Randomly shutting down

[art20]

Hey all, I've been getting the strangest kind of crashes I've seen on a pc so far. They can happen pretty much at any time, from the BIOS screen to playing a game. What happens is, the computer will completely freeze including sound, pointer, keyboard... everything and the only thing to do is restart. Or the computer will just simply shut down, not as in a full shut down cycle but it will just power down out of nowhere.

I've had this setup (the one in my profile)for about 9 months now, started using Win 7 about 4 months ago and these problems just started happening about a week or 2 ago. There is no recently installed software. I just did a fresh install of Win 7 32bit and the problem is still there.

My first idea was that this was a hardware issue but there really arent that many components that can literally just shutdown the computer like that other than the PSU and that is brand new.... because of this problem my old psu unit was no longer to power the system during boot because of so many restarts.

This happens with both Win 7 32bit and 64bit so that rules out the OS... kind of... because the system runs perfectly on Ubuntu

I really dont know what the problem is, at one point I thought it might be the video cards but I've tested them both separately and they run just fine up until I start loading stuff up like a game or even something like youtube is enough to make it crash.

But the strange thing there is that it will also crash while idle after a while and even while at the BIOS screen it will power down.

Ive ran memory tests and disk checks and it all seems to be fine. The PSU is new and more powerful than the old one and the video cards appear to be in order. Please help.

Thanks

 

[usasma]

Try these free stress tests:

Try this free video stress test: FurMark: Graphics Card Stability and Stress Test, OpenGL Benchmark and GPU Temperature | oZone3D.Net

FurMark Setup:
- If you have more than one GPU, select Multi-GPU during setup
- In the Run mode box, select "Stability Test" and "Log GPU Temperature"
Click "Go" to start the test
- Run the test until the GPU temperature maxes out - or until you start having problems (whichever comes first).
- Click "Quit" to exit

Try this free stress test: Free Software - GIMPS

Prime95 Setup:
- extract the contents of the zip file to a location of your choice
- double click on the executable file
- select "Just stress testing"
- select the "Blend" test. If you've already run MemTest overnight you may want to run the "Small FFTs" test instead.
- "Number of torture test threads to run" should equal the number of CPU's times 2 (if you're using hyperthreading).
The easiest way to figure this out is to go to Task Manager...Performance tab - and see the number of boxes under CPU Usage History
Then run the test for 6 to 24 hours - or until you get errors (whichever comes first).
The Test selection box and the stress.txt file describes what components that the program stresses.

 

[art20]

Try these free stress tests:

Try this free video stress test: FurMark: Graphics Card Stability and Stress Test, OpenGL Benchmark and GPU Temperature | oZone3D.Net

FurMark Setup:
- If you have more than one GPU, select Multi-GPU during setup
- In the Run mode box, select "Stability Test" and "Log GPU Temperature"
Click "Go" to start the test
- Run the test until the GPU temperature maxes out - or until you start having problems (whichever comes first).
- Click "Quit" to exit

Try this free stress test: Free Software - GIMPS

Prime95 Setup:
- extract the contents of the zip file to a location of your choice
- double click on the executable file
- select "Just stress testing"
- select the "Blend" test. If you've already run MemTest overnight you may want to run the "Small FFTs" test instead.
- "Number of torture test threads to run" should equal the number of CPU's times 2 (if you're using hyperthreading).
The easiest way to figure this out is to go to Task Manager...Performance tab - and see the number of boxes under CPU Usage History
Then run the test for 6 to 24 hours - or until you get errors (whichever comes first).
The Test selection box and the stress.txt file describes what components that the program stresses.

Hey, thank you for your reply, I downloaded the FurMark test and ran it a couple of times, the first time 1 GPU was stable at 52C while the other went up all the way to 88C.

The second time I ran it, I activated SLI and one GPU was still at 52C and while the other one was at 75C (ish) the computer shut down.

Im going to try the other test you recommended now.

Thanks again

 

[art20]

Ok.... Quick update.

after running the Prime95 test twice, the computer will totally freeze within 15 seconds of starting the test. Any idea what might be happening here?

 

[art20]

Hey, sorry to keep bumping my own thread but this is driving me nuts -.-

I just discovered that one of the GPU's seems to be overclocked. Its running at 700MHz when it should be at 650... all the other features in the card are also above what they should be like Shader domain, memory clock and even core temperature.

Could this be causing all my problems? Any ideas how to underclock it?

 

[art20]

So I removed both Gpu's one at a time, removed all but a single 2 GB Ram stick (swapped it for an identical one after 1st try) also removed all but the system hard drive and last the sound card. So all I was left with was the CPU a single GPU and 1 2GB stick of ram

The problem was still there, havent been able to do any major stress tests other than FuMark and the memtest that comes with the Ubuntu live Cd because the computer willl either freeze or shut down.

So the only thing I can think of is the CPU but I'm surprised a faulty cpu can perform some tasks but not others as the system only crashes at really random times sometimes when the processor is barely being used.

Any ideas? im desperate, the week is starting and im a freelance designer and this is my work box

 

[win7clutz]

If prime failed check CPU Temps, test Ram with Memtest, if both are fine I'd be changing Power Supply... Luck

 

[art20]

Hey I think the problem is solved now, thank you guys for your advice

What happened was: the heat synk wasn't keeping the processor at a stable temp, therefore causing it to shutdown like that. Thanks a lot to my mistaken instincts and the many restarts I did, the power supply wasnt able to keep the system running anymore, so I got a new one. Then I read on a post here someone had a similar issue and someone reccomended he installed CPUID to see voltage issues so I thought it'd be worth a shot to see if I could get a reading of stuff at crashing time and thats when I noticed the processor temp going WAY off when it was stressed. So today I got a new heat synk and the problem seems to have stopped