010! I was concerned with future-proofing it a bit back then (as well as Final Fantasy XIV 1.0), so my only need this time around was upgrading my graphics card to a GTX 970. The new card arrived this weekend, and I soon had my case open and my old cards replaced. The PC started up, I got my updated drivers through the Nvidia application, and everything looked good. I started up Steam and began downloading the latest client update. Then the system suddenly rebooted. After POST and check disk, Windows came back to the login screen. While idling there (and without any user input), I experienced another reboot. No blue screen, just powering down into a reboot.

I tested each DIMM of RAM separately to rule that out, and I experienced the same reboots with each of the three while idling on the Windows login screen. With each restart, my POST code worked its way up to FF on the on-board reader and the recovery check disk passed. After shutting down and adding back in all my RAM, I re-seated the 970 and started it back up. I was able to log in and idle on desktop for about an hour without any problems, so I thought maybe that had done the trick. 3D Mark ran to completion (I have never seen Fire Strike run so smoothly), so with that settled, I updated and started up Steam and began to download Skyrim. The Steam client seemed to be hanging, so I pulled up Task Manager. Task Manager came up fine; but when I hit show processes from all users, both Task Manager and Steam ended up in "Not Responding". Suddenly, another reboot. Frustrated, I shut down and pulled the plug for the night.

At this point I'm ready to believe that my issue could be with the 970 itself, because that's the only hardware delta. The system was running just fine a couple days before, but I'm not ready to dismiss the improbable option that any other piece of hardware could have failed overnight. I've never experienced a bad piece of hardware or anything at all like this, though, so I wanted to ask for second opinions / more suggestions before contacting Amazon for a replacement. I plan on pulling the card and popping one of my old 480s back in tonight and running some more tests, but I'm not sure what to do other than check the Event Viewer, run 3D Mark again, and just try to generally use the computer.

Does anyone have any suggestions? In case it helps, I've put my tower configuration behind the spoiler tag. Thanks in advance.

Gigabyte GA-X58A-UD7
Intel i7-980X
Mushkin 12GB (3x4GB) 240-pin DDR3 1333 RAM
2x SLI PNY XLR8 GeForce GTX 480 EVGA GTX 970 FTW+ (04G-P4-3978-KR)
Corsair TX850M 850W PSU
Windows 7 64-Bit

davidsdurions

First thing I'd do is uninstall the drivers and reinstall then from scratch. If that doesn't fix the problem I'd suspect the PSU of crapping out so go about checking that.

After that I'd put the old cards back in and make sure it is all working like before to determine of the 970 is borked. If it seems that way then it's time for an RMA.

Draygo

850W PS should be enough

but it is possible it is not satisfying the power requirements (faulty?) Do you have another power supply you can give it a go on and see if there is any difference?

fightinfilipino

speaking of power, the PSU is one of those Active Power Factor Correction types which can be sensitive to power grid fluctuations. do you have your computer plugged in to a UPS of some kind? does your place of abode experience issues with electricity fluctuations?

the problem you're describing is one i was experiencing quite a bit because of my APFC PSU. getting a UPS that provided a sine wave current (like this one) fixed the problem.

i know this is a pricey buy, so it may be worth checking your house's electrical first for fluctuations.

Xavier1216

My PC is plugged directly into the wall outlet in my apartment. It's kind of an old building, but the landlord had the electrical system (and a bunch of other stuff) updated when he bought the place about a decade ago. I've also been living here for about a year and using the computer in the same room without any problems, so I would think it would have to be a relatively new change or issue. Thanks for the info, though; I'll definitely have to read up on that (knowing more about it couldn't hurt).

I did a clean re-install of the Nvidia drivers and tested my machine tonight with the old graphics card, and I ended up getting another reboot fairly quickly. It looks like it's not the new card, at least, so I put the 970 back in. I had a possibly-useless thought that it might be the PSU's PCI-E power cables, so I swapped those out and marked the ones I was using before.

It's running pretty stable right now, but I'm not sure for how long. I figured taxing the CPU could be a good test, so I wrote a quick multithreaded program to run on a loop and mark a series of files every five minutes while working the CPU to 100%. I'm going to run that until I shut down for the night and see if it triggers anything. In the meantime, I might be able to borrow a friend's PSU this weekend and run some more tests.

Thanks for the replies, everyone! I'll have to do some more investigative work on this one and let you know what I find.