My PC has been intermittently hard rebooting for a few weeks now. The physical appearance of this reboot would look as though the power has gone out. All the PC fans/lights lose power and the system is "dark" for maybe 2 seconds. Then the lights and fans come back on, and it boots normally.

At first, this was a rare occurrence, happening maybe once per weekend, maybe not at all for a week of gaming. But as the weeks have passed, it's seeming more common. Today it happened twice, within about 10 minutes of starting Elite Dangerous.

Specs:
i7 920
6GB ram
GTX 950 2GB
1 SSD, 1HDD, 4 case fans.
Corsair HX750W PSU

All stock, no overclocks.

My troubleshooting began with monitoring the fans and temperatures. All logged data seems to indicate that the CPU is rarely crossing 60C, even under load. The GPU does run hotter, up to 75C at peak, but this appears within spec. CPU fan speed appears constant, Automatic GPU fan speed appears functional.

Next I checked the memory with Memtest86+, and completed 1 wall in 40 minutes with no errors.

The case/fan lights going dark tends to make me think the power supply could be an issue. Data logging on the voltages (captured before and during a reboot instance) show the following:

CPU VCore: min 1.20v, max 1.21v (1.20v last logged prior to reboot)
+12V: min 11.79v, max 14.2v (13.3v last logged)
+3.3V: min 2.73, max 3.05v (2.83v last logged)
+5V: min 4.92v, max 5.37v (5.37v last logged)

I'm not at all an expert on this, but those voltage variations seem a bit wide. However, there don't appear to be any signs of major voltage changes logged on the last data entry before the reboot.

And that's where I'm at, currently. I wonder if any of the expert eyes here might spot some clues before I go off spending money.

About the only thing I have not done is try running the PC plugged into a different surge protector/power strip. My current one is pretty old, but I've no idea if that could be an issue. At this point, due to concerns about the PSU perhaps faulting, I'm hesitant to keep testing with games and forcing the problem to occur.

Oh, and I did disable the auto-restart function in Windows 10, but didn't have another reboot since doing that. So no certainty on whether there might be a suppressed Blue Screen, or whether it's worth trying to test that.

Thanks for any advice.

Sounds like PSU issue to me. Do you have another computer in the house that you could swap with to test? Try again without any non-essentials such as lights and extra fans?

If it just reboots by itself with no warning, then I think it is PSU. Obviously if it was a software issue, you probably would've noticed and the PC would've probably thrown out some warnings before rebooting.

So this only happens when 3d gaming?

Does your motherboard have onboard video?

I remember a reboot once (maybe twice) while just at the desktop, no 3D engaged (that I was aware of, anyway).

No onboard video, motherboard is an ASUS P6T.

Best guess from me in my limited experience is your vid card is dying or it's the PSU.

Well, if it reboots while idle on the desktop, sounds more and more like a power issue.

+1 for failing (or possibly overheating) PSU, I had a similar issue a few years ago. Over maybe 2-3 days, my pc would randomly shut down w/o any reason, then one day it wouldn't power on at all.

It was even a SilverStone 80+ Bronze PSU and lasted for ~5 years w/o any issues, then *poof*.

Agreed on psu. Still, I wouldn't trust temp readings too much. Go in there and check for dust. In the wrong place it can cause odd issues, particularly with power. Blow out the case and the psu. Check your heat sinks.

Thanks for the responses, everyone.

I'm moving ahead with making a final attempt to isolate any useful detail, and let it happen again.

As a final checklist:

Visual inspection of all fans, heat sinks, and PSU.... All clean.
Updated nVidia graphics drivers (my prior drivers were dated from like November of 2016).
Plugged a lamp into the PC's power strip (so I know if the power strip could be the problem, blinking out the lamp to make it obvious).

Just completed a 3 hour session in Elite Dangerous, no problems. All temp and fan readings are good. Inspecting them in person during and after the session shows them all running, and everything cool. The PSU is cool to the touch during play, the fan slowly spinning, and a cool breeze coming out the rear vent.

