GPU Failing?

[Ziggy0511]

Specs:
i5 3570k (stock speeds atm)
Xigmatek Dark Knight CPU Cooler
ASUS P8Z77V-Pro
MSI GTX 780 Lighting
Antec 750W HCG
Intel 256 GB SSD (Games)
Samsung 128 GB SSD (OS)
Seagate 3TB HDD (Media)



Problem:
Hard restarts during games/benchmarks. Computer shuts off like the power cord has been pulled and restarts a few seconds later (Event Log: Kernal Power 41(63)). It first started happening intermittently a few months ago (restart every couple of days) and has increased to the point where i can no longer run any game or benchmark without a restart. It only lasts 3-5 mins now before restarting. I have performed all sorts of tests trying to determine the cause which most likely would seem to be a heat or power supply issue. Multiple passes on memtest came back clean. Ran prime95 and intelburntest for extended periods just fine. Ruling out memory and CPU (which is at default clock speed for the moment) leaves me to the GPU and PSU as the likely sources of the error. This is where I came across some interesting. As I was running furmark to stress test the GPU i was getting fairly consistent restarts a couple of mins though. However on one run i decided to manually run the GPU fans at 100% for the entirety of the test and what do you know it makes it though the whole test for the first time in days. This made me rule out the PSU as the culprit because the system made it though the whole test just fine. So I started testing as i slowly decreased the fan % down each test. Basically it crashed everytime the fans were significantly lower than 100%. So it must be overheating right? No, well maybe. Obviously I was keeping an eye on the GPU temps as I was running the benchmarks. Most of the restarts are happening when the card is 60-72C which from my understanding is pretty normal GPU temps. With the fans at 100% the GPU stays right around 56C max. I dont think it is drivers as this has been happening though multiple driver versions. It seems like it must be something to do with the GPU, but i just dont buy that it is overheating. It doesnt throttle or drop fps before it happens it just goes down hard. I have thoughly cleaned the heatsinks and whatnot with compressed air.


I have put it a online ticket with MSI to troubleshoot the card, possibly RMA.

Any other ideas?

Thanks in advance for any consideration and help.

[Ziggy0511]

Forgot to mention I am running Widnows 8.1 and am currently on Nvidia driver 355.82.

[Grady666]

Specs:
i5 3570k (stock speeds atm)
Xigmatek Dark Knight CPU Cooler
ASUS P8Z77V-Pro
MSI GTX 780 Lighting
Antec 750W HCG
Intel 256 GB SSD (Games)
Samsung 128 GB SSD (OS)
Seagate 3TB HDD (Media)



Problem:
Hard restarts during games/benchmarks. Computer shuts off like the power cord has been pulled and restarts a few seconds later (Event Log: Kernal Power 41(63)). It first started happening intermittently a few months ago (restart every couple of days) and has increased to the point where i can no longer run any game or benchmark without a restart. It only lasts 3-5 mins now before restarting. I have performed all sorts of tests trying to determine the cause which most likely would seem to be a heat or power supply issue. Multiple passes on memtest came back clean. Ran prime95 and intelburntest for extended periods just fine. Ruling out memory and CPU (which is at default clock speed for the moment) leaves me to the GPU and PSU as the likely sources of the error. This is where I came across some interesting. As I was running furmark to stress test the GPU i was getting fairly consistent restarts a couple of mins though. However on one run i decided to manually run the GPU fans at 100% for the entirety of the test and what do you know it makes it though the whole test for the first time in days. This made me rule out the PSU as the culprit because the system made it though the whole test just fine. So I started testing as i slowly decreased the fan % down each test. Basically it crashed everytime the fans were significantly lower than 100%. So it must be overheating right? No, well maybe. Obviously I was keeping an eye on the GPU temps as I was running the benchmarks. Most of the restarts are happening when the card is 60-72C which from my understanding is pretty normal GPU temps. With the fans at 100% the GPU stays right around 56C max. I dont think it is drivers as this has been happening though multiple driver versions. It seems like it must be something to do with the GPU, but i just dont buy that it is overheating. It doesnt throttle or drop fps before it happens it just goes down hard. I have thoughly cleaned the heatsinks and whatnot with compressed air.


I have put it a online ticket with MSI to troubleshoot the card, possibly RMA.

Any other ideas?

Thanks in advance for any consideration and help.

Could be a number of things: You have a 750W+ PSU, so there *shouldnt* be anything going on w/ the PSU Overloading and Cutting DC Power(which will instantaneously shut down the PC); Have you tried using a different keyboard/mouse and switching them out(as well as any other peripherals running off the US Bus(USB);

Also, have you tried doing a clean install w/ Display Driver Uninstaller: http://www.guru3d.com/files-details/...-download.html
Do the clean install option(Itll say "for installing a new graphics card"), uninstall your current drivers, save the installation file for the latest Nvidia WHQL driver on a flash-drive,Shut-down, Reseat your Graphics card, and run the installer off the Flash-drive then reboot once more- That's the most thorough way to cleanly uninstall and re-install your Graphics card drivers; It might work, might not, but its worth a try-

Give me an update if/when you try this

[Ziggy0511]

Tried a clean driver install. Unfortunately the PC still restarted while running furmark. It did seem to stay up for a noticeably longer amount of time though (7.5mins as opposed to 3-4, could have been due to the cold boot though). I repeated my 100% fan test and the system stayed up for the whole 10 minute run again. Temps were about the same. The restart occurred at 72C. The 100% run floated between 57-58C once it got going.

I haven't messed with any of my peripherals yet aside from unplugging my 360 controller. All that is connected is keyboard, mouse, headphones, speakers, and a mic. The only other keyboard and mouse i have is a wireless combo that runs off a USB adapter deelibop. Would it be better to plug my current keyboard/mouse into other USB ports? or should I setup the wireless ones and see if anything changes that would require installing the wireless software as well. Regardless, I dont think that they are the problem though it cant hurt to test it out.

[13uckFUtter]

780 lightning shouldn't being dieing that fast, but it's always possible.

[Ziggy0511]

780 lightning shouldn't being dieing that fast, but it's always possible.

Yea I purchased the card new in Jan 2014, luckily it is still under warranty though (2 years). I am considering a clean OS install. Would that effectively rule out any driver or software issues?

[Marrv]

Also worth considering Mobo issue, not that they are easy to diagnoise (aka rule everything else out - by putting known good parts in & testing, which is a pain to most people who do not have spare parts lying around....), but when it has issues it can cause all sorts of mayhem including what your having

Just to be troublesome - while you have tested that your PSU can supply the power to the components adequately it may not be providing steady power (it could be spiking, which causes issues, if you know about OC-ing before all the modern tools you did it by adjusting voltages to the ram modules, if the psu is not supplying correct voltages it will cause a hard crash. This is very rarely the case these days as most psu's have gotten better but some still get through).

Also have you disabled automatic restarts to see if the is a full error report created/any additional information (judging by what you done above I assume you have, but best be sure)