Need help: PC shut-downs (no BSOD) and stutter.

[yoadknux]

Hi guys,

I'm trying to troubleshoot my girlfriend's PC. It's a nice computer with i7 2600k, z68ma motherboard, 970 GTX, Seasonic X-Series 650W gold, 2x8 ram (I think G-Skill Aegis). It has Windows 10 installed on 500gb ssd (relatively new) and it also uses some 500gb wd blue for storage.

The exact problem: The computer hard crashes into a black screen and reboots under gaming load, sometimes. It seems to happen randomly, things can run fine for weeks then crash. Additionally, the sound itself seems to stutter at times (usually associated with the crashes).

The components are not overclocked. I disabled XMP on ram to run it at stock speed. Temperatures, I think, are fine, even under load. Here are some of the things I tried to troubleshoot:
- GPU testing: I tried Heaven/FireStrike Ultra a few times. Heaven had runs of about 4 hours that passed. FireStrike ultra I didn't run too many times, but it passed once or twice with no crashes.
- CPU testing: I tried Prime95, also a total of about 5 hours. Mostly blend, some maximum power too.
- RAM testing: Still haven't tried Memtest, I intend to.

There are no BSODs at all, so it's hard to find the exact problem. Here are some things I intend to try:
- Run Memtest. Although it doesn't really feel like a RAM issue.
- Install external sound card and disable onboard. I think maybe the sound card drivers cause this issue.
- PSU issues are also associated with those type of crashes, and although this PSU is kinda old, it should be of decent quality, and this hardware doesn't consume a lot of power.
- HDD/SSD issue? The SSD is new, HDD is old, HDD is slow as hell, might be faulty, but I don't think it has drivers installed onto it.
- Hopefully it's not the motherboard/cpu dying. It will be a pain.

Any other tips guys? Help appreciated.

 

[trents]

Run Crystaldiskinfo to check the SMART drive info on both drives. You can toggle between drives with the > arrow symbol in the CDI user interface. If you get any colors besides blue (either yellow or red) you likely have a failing drive. With the age of that system it would not surprise me if the spinner hard drive is starting to have problems.

I would certainly be suspect of the PSU because of age. Do you have one you can swap in temporarily to test that? And I would certainly run memtest86.

I would also try removing the video card and see if the instability disappears if using the IGP. This can also lesson the load on the PSU if power delivery is the problem. In fact, I would remove all unnecessary components to see if the instability disappears and if so, add them back in one by one.

 

[EarthDog]

CPU testing: I tried Prime95, also a total of about 5 hours. Mostly blend, some maximum power too.

You should actually test the CPU, now. Small FFT.

In P95, people typically leave Large alone. Blend tests more memory and little CPU. Small FFT tests mostly CPU, little RAM.

Try running P95 and a GPU benchmark (Unigine Heaven or something similar you can loop) and see if it craps out. If it doesn't, then chances are that PSU is ok as gaming and a 'cpu' stress test won't use that much power.

PSU issues are also associated with those type of crashes, and although this PSU is kinda old, it should be of decent quality, and this hardware doesn't consume a lot of power.

That doesn't mean it isn't part of the problem...

That PSU was a high-quality unit back in the day...

Have you......

...checked event viewer to see what the issue was?

...tried an OS reinstall? Sounds almost software related because of the sound.

...tried reinstalling/updating the audio drivers?

 

[Mr.Scott]

Black screen and sound stutter is a GPU driver crash usually.
Best guess is your 970 is getting ready to take a dump.

 

[Bill_Bright]

- RAM testing: Still haven't tried Memtest, I intend to.

Make sure you use PassMark’s MemTest86. There are many testers with very similar names but this is the only tester currently under regular development and frequently updated for the most recent hardware.

That said, no software based memory tester is conclusive. So testing with one stick at a time may help (though sadly, many will run fine alone but fail when paired). So testing with all different RAM is often the best bet for normal users - though obviously this may not be an option.

To conclusively test your RAM, you need to use sophisticated and very expensive test equipment, like this $2,495 Memory Tester (and that's for the cheap model)! So it is usually easier (and cheaper!) to swap in known good RAM and see what happens.

I would also swap in a spare PSU since everything inside the case depends on good, clean, stable power. Is the computer on a "good" UPS with AVR (automatic voltage regulation)? Note that power issues often result in no warnings, no BSODs, or crash dumps as they occur so suddenly, the OS has no time to issue an warning or write any dumps. While that is a good PSU, even the best makers will occasionally make a lemon. Or the FedEx driver might have drop-kicked the box. Or Mother Nature might have tossed an excessive surge down the line (referring back to that UPS).

Temperatures, I think, are fine, even under load.

Specifics would help as these symptoms are common with heat related issues too. And of course, the CPU is not the only heat sensitive device in there. Is the case free of heat-trapping dust? Is there a decent amount of air flowing through the case?

 

[yoadknux]

Thanks for the comments guys. You've been super helpful. I'll list all the things I've done thus far:
- Removed the old HDD
- Disabled on-board sound
- Ran Memtest, Prime95 Small FFT, Heaven.

Thus far, no crashes, but a day of normal usage is a sample size. Will keep updating this topic. As for the specific comments:

Run Crystaldiskinfo to check the SMART drive info on both drives. You can toggle between drives with the > arrow symbol in the CDI user interface. If you get any colors besides blue (either yellow or red) you likely have a failing drive. With the age of that system it would not surprise me if the spinner hard drive is starting to have problems.

