Good Samaritan with beast PSU/GPU needed

Arroyo1010

Edit:
Sorted, one of my colleagues has a PSU, and is willing to borrow. No need to read further
Original message below.

Hi all,
First post here

I have bought a gaming rig ~6 months ago @dabs :

1KW PSU - Fractal Design Newton R3
CPU -Core i7 4770 (never clocked)
RAM - 32 GB ram (4x8GB) Kingston HyperX
Mobo - Gigabyte Q87M-D2H
Graphics: Gigabyte's GTX 770 (slightly factory OCd)
2 SSDs
Integrated sound
DVD rom
Windows 7 64bit Professional

The problem is that the computer would shut down after 5-20 minutes of gameplay of GPU-intensive games, and restart.
A good way to reproduce the problem now is to run the Unigine Valley demo, in which the shutdown occurs after 2-5 minutes. Playing a low-end game (Binding of Isaac: Rebirth) wold never cause these events, even after a few hours. Also, I can't reproduce the issue with the Furmark burn-in test, which is why I turned to Unigine. Games in which the shutdown happens: Crysis 3, Titanfall, Battlefield 3 (and the Unigine Valley demo).

Observations:
I've been tracking the PC stats via the "Remote System Monitor" app on my phone. At the time of the crash, GPU is about 63 c, CPU at below 50 c ; voltages don't have any oscilations (straight lines)

Things I've done:
I have made sure that I have the crash dump enabled, and prevented automatic reboots. Regardless, there are no crash dumps recorded on these events. No BSOD. Just a hard shutdown.
Event Viewer in Windows just reports the unclean boot.
Tried utilizing other 12v PCIe rails, as well as different combinations. I have 4 cables for the GPU available, and my GPU needs 2.
Tried downgrading the Nvidia drivers, and then upgrading to beta. With old drivers I get the shutdown immediately in the Unigine demo
Checked for the latest mobo BIOS - Already using latest
Bios defaults - no change
Cleaned up the case (there was barely any dust)
Unplugged/re-plugged connectors.
Taken out the GPU, blew some (barely any) dust out and re-seated it.
Replicated the issue in the other part of the house (I also suspected the building's power.)
Furmark does not cause the event (GPU at 69% usage, though)
CPU stress-test does not cause the event
CPU stress-test + Furmark does not cause the event

My conclusions:
I believe it's either GPU or the PSU. Maybe the mobo?

In order to verify, I would kindly ask someone in Dublin 2 (and therabout) to volunteer their HW for testing: GPU, PSU or both. Note that I need at least a GTX 770, and/or a PSU which is 650W+

I know that this is a long shot, seeing that I'm asking people to contribute serious HW, but I guess I kind of have to do this. A guy from a local PC shop told me that they don't have the components with which they would troubleshoot.

It goes without saying, a pint is on me.

P.S.
Please feel free to comment on what else I could try

Alanstrainor

Good luck mate. No one in their right mind would volunteer such hardware. You should stilll be able to verify/root cause this issue more with what you have.

Does the Mobo have an onboard VGA/HDMI? If so, remove the graphics card, and you may need to tell the bios to output to the onboard boards, and boot. Since you wont be able to run any games as a means of testing, you can download some CPU benchmark/stress test applications and run through tests to see if is the CPU.

What wattage is the PSU, and what is the total wattage of the setup?

I'm not familiar with you temp monitor app for your phone. Get some temp logging software for your machine that will just create logs/graphs and save them as they occur. And review after a system crash. It sounds to me like something is overheating, and that it is most likely the GPU.

Have you tried reseating the RAM? Have you googled around for troubles with that card?

PS: I would get onto Dabs, and start sending things back with that kind of track record....when in warranty I wouldn't waste too much time troubleshooting.

Arroyo1010

Alanstrainor wrote: »

Good luck mate. No one in their right mind would volunteer such hardware.

I know... no harm in trying, though. My colleague turns out to have a 1250w PSU - AND is willing to come over, so I'm set!

Does the Mobo have an onboard VGA/HDMI? If so, remove the graphics card, and you may need to tell the bios to output to the onboard boards, and boot.

Yes, but that would not test the PSU very much. The graphics card draws more than 230W: (can't post URLs. Nvidia ref. spec for GTX 770. Factory OC'd slightly)

Since you wont be able to run any games as a means of testing, you can download some CPU benchmark/stress test applications and run through tests to see if is the CPU.

Did that, prime95 + some other. I leave the computer for two hours, and the temps don't go over 65c on the CPU.

What wattage is the PSU, and what is the total wattage of the setup?

465 minimum recommended vs 1000 w PSU

I'm not familiar with you temp monitor app for your phone. Get some temp logging software for your machine that will just create logs/graphs and save them as they occur. And review after a system crash. It sounds to me like something is overheating, and that it is most likely the GPU.

My first guess was overheating, too. I tried using speedfan with logging first, but the log file gets corrupted after boot. This app is pretty good, seeing that it shows me all the values that speedfan does + some extra GPU values which speed fan does not. It's not the temps: GPU is at 55-65c when the computer shuts down.

Have you tried reseating the RAM? Have you googled around for troubles with that card?

Re-seated the RAM (forgot to mention earlier). I have also re-seated the GPU, and re-plugged every power connector.

PS: I would get onto Dabs, and start sending things back with that kind of track record....when in warranty I wouldn't waste too much time troubleshooting.

Seeing that my colleague will be coming over on the 24th, I'll be able to determine for sure which component is faulty, and then I'll send the faulty component over.

Thank you for taking the time to respond!

Alanstrainor

I have had faulty GPU issues where it was not overheating, but rather memory issues. However the fact that you do not get any bluescreens is a bit odd to me.

The PSU is also likely, that would explain why you have no logs to show for it. Best of luck!

yoyo

Faulty GPUs would tend to lock up/cause a BSOD rather than cause a system shutdown in my experience anyways. Likewise faulty memory. If the system is overheating (particularly the CPU) the system may shutdown. You could try using Furmark to stress the GPU, but don't use it for too long as it will stress it to the max!
Testing another PSU is a good idea, I'd check the CPU cooler is also firmly on as well. Sometimes temp sensors in motherboards aren't always that accurate, not sure that's a big issue these days though.

Nick