Troubleshoot PC Turns Off Whilst Gaming

[Theoretical]

PC Specifications:

CPU: i7 4790K
GPU: Nvidia GTX 970
PSU: Corsair CX750 750W
Motherboard: Gigabyte Z97-HD3
RAM: 16Gb

Hi, so for a while now I have been experiencing an issue where I would be in game and suddenly my PC would turn completely off, an example of game I have played where this has occurred was DayZ SA.

I first experienced this issue a few months back but took little notice and uninstalled steam because of college work, however recently I tried playing DayZ SA again and the issue occurred again.

I have in the past tried monitoring CPU and GPU temps whilst in game and they were normal when the issue occurred.

I have also considered it might be my PSU causing the issue but considering reliability and stability of corsair PSU's I have not concluded that this is the issue yet.

Any help to assist me in resolving my issue would be very appreciated.

 

[SHvFl]

Cache + GPU + CPU or CPU + Cache + GPU + System Memory?

EDIT:
Just stress tested CPU + Cache + GPU. Test ran for 00:01:55 hrs (1 minute and 55 seconds) until computer turned off. CPU Temperature: 63 Degrees Celsius. GPU Temperature: 68 Degrees Celsius.

Is that cpu cache? If yes test only cpu and cache. I assume you can't test cache alone but if you can do that.

Every time you test the cache the pc shutdowns in minutes but i am yet to figure out which cache you test tbh.

Do you use Process Lasso or any other weird crap that controls speeds and system processes?

EDIT: Please reset your bios to start with before doing anything else.

 

[Theoretical]

Is that cpu cache? If yes test only cpu and cache. I assume you can't test cache alone but if you can do that.

Every time you test the cache the pc shutdowns in minutes but i am yet to figure out which cache you test tbh.

Do you use Process Lasso or any other weird crap that controls speeds and system processes?

EDIT: Please reset your bios to start with before doing anything else.

I don't not use any Process Lasso or anything like that.

I'm also not sure what is meant by "cache" it doesn't say.

I haven't actually changed any BIOS settings but I will reset it anyway, also for some reason you can test cache alone.

 

[SHvFl]

I don't not use any Process Lasso or anything like that.

I'm also not sure what is meant by "cache" it doesn't say.

I haven't actually changed any BIOS settings but I will reset it anyway, also for some reason you can test cache alone.

Ok check cache alone to see and do the full night ram test when you can but remember to reset the bios first.

 

[nclr11111]

I don't not use any Process Lasso or anything like that.

I'm also not sure what is meant by "cache" it doesn't say.

I haven't actually changed any BIOS settings but I will reset it anyway, also for some reason you can test cache alone.

According to Aida´s supportforum crashes during "Cache stress test" can be due to a number of reasons:

It could be due a number of issues. CPU cache malfunction, CPU hardware failure, RAM CRC error, motherboard failure, power supply unit failure, too excessive CPU overclocking, too strict memory timings configured, an overheating system component, etc.

I'd try updating the motherboard BIOS first, then restoring all BIOS Setup settings to their default values. If it doesn't help, then check the temperatures and voltages both idle under load. If they show no issues, then you will need to try replacing parts one by one. It's not easy to diagnose such issues remotely though

Thread.

You should get a result like this if the CPU´s cache is tested.

 

[Theoretical]

I tested the cache on its own for 1hr and 4 minutes the system did not turn off, can we conclude it's the power supply after running memtest?

I also reset the bios settings.

 

[SHvFl]

I tested the cache on its own for 1hr and 4 minutes the system did not turn off, can we conclude it's the power supply after running memtest?

I also reset the bios settings.

It's possible. Your results are really not making any sense.

Do the CPU + Cache + GPU again as before and see if it shutdowns in minutes again. Wondering if it's random or there is a pattern we don't see.
Whatever happens after 15 minutes do Cpu and cache for 15 and then do gpu and cache.

At the end of all this do an all nighter of ram and hopefully it's that.

