PC shuts down during climbout (P3Dv4 thunderstorms)

[Wanthuyr Filho 83]

Hello ALL,

It's been like 3 times already over the last months, temperature in the room is good (cold enough) since we're in the winter down here, thus I suspect it's a PSU issue. What happens is that the PC simply turns itself OFF abruptly during climb when the weather is stormy (in the simulator session of course) with CBs and lightning all around. It's not every stormy situation that it will happen, since I've had successful flights in these conditions, but each time that the issue occurred it was during this kind of scenario. Then I have to hold pressed the ON/OFF button to "reset" and turn on the PC again. It starts up normally and when I check Event Viewer it has no clues other than the "critical error/this PC shut down unexpectedly" thing. I have this PC hardware/configuration running stable for many years and only in the last few months this thing started to happen, and I can't even blame temperature. The other day I made a stress test and temps all around seemed to stay within throttling limits. Apparently it's only occurring during some stressful simulation sessions, such as climbouts in stormy weather with lots of clouds (and textures) being rendered. The airplane I've been flying is the FSL A320, which as we know is one of the most hardware demanding out there nowadays. My P3D versions is 4.5 HF2.

Any comments on this?

Thanks.

Edited August 8, 2021 by Wanthuyr Filho

[stans 703]

Could certainly be that during graphically intensive scenes, the CPU and GPU are pulling more power than the PSU can provide and you computer shuts off.  I've had computers that would freeze or turn off when stressed, then the PSU would fail, sometimes in a spectacular fashion... POP!, then smoke.

[Bob Scott 7,720]

It does sound like your PSU is being overloaded when the GPU load spikes up.  What kind of GPU and PSU are you using?

 

[Wanthuyr Filho 83]

2 hours ago, w6kd said:

It does sound like your PSU is being overloaded when the GPU load spikes up.  What kind of GPU and PSU are you using?

 

My GPU is an EVGA FTW 1080 Ti and PSU is a Corsair 750w (circa 2013).

[Bob Scott 7,720]

52 minutes ago, Wanthuyr Filho said:

My GPU is an EVGA FTW 1080 Ti and PSU is a Corsair 750w (circa 2013).

I think that more than likely you're seeing degradation in your PSU.  There are a couple of one-off possibilities worth mentioning--first, make sure your GPU aux power connectors are not daisy-chained on the same cable--they should be on two separate cables.  Also, if that PSU is not a single-rail supply, it might be worth trying to connect one of the aux connectors to a cable coming off a different DC rail.

I'd also sniff around and see if there is any hint of a burned odor coming off the PSU.

Personally, given that it's 8-years old, I think I'd just go ugly early and replace the PSU with a new 850W-1000W unit.  Some PSU failure modes can be catastrophic to connected components.  It might just be that the ghost of Georg Ohm may be trying to send you a message...

[turbomax 3,285]

there are certain components that dry out over time and can lead to malfunction of power supplies, the so called electrolyte capacitors, especially the big ones in power supplies (smaller ones are also used on motherboards). they are used to reduce ripple and "noise" on heavy duty current lines. the better power supplies use japanese capacitors (nippon chemicals), supposedly these last longer and are slightly more expensive than cheap chinese capacitors. 

also often overlooked issue in so-called multi-rail power supplies: refer to your power supply's manual to make sure you use both pci-e 12 volt rails that go into your GPU, instead of a single y-split cable. an average 12 pci-e rail may supply typically (based on wattage of power supply) some 20-25 ampere, multiplied by 12 volts = 240 - 300 watts, which might be insufficient because a power hungry and overclocked RTX 3090 GPU draws up to 350-400 watts. in other words: even if a power supply is rated for lets say 650 watts, it might still not be able to actually deliver those 650 unless all 12 Volt power rails are used up properly. GPU power, temperature, volts etc. can easily be controlled while your game is running using GPU-Z:

https://www.techpowerup.com/gpuz/

https://thinkcomputers.org/settling-the-debate-multi-rail-vs-single-rail-power-supplies/

https://www.pcpartguide.com/single-vs-multi-rail-power-supplies/

https://askingthelot.com/what-is-a-multi-rail-power-supply/

Edited August 9, 2021 by turbomax

[stans 703]

12 hours ago, Wanthuyr Filho said:

My GPU is an EVGA FTW 1080 Ti and PSU is a Corsair 750w (circa 2013).

Old PSU, it's likely breaking down. Don't buy just any PSU, try to get a good one as some makers/resellers are better than others.  My current computer's EVGA power supply died after 16 days of use.  Evidently, Gigabyte is now a PSU reseller and they may be selling some really poor units.  See this thread.

 

 

Edited August 9, 2021 by stans

[martin-w 3,235]

1 hour ago, stans said:

My current computer's EVGA power supply died after 16 days of use.

