MSFS Causes Computer to Suddenly Restart

[ESzczesniak 128]

1 hour ago, flyinion said:

You need to use 2 separate cables.  Especially on a 3090 you are exceeding the power limit for a single cable and probably overloading that single output on the PSU.  Those dual pigtail cables are intended for much lower power cards.

Maybe if I'm lucky, that will fix my problem.

[peloto 213]

Curious, the best day after maybe around two months that I came back to flight FS20, no one error or issue, a VR flight with some disconnectings from VR, flight of three hours with the headwind 330-900 and then a normal PC restart for load another new flight and a bit more than a hour with the helicopter Airbus H125.

I have to test more, but the two different things today was the W11 update and the raise of the Rams voltage to 1.29. I did a test some weeks ago with 1.26 without no success, and decided this two days ago when I had a "memory could not be read” as last resource before a new basic install of W11 only with FS2020 in a new partition to test primally.

The name of both setting are VDD and VDDQ.

It has sense once one stick is 1.25v, and when you add more maybe it is normal to have to raise the voltage. I bought the exact same RAM memory but two sticks separately.

 

I did too a pair of tests of RAM during hours of Memtest with 0 errors some month ago, and a lot PRIME95 really intensive with Ram with 0 errors OCing, curious, hope find and discover the solution already...

[Bob Scott 7,722]

11 hours ago, ESzczesniak said:

Maybe if I'm lucky, that will fix my problem.

It very well might.  Running a 3090 off a single power cable (with a splitter at the end) causes a pretty significant voltage drop at the card and might be causing the overcurrent protection to trip. 

The 3090 is a pretty power-hungry GPU--in fact some of the 3090s actually use *three* power cables instead of two.

Seems to me a better design standard for high-power PSUs (1000+ watts) would be to add a few more modular connector sockets on the supply and only provide cables with a single connector on the device end.

Since this is really a hardware topic, I'm going to move it to the hardware forum now.

[martin-w 3,242]

On 5/10/2022 at 5:27 PM, ESzczesniak said:

My system: i9-12900k, Asus Z690-E motherboard, G.skill 32 gb DDR5 RAM, PNY RTX 3090, eVGA 1000W PS, NVME drives x 3, 1 2.5" SSD, Corsair H150i liquid cooler, Windows 10 Professional

 

If you check out my comments in the thread below you'll see I had the same issue. 12900k RTX 3090 Trident Z DDR5.

Quite a few with 12th gen are seeing this.

Change XMP to auto so it runs at the JEDEC frequency. All issues for me have now gone.

12th gen and DDR5 are new, quirks like this are to be expected.

 

I havent read the entire thread yet so if you've determined it's something else, ignore me.

Edit: I see you do have XMP off. You mentioned using your card on a riser cable. Riser cables are notoriously troublesome. And yes, you should use a separate PCIe cable for each connector on a 3090, for my EVGA I use three separate.

 

 

Edited May 11, 2022 by martin-w

[ESzczesniak 128]

14 minutes ago, martin-w said:

 

If you check out my comments in the thread below you'll see I had the same issue. 12900k RTX 3090 Trident Z DDR5.

Quite a few with 12th gen are seeing this.

Change XMP to auto so it runs at the JADEC frequency. All issues for me have now gone.

12th gen and DDR5 are new, quirks like this are to be expected.

 

I havent read the entire thread yet so if you've determined it's something else, ignore me.

 

 

 

I have XMP disabled.  I haven't been through the whole thread, but what I can see this wouldn't apply if I have XMP disabled.

[peloto 213]

Just started a flight very happy, and... an excellent freeze screen very decorative....

[ESzczesniak 128]

Unfortunately adding the second power cable to the graphics card did not resolve the issue.  I had more or less the same experience, with the computer rebooting right at Vr/liftoff.  This time the computer actually had to reboot a few times.  It would get to the Windows login screen, but reboot a couple times before I could get in to Windows.  During one of these, the computer seemed to POST, but the screen remained black for a while.

My PSU was supposed to be here from Amazon today, but is delayed.  I'm about 20 minutes in to an OCCT stress test for the CPU.  Temps seemed to have plateaued at 70-75C and the coolant temp just under 40C. 

I still have this lingering concern over the video card.  Nothing hard, but lingering.  It definitely runs hot.  And if I'd leave my system on for a couple days, would start to get some graphic artifacts. 

[martin-w 3,242]

11 hours ago, ESzczesniak said:

still have this lingering concern over the video card.  Nothing hard, but lingering.  It definitely runs hot.  And if I'd leave my system on for a couple days, would start to get some graphic artifacts. 

 

What's the DDR6X Memory Junction Temp? Junction temp for DDR6X runs VERY hot. 90 degrees plus is common, mine hits 95 degrees in the sim, 90 now I've replaced some of the thermal pads.

Check the Junction temp with GPUZ latest version. Older version didn't read it. Or Hardware Monitor latest version. When the new versions came out the temp shocked the pants of people. Some users are seeing 110 degrees, throttling and shutdowns. 

In GPUZ tick the box to read to file, it will then save the temps for you. 

I have a thread about this too.

When you say hot, how hot, what's the GPU temp under load. Check that too in GPUZ. 

Check Junction Temp and GPU Temp. 

 

 

Edited May 12, 2022 by martin-w

[ESzczesniak 128]

2 hours ago, martin-w said:

 

What's the DDR6X Memory Junction Temp? Junction temp for DDR6X runs VERY hot. 90 degrees plus is common, mine hits 95 degrees in the sim, 90 now I've replaced some of the thermal pads.

