Computer shuts off when playing GPU intensive games

[Oifish]

Recently when I've been playing games my rig will all of the sudden shut off, then when it restarts, it only makes it a couple seconds into the start up process before restarting again. This loop will continue until I manually shut off the PC. If I let the PC rest for a while and turn it back on there are no problems, unless I go play a GPU intense game. I have noticed that this happens more frequently if the temperature in my room is hotter. I never get a BSOD, the rig just restarts. I''m assuming it's either the PSU or GPU. Does anyone know of any software that tracks and logs temps and voltages? I have no way of knowing if this is caused by a temperature spike or voltage drop. Or am I barking up the wrong tree entirely. Thanks!

 

[Murloc]

I suggest EVGA Precision X if you have a nvida video card (it doesn't matter if it's not an EVGA), use it to set a fan curve that goes to 100% at a reasonable temperature (for me it's 90°C, and 80% at 80°C) and see if it keeps happening. It also shows the GPU temp.

EDIT: disregard, I see you have an AMD card. Use an equivalent program and do the same (idk if catalyst does it).

I don't think you can measure the PSU temperature from the computer.

Trying another PSU would be an idea too if you can't solve your problem. Idk how old yours is but 600W should be enough.

 

[redrider4life4]

I would say it is def PSU or CPU, those two would cause your computer to shut off. If your GPU overheats you just get crazy artifcating or it will freeze.

 

[denis280]

The Temp. and also the psu to check.

 

[monkeydelmagico]

download HWinfo64 or hwmonitor
run prime95 or AIDA64
check CPU temps and see if running benchmark for a while shuts compy down
if yes, check CPU cooler for dust and fans working. Maybe take off and re-apply thermal paste.

 

[Oifish]

Alright I have HWinfo64 running and logging its info while I run prime95. I'll get back with the results later. After about 15 min the core temp has reached a max of 42.0* celsius and is staying steady at 41.0*. I know 15 min is not very long, but I figured it wouldn't hurt to let you know.

 

[Ketchup]

Alright I have HWinfo64 running and logging its info while I run prime95. I'll get back with the results later. After about 15 min the core temp has reached a max of 42.0* celsius and is staying steady at 41.0*. I know 15 min is not very long, but I figured it wouldn't hurt to let you know.

Is the 42 at idle? That's not too bad for idle. If it is saying 42 under load, I would question the sensor it is reading off of. Are both cores reporting the same temp?

 

[Oifish]

at idle core 0 reads at 27*, core 1 reads at 35*. At load both read at 42*.

 

[Ketchup]

Are you running aftermarket cooling? Those temps seem quite low to be correct, for a Wolfdale.

Anyway, it's a good bet your problem is the power supply, based on what it takes to shut it down. Seeing that it's OCZ puts another nail in it, for me anyway.

 

[pandemonium]

Aside from the PSU being bad - reading your sig - I'm guessing you've been running this rig for a bit now. The voltage on the GPU may be too low, causing it to crash and Windows not returning an error code and restarting the graphic controller from the drivers. Once you get some software installed to tweak voltages, I'd recommend posting the current voltage it's set at, searching for similar cards on the net, and compare it to their voltages with the same card. It's quite possible the gates on the GPU processor have diminished to the point of requiring more juice to operate.

 

[Oifish]

Are you running aftermarket cooling? Those temps seem quite low to be correct, for a Wolfdale.

Anyway, it's a good bet your problem is the power supply, based on what it takes to shut it down. Seeing that it's OCZ puts another nail in it, for me anyway.

Yes I have aftermarket cooling. After running Prime95 for 17 hours both core's max temp was 43*. And yes, pandemonium I've had this rig for about 5 years now. It honestly works great for everything I need...well except for this issue. HWinfo64 reports the GPU voltage at 1.263 V. I'll search for what the normal voltage readings should be, but if anyone knows where I can find this info, that'd be great.

 

[AnonymouseUser]

I would suspect the PSU is overheating. Turn a tabletop fan towards the (open) side of the case and stress test it. If it doesn't crash then something is overheating.

 

[monkeydelmagico]

Aside from the PSU being bad - reading your sig - I'm guessing you've been running this rig for a bit now. The voltage on the GPU may be too low, causing it to crash and Windows not returning an error code and restarting the graphic controller from the drivers. Once you get some software installed to tweak voltages, I'd recommend posting the current voltage it's set at, searching for similar cards on the net, and compare it to their voltages with the same card. It's quite possible the gates on the GPU processor have diminished to the point of requiring more juice to operate.

Agreed. Next test would be GPU using furmark, heaven, or valley benchmarks. Keep an eye on temps and voltages

 

[taq8ojh]

You didn't say what PSU do you have. What's the watts rating on it?

 

[Oifish]

Agreed. Next test would be GPU using furmark, heaven, or valley benchmarks. Keep an eye on temps and voltages

Ran a furmark 1920x1080 benchmark. Voltage stayed constant at 1.263V and the temperature reached a max of 84*. PC didn't shut off or freeze.

 

[pandemonium]

That voltage appears to be stock for that card, so a voltage increase may be called for here.

Prior to doing so, I'll ask if you've done a fresh driver update recently? >>Reboot to Safe mode, uninstall GPU drivers, run CCleaner and scan registry (clean all), reboot, install latest version of GPU drivers.

 

[Oifish]

That voltage appears to be stock for that card, so a voltage increase may be called for here.

Prior to doing so, I'll ask if you've done a fresh driver update recently? >>Reboot to Safe mode, uninstall GPU drivers, run CCleaner and scan registry (clean all), reboot, install latest version of GPU drivers.

I've updated it to the newest Catalyst version (13.1) but not from scratch. I'll do that. If the problem still persists what is a good program to increase the voltage?

 

[taq8ojh]

Newest Catalyst version is 13.4 or 13.9 (released today). Not sure why are you using 13.1

 

[pandemonium]

I've updated it to the newest Catalyst version (13.1) but not from scratch. I'll do that. If the problem still persists what is a good program to increase the voltage?

Get a clean driver install done and see if it helps. Either way, Sapphire has their own OC tool: https://www.sapphireselectclub.com/ssc/TriXX/

 

[Oifish]

Newest Catalyst version is 13.4 or 13.9 (released today). Not sure why are you using 13.1

From what I've read the 4xxx series is not supported by anything newer than 13.1. Pandemonium, I'll do a fresh install and tweak the voltage if needed. I'll you all posted if there are any changes.

 

[taq8ojh]

Oh. I didn't see that part.
You might want to check Guru3d forums, people sometimes mod drivers for older cards.

 

[frozentundra123456]

I would suspect the PSU is overheating. Turn a tabletop fan towards the (open) side of the case and stress test it. If it doesn't crash then something is overheating.

This sounds like the most reasonable explanation. I had a very similar issue with one of my computers shutting down under load, and I finally found that the PSU fan was sticking and allowing it to overheat.