 

[nclr11111]

Just to conclude!
- Your CPU is within thermal limits!
- Your GPU is within thermal limits!
- Your RAM is reseted in slot one and three and passes memtest?
- Your PSU passed PSU stresstest?
(If not, dl "HWMonitor" and note the voltages in idle for +5V/+12V and make sure there are no values that are way off. Then run Prime 95 first and Furmark once Prime 95 is done.
Make sure +12V stays above +11.4V and below +12.6V, +5V must be above +4.6V and below +5.25V when going from idle to full load. The less drop in +5V/+12V when going from idle to full load the better the PSU performs!
If any of these values are broken when it goes from idle to full load then discard the PSU!)

One more thing!
Move the GPU to the second slot. It will only be x4 in that slot but enough to see if the problem persists. If it does´nt we probably can narrow down the issue to either the PCI-slot itself or the PCI-E 3.0 controller in your CPU failing.

 

[aragornnnn]

how old is your PSU?

 

[Theoretical]

Just to conclude!
- Your CPU is within thermal limits!
- Your GPU is within thermal limits!
- Your RAM is reseted in slot one and three and passes memtest?
- Your PSU passed PSU stresstest?
(If not, dl "HWMonitor" and note the voltages in idle for +5V/+12V and make sure there are no values that are way off. Then run Prime 95 first and Furmark once Prime 95 is done.
Make sure +12V stays above +11.4V and below +12.6V, +5V must be above +4.6V and below +5.25V when going from idle to full load. The less drop in +5V/+12V when going from idle to full load the better the PSU performs!
If any of these values are broken when it goes from idle to full load then discard the PSU!)

One more thing!
Move the GPU to the second slot. It will only be x4 in that slot but enough to see if the problem persists. If it does´nt we probably can narrow down the issue to either the PCI-slot itself or the PCI-E 3.0 controller in your CPU failing.

Hi, I did not run any PSU stress test so I will definitely check those voltages.

I also still have not done the all nighter Memtest.

how old is your PSU?

My PSU is as old as my computer, so around a year and a half. It's a Corsair CX 750.

It's possible. Your results are really not making any sense.

Do the CPU + Cache + GPU again as before and see if it shutdowns in minutes again. Wondering if it's random or there is a pattern we don't see.
Whatever happens after 15 minutes do Cpu and cache for 15 and then do gpu and cache.

At the end of all this do an all nighter of ram and hopefully it's that.

Ok I will try running the stress test on those three again.

I'm starting to think the power supply is the issue personally because when stress testing the CPU + Cache and GPU you are consuming more power than just stress testing the CPU + GPU?

 

[SHvFl]

Hi, I did not run any PSU stress test so I will definitely check those voltages.

I also still have not done the all nighter Memtest.



My PSU is as old as my computer, so around a year and a half. It's a Corsair CX 750.



Ok I will try running the stress test on those three again.

I'm starting to think the power supply is the issue personally because when stress testing the CPU + Cache and GPU you are consuming more power than just stress testing the CPU + GPU?

Not really. We don't know what cache it is but no way it adds more than a few W. Obviously the first thing i told you to check was the power supply but you can't borrow or have one so we have to keep guessing until we are sure it's the psu so you can return it for warranty.
I would have told you take it back and tell them it's bad but wouldn't they check it at the shop first and because it works fairly normal they would say it's fine?
Do what we said and unfortunately it seems time is getting closer to have to take it to a shop to physically access. Not much else we can test without spare parts.

 

[Theoretical]

PSU Voltage IDLE:

+3.3V - 3.05V

+5V - 5.040V

+12V - 12.02V

PSU Voltage LOAD (Stress Testing CPU + GPU):

+12V - 11.808V

+5V - 5.010V

+3V - 3.285V


The system turned off again whilst running a stress test on the CPU + GPU + Cache after stressing for 1 minute and 41 seconds.

Just from what's happened so far, what do you guys think the issue might be?

 

[nclr11111]

PSU Voltage IDLE:

+3.3V - 3.05V

+5V - 5.040V

+12V - 12.02V

PSU Voltage LOAD (Stress Testing CPU + GPU):

+12V - 11.808V

+5V - 5.010V

+3V - 3.285V


The system turned off again whilst running a stress test on the CPU + GPU + Cache after stressing for 1 minute and 41 seconds.

Just from what's happened so far, what do you guys think the issue might be?