I'll report back next time it happens.

If it's just the PSU that went *poof!* and none of your internal components were affected, I'd count that as a sign of a good PSU. Time to retire it, but at least it went without taking others with it.

Oh I forgot to post about the unit I'm considering to replace my PSU with. It's a lower spec Corsair, but I think will do the job fine. The prime interest for me is that this model is detailed by Corsair to include the various system protections in case of it's having trouble down the road.

https://www.amazon.com/CORSAIR-CX550-Bronze-Power-Supply/dp/B01MRW2K79/ref=sr_1_1

Downgrading from a 750W PSU to a 550W will not likely be a big deal. This computer originally had the i7 (138 TDP) paired with an AMD HD 5780 (188 TDP), with headroom for overclocking. The currently installed EVGA GTX 950 only has a TDP of 75W. The total power demand of the system looks to be around 300 watts or so, the 550W unit will seem to do the job with reserves to spare.

I'd heartily recommend EVGA SuperNOVA G2 PSUs. I would also recommend going with a 600-650W unit as well. I agree that a 550W may have room to spare, but most PSU's efficiency is around the 50% mark so the closer you are to the limit, the less efficient it actually becomes. Also might want to consider a modular or semi-modular PSU as well... less cable to clutter your case.

Thanks Ice, I'll look those over.

BTW, I've logged another 2 hour play in Elite Dangerous, then a very taxing session in Arma3 (with 40GB of addons loaded and doing infantry/vehicle combat on XCam Taunus). Zero crashes, reboots, or performance issues.

One weird thing, since updating those old nVidia drivers I'm seeing some different behavior. Before, there used to be a feature early on in the Windows 10 release called "XBox App" or something, which offered a few features like overlays showing FPS, and video recording. Well back in those early days, I had enabled a few of those features and set some hotkeys.

Then somewhere along the way, a Windows 10 update removed access to that XBox App, without first signing in. I never did this originally, and since there was no way to try to adjust any settings otherwise, I did create a sign-in hoping access those old settings. To my annoyance, even with a sign-in all those settings and controls had been removed from the XBox App settings menus. So there was never any way to disable all those FPS overlays.

Well, after updating the nvIdia drivers while troubleshooting this issue, the FPS overlay stuff is all gone now. Makes me wonder if a Windows Update could have been causing trouble with my old nVidia drivers.

That was happening to me all last week. This morning, my son was on the computer playing a game and there was a loud BANG and the thing shut down for good. PSU exploded, luckily nothing else damaged. new PSU and back up and running/

apart from the ones that are certified 'bronze/silver/80% Plus'' etc in which the certification is guaranteed at 100% load? In my opinion the PSU is one of the most important parts of the system and it's well worth going above and beyond what you think you need purely because it's a component that will last you several upgrades. It's also worth paying out for a brand name with the 80%+ certification as it's definitely a component that is 'buy cheap/buy twice'. I learned that the hard way myself with an el-cheapo 800w unit that fried itself and unfortunately took my brand new cpu and motherboard with it many years ago.

FWIW I have running this on my ASUS Z170 Machine with an I5 and a NV 950 and 16GB of RAM for almost a year now.

Another Corsair 550 watt

Not even hint of a hiccup.

That's not what I meant.... I was talking about efficiency...


link

Thanks, Nixer. That model was on my short list.

Comparing that RM-X unit, and Ice's recommendation, I split the difference and went with a Corsair CX550M. It offers a few better specs and the 5 year warranty for a few bucks more than my first choice unit, but still saves a bit over a more enthusiast oriented unit. My brother has this same PSU in his build, and it's been solid for him.

I should have it by Monday.

Just got the new PSU installed, and tested under gaming load.

The motherboard voltage readings are already looking vastly better. Those swings on the 12V line from 11.7V to 14.2V are now a rock steady 12.1, from idle to full gaming load. Same with the 3.3V and 5V lines.

Thanks for the final verdict. It's useful to know what a good and bad PSU voltage reading look like, I wouldn't have had an idea myself.