The Dreaded PC Reset/Shut Down

Scooby_Doo

Scooby_Doo

Honorary Master
Joined
Sep 4, 2005
Messages
12,190
Reaction score
6,207
Location
In a Box, Next to a tree.
Hi all,

Looking for some additional testing ideas to help narrow down my issue.

My system

Motherboard - Asus P8Z77-V pro
CPU - Intel 3570K
RAM - 8gigs 1600mhz
PSU - Generic 600 watt
GPU - Nvidia 560Ti/ Nvidia 970 GTX
Other - Assorted SSDs and HDDs

When my PC is under GPU load the system resets/shuts down as if the power has been pulled. No BSOD, no memory dumps no errors other than the system lost power for some reason in the event viewer. The loss of power happens in less than 20 seconds from the start of the test using OCCT PSU stress test.

I had this issue earlier last year but i swapped out my 650watt Corsair PSU for a generic 600watt PSU. I sent in the Corsair to Wootware for testing but they were unable to find a problem with it.
However my system was running fine on the generic PSU so I settled on using that.

However I have recently upgraded to a 970 GTX and first try on Dragon Age after installing the GPU it reset. I have done the following tests.

Tested 2 GPUs (560 Ti and 970 GTX) - 560TI works on the Generic and Coolermaster PSU but crashes on the Corsair. 970 GTX crashes regardless of which PSU I try it on.
Tested 3 PSU (Corsair TX650M, Generic 600Watt and a Coolermaster 500 watt)
Tested using prime95 for 6 hours, no errors, CPU never went over 50 degrees C.
Rebuilt the system and "cleaned" it, was not much to clean.
Removed everything but the essentials, SSD, RAM, MB, CPU, GPU.
Tied different power cables to the PSU
Different plug socket and power strip

So before I reveal what I think it is, would anyone else like to provide a suggestion and/or guess as to what my problem is?
 
The 970 is actually more efficient than the 560ti.

Tests:

970 w/ different MB, different PSU

970 w/ different MB, same PSU

Is your CPU OCd?
 
ashfaak said:
The 970 is actually more efficient than the 560ti.

Tests:

970 w/ different MB, different PSU

970 w/ different MB, same PSU

Is your CPU OCd?
Click to expand...

Asus MB does some random automatic O/C as standard. My guess is also motherboard, but for the life of me its the one thing that I don't have a spare of.

Going to send it back to the supplier for testing, R100 bet that they don't find anything wrong with it.
 
Sounds like your system is now running overloaded for the PSU you currently have.

Take into account the other components that use the most power...and you may find that the 970 has just sucked the life out of the PSU when under load.

I would upgrade to at least an 850W 80+ PSU to guarantee output power. Until the n, I would recommend disconnecting any un-necessary components and running the GPU under full load to see what happens
 
CriticalConsumer said:
Sounds like your system is now running overloaded for the PSU you currently have.

Take into account the other components that use the most power...and you may find that the 970 has just sucked the life out of the PSU when under load.

I would upgrade to at least an 850W 80+ PSU to guarantee output power. Until the n, I would recommend disconnecting any un-necessary components and running the GPU under full load to see what happens
Click to expand...

As stated, already tried removing everything but the minimum to get the PC running. 850 seems like over kill for 1 970gtx. Can anyone confirm the requirements? Nvidia states min of 500watts.
 
Automatic OC sucks from my experience, though I have an older generation motherboard (Z68), so perhaps they have improved.

I would reset it to stock at least to rule it out.
 
I run an R9 290 and OC'd i5-2500k on 600W, which should draw more than your system. I would definitely set cpu back to default, automatic OC's almost never work.
 
I'd upgrade to the latest BIOS, then reinstall the GTX 970 and test to see if it still happens. Given all that you've tried, I also have to suspect the motherboard. OCCT also stresses the memory subsystem if I'm not mistaken, so run Memtest86+ and see if it throws up any memory errors. Use the USB auto-installer option, its quicker than all the others to set up.
 
CataclysmZA said:
I'd upgrade to the latest BIOS, then reinstall the GTX 970 and test to see if it still happens. Given all that you've tried, I also have to suspect the motherboard. OCCT also stresses the memory subsystem if I'm not mistaken, so run Memtest86+ and see if it throws up any memory errors. Use the USB auto-installer option, its quicker than all the others to set up.
Click to expand...

Tried to upgrade and down grade the bios, cleaned the Nvidia drivers and re installed. I have not tried memtest, although prime 95 does some memory testing and that worked successfully for hours.
 
Scooby_Doo said:
Tried to upgrade and down grade the bios, cleaned the Nvidia drivers and re installed. I have not tried memtest, although prime 95 does some memory testing and that worked successfully for hours.
Click to expand...

Prime also doesn't stress things in the same way as other load inducers, so it should only be one of the programs you use for testing. Try out Intel's Xtreme Tuning Utility, Linpack or Sissoft Sanda Lite. If the overclock is somehow not stable, XTU and Linpack will show it up almost immediately, just like OCCT does in your tests.
 
CataclysmZA said:
Prime also doesn't stress things in the same way as other load inducers, so it should only be one of the programs you use for testing. Try out Intel's Xtreme Tuning Utility, Linpack or Sissoft Sanda Lite. If the overclock is somehow not stable, XTU and Linpack will show it up almost immediately, just like OCCT does in your tests.
Click to expand...

I will give those a try, I must mention that the CPU test in OCCT works just fine, its only the PSU and GPU tests that cause the system to die on me.
 
Only recommendation I could give is that if your motherboard supports and is running Intel Turbo boost, turn that off and see if the problem is fixed. I had a similar issue but I did not lose power, it just caused the pc to freeze without any BSOD.
 
Scooby_Doo said:
My concern with it being a temp problem is that it resets so quickly that I doubt it has time to even warm up, never mind over heat.
Click to expand...
Scooby_Doo said:
I must mention that the CPU test in OCCT works just fine, its only the PSU and GPU tests that cause the system to die on me.
Click to expand...
Just thinking your CPU/GPU needs thermal paste refresh. Yes, I read that CPU temperature is correct. It can be fake indicator, or temperature changes are so quick that monitoring do not pick it up.
 
sajunky said:
Just thinking your CPU/GPU needs thermal paste refresh. Yes, I read that CPU temperature is correct. It can be fake indicator, or temperature changes are so quick that monitoring do not pick it up.
Click to expand...

Would a CPU warm up in literally under 1 second, I know it goes up to 50 degrees but but...? I do not think it can be the GPU as its doing it on 2 different cards, coupled with that, stripping the heat sink off a GPU is bound to kill the warranty so will avoid that at all costs..

Also the CPU stress tests are working fine over many hours, surely it would crap out then if it was going to?
 
CataclysmZA said:
Prime also doesn't stress things in the same way as other load inducers, so it should only be one of the programs you use for testing. Try out Intel's Xtreme Tuning Utility, Linpack or Sissoft Sanda Lite. If the overclock is somehow not stable, XTU and Linpack will show it up almost immediately, just like OCCT does in your tests.
Click to expand...

Tried the intel's xtreme tuning utility. Used the memory and cpu stress test everything was fine. CPU never went over 55 degrees.
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X