Check the Junction temp with GPUZ latest version. Older version didn't read it. Or Hardware Monitor latest version. When the new versions came out the temp shocked the pants of people. Some users are seeing 110 degrees, throttling and shutdowns. 

In GPUZ tick the box to read to file, it will then save the temps for you. 

I have a thread about this too.

When you say hot, how hot, what's the GPU temp under load. Check that too in GPUZ. 

Check Junction Temp and GPU Temp. 

 

 

Under stress testing through OCCT or MSI Kombuster, the chip tops out about 78C and the memory junction 86-88C. In games, I’ve seen 6-8C higher for both, but not for very long. 
 

I know these temps shouldn’t be causing a reset. The card itself is older than system. I got it fairly early after they came out before the scalpers shot the prices sky high, so maybe around 2 years old. I don’t recall it running at those temps early, but once a system is stable, I don’t watch temps a lot. 

Edited May 12, 2022 by ESzczesniak

[martin-w 3,242]

4 hours ago, ESzczesniak said:

he chip tops out about 78C and the memory junction 86-88C. In games, I’ve seen 6-8C higher for both, but not for very long. 

 

110 is the temp where DDR6X throttles and then shuts down. If you aren't exceeding that it shouldn't be an issue. 

EVGA told me that they aren't concerned unless I'm hitting 110. DDR6X runs hot, its as simple as that, and with the 3090 having chips on the back, just cooled by a thin aluminium plate, its not exactly top-notch cooling.

So difficult to diagnose with the symptoms you have because they can be caused by a number of issues. Luckily for me, I have spare components to swap out. 

Have you updated your BIOS by the way?

See how you get on with your new PSU.

 

[tfm 169]

I see you have no overclock.

 

But I see you are using an Asus motherboard. So I would be interested to know how you managed to avoid an overclock? If you think that “optimised default” is the same as “no overclock” then (in my experience) that’s sadly not the case.
 

I have always used Asus motherboards for my home builds. I assumed that the “optimised default” setting, or whatever they call it, would use Intel’s specs. But in fact (on all my home builds, anyway) even this setting attempts an overclock. I have never found out how to force them to use Intel defaults.

 

 

[ESzczesniak 128]

Something seems to have worked!...although with these bugs, they can be elusive and pop back up unexpectedly. 

Unfortunately, I have to say I was a bad boy and did not change one thing at a time to know for sure what was the cause.  However, I was able to complete a full flight in the Maddog X without any problems.  Some combination of the following:

1. Changed the PSU.  On an interesting note, these EVGA PSU's have an "eco mode" where they will not run the fan if it's cool enough.  I wasn't really aware of this until the new PSU was in place.  On paper it shouldn't cause an issue, but there are reports of people solving random restarts by turning the switch off.  The old PSU was "on", leaving the new one off.

2. New RAM.  I had some new low profile RAM available, but not yet installed.  I've been wanting to fit a 280 mm radiator in my case instead of 240 mm, but this needs low profile RAM sticks.  I also put them in new slots (A1/A2 vs B1/B2).

3. Removed some LED light strips from my case.  It seemed fun at the time, but I've been less than impressed and decided to get rid of them.

4. Unplugged my Reverb G2 head set USB.  There were a few reports on the MSFS forums of people seemingly having restart issues when using MSFS on a screen, but still having the head set plugged in.

5. New flight/scenery.  This really shouldn't cause anything more than a lockup or CTD, but moved from KSAN-KLAS to KSTL-KORD.

Ideally it would have been best to do one at a time, but at 20-30 minutes to load a flight, setup at the gate, and take to "line up and wait", it would have taken me 2 weeks to test all these variables. 

  

10 hours ago, tfm said:

I see you have no overclock.

 

But I see you are using an Asus motherboard. So I would be interested to know how you managed to avoid an overclock? If you think that “optimised default” is the same as “no overclock” then (in my experience) that’s sadly not the case.
 

I have always used Asus motherboards for my home builds. I assumed that the “optimised default” setting, or whatever they call it, would use Intel’s specs. But in fact (on all my home builds, anyway) even this setting attempts an overclock. I have never found out how to force them to use Intel defaults.

 

 

I will not claim to be an expert here, but I have XMP set to disabled and my CPU reports 3.2 GHz at no/light load, and doesn't top 5.2 GHz on any core.  This matches the advertised specs from Intel, so it seemed fair to think there wasn't an overclock.  And my PNY 3090 is basically an FE edition and has not overclock.  I've been eyeing a Asus ROG 3090 for the OC, but also it probably won't make much difference (I do at least use 4K, which is becomes GPU limited).

Edited May 14, 2022 by ESzczesniak

[martin-w 3,242]

You can turn off Asus performance enhancements and stick to Intel power limits in the BIOS. It also asks you which you prefer when you first fire up the board. I see no auto overclocking at all on my board.

I've just discovered something (hopefully). In addition to crashes and lock up's, I was also noticing my Corsair keyboard assignments going bonkers in game.

My daughter had installed Citrix on my system for working at home. Uninstalled Citrix and now my keyboard behaves itself. According to Google, many have had issues with Citrix causing game crashes. 

I'm wondering if I had an issue with XMP and Citrix or if both issues were down to Citrix. 🤔

 

[peloto 213]

Try to disable HAGS in FS20 in game settings, the graphic controller, Zendesk have been pointing to disable it and the culprit, or maybe deleting MSFS from the list of apps. I am having really good flights without any problem in VR the majority.

[peloto 213]

Only disabling hyperthreading in BIOS is the most success that it is giving to me.