Computer reboots under load

[David Barishev]

Hello all !

Around a year and a half ago, i upgraded my system from gtx 550 ti , to 970 g1 , and added Samsung 850 evo.

Everything was fine for a year, except couple of crashes hear and there, and sometimes didn't boot and made 3 short beeps, i mostly solved this by removing one ram stick and reinserting.

 

One day when i played my computer crashed, no blue screen, I tried to boot him up, and got short quick beeps when trying to boot. 
I was like , OK this has happened before so I got one ram stick out, and now the computer booted, was stuck on bios , and beeping every 30 seconds. 
I tried to unplug Hard drive , optical drive reader , every external phariperal and still same error- stuck on bios with long beeps every 30 seconds. 
I tried to boot without the GPU , which the computer did , and I was presented with this screen( I was using the onboard graphics),

.
I waited till it finished, and the computer booted up normally.
When I tried to insert the graphics card the computer doesn't want to bot and I'm getting 1 long beep , and 3 short ones. 
 

I sent the card to RMA (stupidly i know), and they returned it to me as the gpu was fine.My friend then tested the card on his system and it worked fine.

 

When i got the card, i inserted it into the pc, booted fine , had a very long bios screen 1-2 minutes, then windows loaded fine.I Fixed this problem by updating my bios version (F5 -> F12),

but now when i try to play a game, my computer just reboot after 20 min or so (depends on the game).

When im not doing anything that requires high gpu usage the computer is stable.

 

I wanted to test my machine to eliminate other problems, so i used a stress test that consisted of Furmark, and prime95, while recording sensors with hwinfo.

My computer rebooted after around 9 minutes.

 

I also did only prime95 for 2 hours, and my computer was fine, cpu peaked at 75c, but average was 65 c, so temperatures are not a problem.

 

Even in my games, and in the benchmark my gpu would work fine, and i got good frame rate.

I attached the stress test log below, and here are my full computer specs : http://speccy.piriform.com/results/d027LREWndsbtpYgHyEv7iT

I'm using Antec 500w 80+ bronze, dont know the exacts model, but here is an image with some data (This might be very useful - it states all the power info):

 

I speculate the cause of the problem im my psu - its not giving enough power, but i want to be sure, and if so, what psu should i buy for my system ?

 

Thanks in advanced.

FullStressTest.CSV

[Light-Yagami]

I don't think PSU is the culprit.. Maybe one component in your system runs too hot and PC shuts off as a result.

Do the usuall stuff. Update your drivers to the latest possible, run only CPU and then only GPU benchmarks, keep an eye on temps and see if anything strange pops up.. 

 

 

[David Barishev]

Hey, thanks for the replay !

When i run cpu only(Without furmark) i can run the test for 15+ minutes and it all works fine, also if you look at the attached file, cpu temp when under full load after around 5 min stays at 70c, and gpu at 80c so i think temperatures are fine.

[Light-Yagami]

3 minutes ago, David Barishev said:

Hey, thanks for the replay !

When i run cpu only(Without furmark) i can run the test for 15+ minutes and it all works fine, also if you look at the attached file, cpu temp when under full load after around 5 min stays at 70c, and gpu at 80c so i think temperatures are fine.

Yeah, 5 minutes of testing really isn't enough. If you plan to accuse the PSU for your rebooting, then run only CPU stress test like Prime95 for 2 hours.. So everything gets nice and toasty. If the PSU is bad, it will fail. You need to eliminate possible culprits. Test multiple conponents one by one and see what happens. It's a trial and error method. That is if you really don't know where to start.. 

 

I had a RAM failure and I refused to believe that it would just die on me (it was making writing errors) so I tested it at a lower speed and higher voltage and looser timings and got it working again (turned out it couldn't quite achieve the 3ghz advertised mark). I could've thrown it away and lost a perfectly fine 16gb kit of RAM. Do you feel me? Don't assume hardware failure till you're absolutely sure. Test your system in multiple ways.. It could be an overclock, could be a voltage too low, could be many things.. You just have to run through all the possibilites

[David Barishev]

Hey, quick update

I just ran prime95 for 2 hours, no crash. Max tmp was around 77c, but average was around 65.

I tried to upgrade all my drivers using the device manager, but windows didnt find any new driver, and my gpu driver is up to date.

 

What should i test next ? Should i try to reinstall windows ?