PC crashing/monitor freeze up

[ziggynerja]

Problem:

- I've been having an issue over the last couple of months with my computer whilst gaming. The screen freezes into a pink, blue or green screen, with vertical lines.
- It ONLY seems to occur whilst gaming, very randomly. I'm finding it really hard to troubleshoot because it can happen 2-3 times a day, and then it'll be completely fine for a few days. It only occurs when gaming, and usually when I've been playing for a good hour or so, at least.

Temps:

CPU
Idle: 40-50c
Peek(gaming): 40-60c

GPU
Idle: 50-60
Peek(gaming): 65-75c

I was under the impression these temps were normal.. the GPU never goes above 80c, and the CPU usually hits 55c at most.

What I've done:

- I've tried to reinstall drivers, different versions of drivers, and even an entire format of my OS.
- I've tried to move my GPU to the second slot on the mobo, and I've tried different combinations for my ram (different slots, single stick..etc).
- Memtest no errors
- FurMark no errors

Spec:

GPU: Sapphire R9 280X
RAM: Corsair Vengeance 16gb
Mobo: Gigabyte Z77X-UP4 TH
PSU: Coolermaster 600w silent pro
Case: BitFenix Neos
CPU: i7 2600
Storage: Samsung SSD 850 Evo

I've no idea what else to try now, I'm worried it might be my GPU failing, but is there any way in which to test this? Any help at all would be massively appreciated!

[satrow]

See if you can get some details from Reliability History (W7 method: http://windows.microsoft.com/en-US/w...bility-Monitor).

[MrComputerSaint]

Hmm, wish I could be more of help. I've had very similar issues to you in the past, not on that hardware, but a while ago. I think one time was the card itself and the other time it was drivers.

You say the screen freezes, does the computer still run, just without graphics?

Have you checked event viewer to see if there is an error message? Maybe there is a dll file that needs re-registering

[ziggynerja]

See if you can get some details from Reliability History

Thanks for the responses!

I've not seen the Reliability History before. Since yesterday morning, I took my card out, and I'm now gaming (barely) with my on-board graphics. I've had no crashes yet. Maybe I should consider putting the 280x back in, and wait for a crash, restart the pc, and then check the history to see if any errors occurred?

Hmm, wish I could be more of help. I've had very similar issues to you in the past, not on that hardware, but a while ago. I think one time was the card itself and the other time it was drivers.

You say the screen freezes, does the computer still run, just without graphics?

Have you checked event viewer to see if there is an error message? Maybe there is a dll file that needs re-registering

Well, unfortunately there have been situations where I'm in a skype call, and it would freeze, and I'd still be able to chat, meaning the PC IS still running. However, there have also been times where the entire PC freezes up, and my game sound stops.

I actually haven't tried the event viewer. I think, as I said above, I will refit my graphics card, and wait for the freeze to occur again, so that I can use the two tools you have both suggested to look for any errors.

Thanks for the help!

[satrow]

Yes, refit the 280x.

I have a feeling that it's a TDR issue (0x116/0x117), best troubleshooting for this that I know of is here: http://www.sysnative.com/forums/bsod...116-0x117.html

It's something I get myself sometimes, both with my HD7850 and R9 270 (though I might have fixed it, no issues in ~2 months - but no gaming either in that time).

[Thorsson]

If it only happens in games, unless it's a specific game, then could be it's happening when the card heats up. It could be the memory on the card rather than the GPU itself. When this happened to me it was a sign the card was on its way out. I'd use some benchmarking software to see if it happened during a particular operation.

[MrComputerSaint]

Hmm, if it happens and your PC can still respond there could be a possibility it might not be the card, but be the PSU not giving the right voltage perhaps... Maybe it drops just for a moment enough to cause a problem of freezing application.

I wonder if there is any monitoring program to show the voltages on graphics card and PSU and see if they're stable... and a logging tool.

[TheDutyPaid]

I think this points to a graphics card which the hardware is dying, not a software fault. A few years ago,

I had a PC that would crash when opening youtube full screen. A friend who had the same graphics card lent me theirs and my machine was fine, tried mine on theirs, and it crashed.

As you have said your on board does run games, all be it slow, shows that your PC health is ok.

[ziggynerja]

Yes, refit the 280x.

I have a feeling that it's a TDR issue (0x116/0x117), best troubleshooting for this that I know of is here:
It's something I get myself sometimes, both with my HD7850 and R9 270 (though I might have fixed it, no issues in ~2 months - but no gaming either in that time).

I've refitted the card, and have been running 3D intensive applications/games since yesterday morning. I've not froze once yet (typical). So will waiting for that so that I can look for any error messages!

@Thorsson That does sound likely! I've been reading a bit into it, people seem to have solved that by flashing their bios? http://www.overclock.net/t/1459922/s...blem-good-news

@MrComputerSaint I've not been able to find any tools which have this monitoring feature unfortunately. The best I've used is Speccy, which can show the voltages.

@TheDutyPaid Well that's the problem, It doesn't happen frequently enough for me to test it properly For instance I played GTA V at 4k res yesterday for a few hours, and no crash occurred on my R9 280X. It happens very randomly. I'll do the same tests today on the card, I'm fairly sure it will crash again. It rarely lasts two days without a crash.

[ziggynerja]

Update:

It just crashed. I can't see anything within the Reliability Monitor, nor within the Event Viewer. The only error I received within the Event Viewer, was "Kernel Power", which I believe is because I had to force restart the PC to turn it off.

Sigh.. is there any way I can test whether the PSU is the culprit? I have a feeling its the vram overheating.. Should I attempt flashing BIOS? I've never done that before!

[TheDutyPaid]

Just some random thoughts,

With your PSU, 80% of 600 is 120, so true power is around 480W Which should be enough, is this enough?

Just checking your card needs two 8 pin, is this off the same rail? If so try one from each / if it is one from each try the same rail.

I read the spec of what you have, but do you have anything else plugged in. water cooling pump, loads of fans, LCD fan speed screen, disco lights ect?

[Thorsson]

Update:

It just crashed. I can't see anything within the Reliability Monitor, nor within the Event Viewer. The only error I received within the Event Viewer, was "Kernel Power", which I believe is because I had to force restart the PC to turn it off.

Sigh.. is there any way I can test whether the PSU is the culprit? I have a feeling its the vram overheating.. Should I attempt flashing BIOS? I've never done that before!

The only certain way to test whether a PSU is the culprit is to plug another PSU in.

Flashing a BIOS can kill your card if it goes wrong and you won't (officially) be covered by warranty. That said, it's not usually a problem. Just you need to be aware that it's a non-trivial decision. That thread is over a year old; if the BIOS were a problem then I would have hoped that Sapphire would have an official BIOS update by now - there's a bunch here.

Also a good description of what to do in this thread.

[Domestic_Ginger]

GPUz can log temps, volts etc and log to text file in the event if a crash.

The 7970 had dual BIOS; I presume the 280 is the same so flashing isn't the end of the world.

I would run unigine valley for a bit and see how it does. You could try blowing out the dust bunnies.

[rapidracer10]

A friend of mine had the exact same problem, he had a MSI R9 270X, with a corsair CX500 PSU
Just returned the card to amazon, im not sure if it was the PSU as he just gave up and stuck with the iGpu for now

[lankeyphil]

What game you playing at the time?
Maybe worth checking the voltage to the car ld I'd maybe take the pc apart and give it a good clean up on connections etc.

Could also be either the card or the psu on its way out.

I'd try give it a good clean up first internally

[Thorsson]

Getting rid of dust is always a good idea. Some of the video card cooling solutions seem to be dust traps.