Computer Crashing (Blackscreen, hard shutdown required)

[Dooms]

Hey guys,

Last couple of weeks my machine has been causing me grief and I've been ignoring it hoping it will go away so I don't have to deal with it... but alas it's got to that point.

<< Spec is in the side (Win 8.1, 8350, 16gb ram, MSI r9 290, Sabertooth 2.0)

Problem: When gaming (desktop work is fine it seems) computer will randomly black screen crash, sound will 'stick' and machine will be completely unresponsive and will require a full hard shutdown - but sometimes only after hours, sometimes soon as I boot up.

Tried: Temps are fine, reset all BIOS to optimized defaults to remove OC, removed OC from GPU. Rebuilt with fresh Win 8.1, rolled drivers back to old old drivers and also tried the new omega ones that came out the other day.

Sooooo easy things out of the way, time to investigate hardware issues... sigh. Times like these that I envy console gamers (only a little bit )

I'm transplanting my machine into a different case anyway so I'm going to take that time to reseat everything, reapplying paste etc etc. Maybe that will be the magic fix all but doubtful.

Plan of attack after is run stress tests (OCCT for CPU, Memtest86 for RAM, 3D Mark for GPU) find something that reliably kills my machine and maybe highlight where the issue is. Take out stick of ram etc.

Basically I'm here asking if 1) those tools are still relevant and 2) am I missing something obvious?

Cheers.

[MrJim]

My guess would be your PSU - the 290 will be drawing quite a bit of power whilst gaming, which may be why it's then that it crashes. Do you have an alternative PSU you can try? I think OCCT has a PSU power test, but I've never used it myself & don't know how reliable it is.

[kalniel]

Right approach - if it's gaming related then it's worth ruling out stressed hardware and finding a certain cause. OCCT is fine - I happen to use prime95 instead. Memtest 86+ is still great for RAM, though if you only experience crashes in games I doubt it's RAM. 3DMark isn't the most stressful for GPUs though it does gives good range of tests. Furmark is more stressful, to the point cards throttle-down on detecting it to reduce overloading, but it's still a good one.

After trying them individually, try combinations (eg prime95+3D Mark or Furmark) to make sure combined loads are stable from your PSU etc.

Reservations are that it's a variable amount of time - if it works some of the time for a given load then trickier to pinpoint, unless something just happens to be right on the edge and temperature/mood of computer fluctuations are enough to tip it over the edge. Can you check the event viewer to see if there's anything kicking off before the crashes? Could be a software updater or something.

[Dooms]

My guess would be your PSU - the 290 will be drawing quite a bit of power whilst gaming, which may be why it's then that it crashes. Do you have an alternative PSU you can try? I think OCCT has a PSU power test, but I've never used it myself & don't know how reliable it is.

Yeah I was thinking it may be PSU as well but obivously a bit harder to test. The only PSU I have kicking about is a old old old generic which I doubt would handle the draw of both the 8250 & 290. I guess if I eliminate everything else it may be the case... Maybe I'll underclock/undervolt the 290 to see.

Right approach - if it's gaming related then it's worth ruling out stressed hardware and finding a certain cause. OCCT is fine - I happen to use prime95 instead. Memtest 86+ is still great for RAM, though if you only experience crashes in games I doubt it's RAM. 3DMark isn't the most stressful for GPUs though it does gives good range of tests. Furmark is more stressful, to the point cards throttle-down on detecting it to reduce overloading, but it's still a good one.

After trying them individually, try combinations (eg prime95+3D Mark or Furmark) to make sure combined loads are stable from your PSU etc.

Reservations are that it's a variable amount of time - if it works some of the time for a given load then trickier to pinpoint, unless something just happens to be right on the edge and temperature/mood of computer fluctuations are enough to tip it over the edge. Can you check the event viewer to see if there's anything kicking off before the crashes? Could be a software updater or something.

Thanks for pointing me to Furmark, I'll check it out.

I've combed eventlog and all I've got really are warnings about the unexpected shutdown

Right so at least I know what the plan is, first step is to swap everything over to new case, give it a clean and reseat everything. Then testing, then diagnosing. Sigh. I remember a time when I used to enjoy doing this stuff.