 

I have to say, EVGA power supplies are excellent. Great quality and never had one fail. I always buy EVGA power supplies. Great warranty too, 10 years on many models, and great support.

Even the best fail sometimes though.

[turbomax 3,285]

another premium brand is bequiet, warranty varies between 5 - 10 years.

https://www.bequiet.com/en/powersupply

[SteveW 1,792]

Before doing anything else, I would blow out all the dust to be clean before starting and then reseat all the power cable plugs (unplug and press back into place), reseat the GPU in the slot and maybe reseat other cables and cards. Over time connections can become resistive and reseating them will clean them up and reduce the contact resistance.

[martin-w 3,235]

On 8/8/2021 at 10:10 AM, Wanthuyr Filho said:

Hello ALL,

It's been like 3 times already over the last months, temperature in the room is good (cold enough) since we're in the winter down here, thus I suspect it's a PSU issue. What happens is that the PC simply turns itself OFF abruptly during climb when the weather is stormy (in the simulator session of course) with CBs and lightning all around. It's not every stormy situation that it will happen, since I've had successful flights in these conditions, but each time that the issue occurred it was during this kind of scenario. Then I have to hold pressed the ON/OFF button to "reset" and turn on the PC again. It starts up normally and when I check Event Viewer it has no clues other than the "critical error/this PC shut down unexpectedly" thing. I have this PC hardware/configuration running stable for many years and only in the last few months this thing started to happen, and I can't even blame temperature. The other day I made a stress test and temps all around seemed to stay within throttling limits. Apparently it's only occurring during some stressful simulation sessions, such as climbouts in stormy weather with lots of clouds (and textures) being rendered. The airplane I've been flying is the FSL A320, which as we know is one of the most hardware demanding out there nowadays. My P3D versions is 4.5 HF2.

Any comments on this?

Thanks.

 

 

Does any other software other than the sim cause this issue? If not, and stress tests don't either, i'd suggest a sim issue. Software issue.

Reinstalling is a pain, I know, but may help. Updating all drivers of course. And if a mechanical HD, then defragging. Updating all add-ons of course. Running a virus scan, making sure Windows is updated. General PC maintenance. You say no temp issues, but have you checked the GPU as well as CPU? 

If you suspect a PSU issue, then I would expect it to rear its head during tough stress tests and you say it doesn't. Try quite a few different stress tests, emphasising the CPU and then the GPU and then both. You can buy a cheap PSU tester on Amazon, it will tell you what the PSU is doing when not under load. 

Checking for memory issues is another good idea.

 

I'm sure you know the drill.  

Edited August 9, 2021 by martin-w

[turbomax 3,285]

stress test benchmark programs don't necessarily recreate the same stress conditions that flight simulators do. I just ran furmark GPU benchmark at highest seetings without any problems:

https://www.geeks3d.com/furmark/

but MSFS 2020 after a few minutes shuts pc completely off (RTX 3090, i9-9900K overclocked, using obviously underpowered 650 power supply). after replacing with a 1000 watt power supply I don't have these problems anymore. MSFS (and I assume P3D as well) can generate milliseconds of increased combined loads, spikes that last only miliseconds, but long enough for the protection electronics within the power supply to shut off immediately, even though the total amount of all pc components does not exceed the nominal wattage of the PSU. it is those short increased spikes in ampere load, when CPU and GPU together draw too much power. using stress tests like furmark I have not observed such spikes, therefore they don't reveal these problems.

Edited August 9, 2021 by turbomax

[OlliePen 88]

Look for a 10-year warranty on a new PSU. Think Gold 80+ ratings from manufacturers such as Sea Sonic, Corsair and EVGA.

[martin-w 3,235]

12 hours ago, turbomax said:

stress test benchmark programs don't necessarily recreate the same stress conditions that flight simulators do. I just ran furmark GPU benchmark at highest seetings without any problems:

 

True. This is why I suggested a range of stress tests. Not just one or two. If power draw spikes are believed to be an issue then the temp testing done while running the sim should reveal temp spikes in synchrony with the power draw spikes.

OPP (Over Power Protection) should kick in if there are excessive spikes. 

[turbomax 3,285]

17 hours ago, martin-w said:

 

the temp testing done while running the sim should reveal temp spikes in synchrony with the power draw spikes.

temps take too long to rise to a critical point and if spikes last only milliseconds you won't see temps rise at all, psu will long have shut down before. GPU temp around 80, vram memory around 88 celsius with no problems, but sudden load spikes and boom - psu black out. I just replaced my power supply, went from 650 watts to 1000 watts, and had not blackouts since, using RTX 3090 and i9-900k, both overclocked.