I would certainly be suspect of the PSU because of age. Do you have one you can swap in temporarily to test that? And I would certainly run memtest86.

I would also try removing the video card and see if the instability disappears if using the IGP. This can also lesson the load on the PSU if power delivery is the problem. In fact, I would remove all unnecessary components to see if the instability disappears and if so, add them back in one by one.

I now removed the HDD. I tried Crystaldiskinfo for the SSD and it seemed fine. Don't have a spare PSU sadly. Doing gaming load using the on-board won't be easy.

You should actually test the CPU, now. Small FFT.

In P95, people typically leave Large alone. Blend tests more memory and little CPU. Small FFT tests mostly CPU, little RAM.

Try running P95 and a GPU benchmark (Unigine Heaven or something similar you can loop) and see if it craps out. If it doesn't, then chances are that PSU is ok as gaming and a 'cpu' stress test won't use that much power.

That doesn't mean it isn't part of the problem...

That PSU was a high-quality unit back in the day...

Have you......

...checked event viewer to see what the issue was?

...tried an OS reinstall? Sounds almost software related because of the sound.

...tried reinstalling/updating the audio drivers?

Thanks for the comment. I followed your suggestion and went with 2 hours of small FFT, then about 2 hours of small FFT + Heaven loop at the same time. Temps seem fine and there were no crashes.
OS is relatively fresh, about a month old. I now disabled on-board sound and plugged an external sound card, and DDU'd the realtek drivers out. As for event viewer, I didn't try it.

Black screen and sound stutter is a GPU driver crash usually.
Best guess is your 970 is getting ready to take a dump.

Could be. If I were losing signal, with other stuff still running, I'd say GPU, but the computer crashes and restarts, so I can't say for certain.

Make sure you use PassMark’s MemTest86. There are many testers with very similar names but this is the only tester currently under regular development and frequently updated for the most recent hardware.

That said, no software based memory tester is conclusive. So testing with one stick at a time may help (though sadly, many will run fine alone but fail when paired). So testing with all different RAM is often the best bet for normal users - though obviously this may not be an option.

To conclusively test your RAM, you need to use sophisticated and very expensive test equipment, like this $2,495 Memory Tester (and that's for the cheap model)! So it is usually easier (and cheaper!) to swap in known good RAM and see what happens.

I would also swap in a spare PSU since everything inside the case depends on good, clean, stable power. Is the computer on a "good" UPS with AVR (automatic voltage regulation)? Note that power issues often result in no warnings, no BSODs, or crash dumps as they occur so suddenly, the OS has no time to issue an warning or write any dumps. While that is a good PSU, even the best makers will occasionally make a lemon. Or the FedEx driver might have drop-kicked the box. Or Mother Nature might have tossed an excessive surge down the line (referring back to that UPS).
Specifics would help as these symptoms are common with heat related issues too. And of course, the CPU is not the only heat sensitive device in there. Is the case free of heat-trapping dust? Is there a decent amount of air flowing through the case?

Thanks for bringing it to my attention. I ran Memtest for 6 hours, in which it completed 2 passes with no errors. My girlfriend doesn't have as much patience as me, had to test the sticks together. It also passed a few hours of Prime95 blend. Errr, those mega ram testers are out of my league, lol
Don't have a spare PSU at the moment. I might get one if I see things still act weird. UPS with AVR? hmm... more like straight to the wall.
As for temperatures: GPU at heaven loop approaches 70c, and is about 71c while gaming. i7 2600k with small FFT hits about 70-75c, around 50c-60c while gaming. Don't really have information about other temperatures. PSU is on a different chamber (it's a dual-chamber case). Don't know about RAM temps. Airflow could be better, exhaust is good, intake is so-so, filters could use some cleaning, but one crash for example occured about 30 minutes after a gaming load, so I don't think it's temperature related (unless there's some overheat which induces delayed issues).

 

[Bill_Bright]

Don't have a spare PSU at the moment. I might get one if I see things still act weird. UPS with AVR? hmm... more like straight to the wall.

Perhaps swapping with another PC?

i7 2600k with small FFT hits about 70-75c

According to Intel, the maximum Tcase temps is just 72.6°C. So this may account for the instability issues. Opening the case side panel and blasting a desk fan in there to see how it holds may provide more clues.

 

[RJARRRPCGP]

Core temps are not Tcase. You can get the actual core temps, which the temp numbers above, don't look too bad, with RealTemp.

 

[EarthDog]

https://www.overclock.net/forum/5-intel-cpus/307372-core-temp-tcase-vs-tjunction.html#post_3561245

 

[yoadknux]

The problem is solved, although I still don't know what was the exact reason.
I've done two things, the first one was to remove the old HDD and keep only the new SSD (no fresh install or anything like that, just plainly disconnecting the power and sata cables). The second is DDU the audio drivers out and use a very basic usb sound card.
No stutters, no shutdowns, everything is working properly.

 

[Alaric]

Maybe Windows 10 updates? It likes to mess with drivers.

 

[UltraTaco]

Black screen and sound stutter is a GPU driver crash usually.
Best guess is your 970 is getting ready to take a dump.

This is true. When I OC gpu and it crashes, sound would stutter and in better caee, you get crash to desktop and says driver has recovered. In worse cases, you get black screen and she hangs. Don't remember if mine would restart..

Hopefully the mysterious resolve is permanent.

 

[Zerileous]

Probably the sound driver. Of course its never a bad time to clean dust out of everything, ensure the CPU cooler is clean, could even go with some fresh thermal paste.