[MrJim]

Yeah I was thinking it may be PSU as well but obivously a bit harder to test. The only PSU I have kicking about is a old old old generic which I doubt would handle the draw of both the 8250 & 290. I guess if I eliminate everything else it may be the case... Maybe I'll underclock/undervolt the 290 to see.

I wouldn't risk your expensive kit by using a cheap generic PSU! As you say, try the zero cost alternatives first. Troubleshooting these sorts of problems is a major pain, it's true

[DanceswithUnix]

I usually start with memtest86 and give it an overnight run. Memory dimm connections must tarnish a bit over time, my wife's PC needed a reseat of the ram recently.

[MrJim]

I usually start with memtest86 and give it an overnight run. Memory dimm connections must tarnish a bit over time, my wife's PC needed a reseat of the ram recently.

Most RAM comes with a limited lifetime warranty, so if it is the cause of the problem, it shouldn't be too difficult to replace...

[DanceswithUnix]

Most RAM comes with a limited lifetime warranty, so if it is the cause of the problem, it shouldn't be too difficult to replace...

Unplug and plug it back in seems to sort it for another 4 years though

[Dooms]

Unplug and plug it back in seems to sort it for another 4 years though

Hahaha yeah. Well troubleshooting starts tonight, will update later/tomorrow

[MrJim]

Unplug and plug it back in seems to sort it for another 4 years though

Ah, reminiscent of 'The IT Crowd'

[Dooms]

Well last night I took computer apart (using this opportunity to transplant back to my R4 Case), gave everything a good dusting, reseated everything, reapplied paste etc etc etc - boring boring job.

Started playing games after, hour later, crash. Heart broke. Was hoping for an easy reseat fix.

Didn't have much time left so just did a 20min burn of Furmark, 20min burn of OCCT CPU and 15min OCCT PSU. Temps and everything fine and no crashing. Looks like I need to do some long testing when I have time.

Sigh.

[Dooms]

Ran 5 hours of Furmark last night, no crash.

I'll run Memtest tonight

[Bonebreaker777]

Ran 5 hours of Furmark last night, no crash.

I'll run Memtest tonight

If temp are good and memtest will pass as well, check the HDDs (just tonight I had to take my 1TB SSHD out cause it was crashing my PC).

[Dooms]

*fingers crossed & touching wood!!*

I think its resolved - which is good because I was close to throwing the whole thing out of the window and throwing £1k at a new machine just for the hell of it.

So... Crashing when playing games.
GPU Burn - No Crash
MemTest - Pass (12+ hours)
OCCT CPU test - Pass (12+ hours)
OCCT PSU test - Pass (12+ hours)
HDD SMART - Clear

Still crashing in multiple different games. ASHDFKWDKGTHSEGKWQ!!!!

Pulled a stick of ram, crash, pulled other one, crash.

*start crying*

When replacing ram notice that the USB3 cable was only half in, push it in and I haven't crashed for the 2 days over the weekend.

My only thought is 1) nothing was plugged into front ports so I can't see how it would be an issue and 2) I transplanted my whole build into another case so it doesn't seem like something that would be in old case and new. ?!

I just hope that was it (as unlikely as it is) because I don't want to deal with it anymore.

[kalniel]

When replacing ram notice that the USB3 cable was only half in, push it in and I haven't crashed for the 2 days over the weekend.

My only thought is 1) nothing was plugged into front ports so I can't see how it would be an issue and 2) I transplanted my whole build into another case so it doesn't seem like something that would be in old case and new. ?!

Front panel connectors are notoriously fragile so even if nothing plugged into them then they might cause problems with an unseated connector. However point 2 is exactly what I thought - to have an unseated cable once might be bad luck, but twice... Then again, maybe there's something that's awkward about it that makes it easily unseat.

[Dooms]

Front panel connectors are notoriously fragile so even if nothing plugged into them then they might cause problems with an unseated connector. However point 2 is exactly what I thought - to have an unseated cable once might be bad luck, but twice... Then again, maybe there's something that's awkward about it that makes it easily unseat.

The port is quite lose compared to the normal USB ports on the bottom of the board so that may be it, maybe the plastic has been bent a bit in my old case due to the thickness of the USB3 cables and pulling cables around the back. I'm just going to keep an eye on it but here's hoping it was just that.