Did you run Prime 95 to test and if so did you use Small , Large or Blended FFT´s in torturetest?

If you used another software please DL Prime 95 and run torturetest in above order (Small, Large and Blended FFT´s) and keep a close eye on temps via HWmonitor since those tests will push your hardware.

Regarding your voltages i´m not too enthusiastic! +12V drops by 0.2 in load which makes me wonder if it might not be able to keep up when the load continues and therefore is the problem in your system.... Also the +3.3V shows strange readings in IDLE. +/-5% is usually acceptible but this is way more and in idle? Kind of confusing. Might be reading wrong!?

You are sure you haven´t mixed up idle and load values??

This is what mine looks like after running Prime95 Small FFT´s:

First run Prime95 Small FFT´s during 1min and make sure HWMonitor is recording - Take a screenshot as above.
Then do the same with Large FFT´s, then Blended and upload the screenshots here.

If your comp manages those 3x1min tests let it run from the top (Small, Large, Blended) ~30min each or until it crashes. Take note which test you run and for how long before any eventual crash occurs and post it here.

Ps. I think your PSU is ******! But since you have no other to try it´s just trial and error until proven to be faulty.

 

[Theoretical]

Hi just to update on the situation, I took it to a computer repair shop and he tested it leaving it running for a few days on a graphics benchmark, he said it crashed within a day however he then said he replaced it with a 1060 6GB and ran it for a few days with no crash concluding it was the GPU. (This was the only test he ran on it)

I just wanted to get your thoughts on this as I always thought it might have been the PSU and don't want to buy another GPU until I know for certain.

 

[SHvFl]

Hi just to update on the situation, I took it to a computer repair shop and he tested it leaving it running for a few days on a graphics benchmark, he said it crashed within a day however he then said he replaced it with a 1060 6GB and ran it for a few days with no crash concluding it was the GPU. (This was the only test he ran on it)

I just wanted to get your thoughts on this as I always thought it might have been the PSU?

He is probably right but you will know when you get it home and try it. Only way because after a point guessing is useless.
It's weird though that he only did this test except if the cache you were testing was the gfx card. Did you tell him the test you do and closes in a minute? That is what he has to try with the 1060.

 

[Theoretical]

I did mention it to him yeah and he did say he used Aida64 but I'm not sure if he ran the exact settings and how long he did that test for.

The only way to know for sure is to buy another GPU but I don't want to waste money on buying one if that is not what's causing the issue, he offered no gaurentee it was the issue and said if it wasn't we would have to bring it back so he could do more tests on it.

 

[SHvFl]

I did mention it to him yeah and he did say he used Aida64 but I'm not sure if he ran the exact settings and how long he did that test for.

The only way to know for sure is to buy another GPU but I don't want to waste money on buying one if that is not what's causing the issue, he offered no gaurentee it was the issue and said if it wasn't we would have to bring it back so he could do more tests on it.

If you have the pc hm take the card off and try the test that shuts the pc in minutes again minus the gpu obviously. If it doesn't shut down then most probably it was the gpu.
I don't believe is the psu because it never shutdown when you did the cpu+gpu test which uses the most power.

Question. IS the 970 out of warranty? How? They usually have 3 years warranty.

 

[Theoretical]

I will do this now and I have only had the computer for a year and a half.

 

[SHvFl]

I will do this now and I have only had the computer for a year and a half.

Then don't buy a new gpu. Send this one for warranty.

 

[Theoretical]

Then don't buy a new gpu. Send this one for warranty.

I tested again, however this time before the test I tried cleaning every bit of dust on it (even though there was hardly any). I ran Aida64 and weirdly it lasted longer when I cleaned it, it lasted around 10 minutes rather than the usual 1 minute before turning off.

I will also be contacting MSI on Monday about getting a replacement, thanks for pointing that out to me.

 

[SHvFl]

I tested again, however this time before the test I tried cleaning every bit of dust on it (even though there was hardly any). I ran Aida64 and weirdly it lasted longer when I cleaned it, it lasted around 10 minutes rather than the usual 1 minute before turning off.

I will also be contacting MSI on Monday about getting a replacement, thanks for pointing that out to me.

You did the test with the gpu on? What about the same test with the gpu off?