[Resolved] P3D v5.4 random lockups driving me mad

[Ray Proudfoot 9,025]

19 hours ago, Bob Scott said:

I recommend watching the CPU package power consumption when the lockups happen, as it's a better indication of workload.

I tried a YSCB-YSSY flight this morning but it froze half-way there. I loaded an autosave and flew the remainder, It was fine and I was able to land too.

I'm attaching two  screenshots showing PCB Temp and CPU package Power for this successful flight. It should serve as a useful comparison against a locked one.

Do you see anything unusual in either of these?

[Bob Scott 7,740]

Ray--the EC is the motherboard's embedded controller.

You said PCB temp--I'm assuming you meant PCH temp, and the temps and CPU power demand look OK.  I'd be interested to see what they look like during a lockup.

How are your DIMM (memory) temps?

[Ray Proudfoot 9,025]

5 hours ago, Bob Scott said:

Ray--the EC is the motherboard's embedded controller.

You said PCB temp--I'm assuming you meant PCH temp, and the temps and CPU power demand look OK.  I'd be interested to see what they look like during a lockup.

How are your DIMM (memory) temps?

Yes, PCH I'm sure. Thanks for confirmation of temp and power demand. I'll try the flight tomorrow morning so expect more analysis then after the probable lockup. I'll include DIMM temps too.

I ran MemTest86 and after 2hrs it completed without error. I also tried 3DMark but you only get one test with the free version and that seemed okay.

Do today's tests rule anything out? RAM?

[anzac1977 140]

Do you have your frames locked in it?

I had random crashes with unlimited frames and not using Nivida to lock the frames.

Maybe try locking your frames to 30 using the sims one.

Edited December 25, 2023 by anzac1977

[Ray Proudfoot 9,025]

6 minutes ago, anzac1977 said:

Do you have your frames locked in it?

I had random crashes with unlimited frames and not using Nivida to lock the frames.

Maybe try locking your frames to 30 using the sims one.

I have a  30Hz monitor so with Vsync on and Unlimited set it produces 30fps naturally.

It's always been advised never to set 30 in P3D. It costs CPU cycles to maintain it and is very inefficient.

[anzac1977 140]

I did that as well and had CTD. The FPS kept spiking all of the time.

Try using Nivida to lock it at 30fps.

In the end it is just a test.

Edited December 25, 2023 by anzac1977

[Ray Proudfoot 9,025]

@anzac1977, I appreciate you're offering suggestions but this problem is not related to video settings. Why force something to 30 when it delivers it naturally?

[anzac1977 140]

It is just a box to tick.

Why my sim with a 12900k and 3090 would freeze every flight at random times at unlimited I do not know. But when I change to limited FPS in sim it runs perfectly.

Not a software engineer, just a user offering a simple test to try out. I really do not notice any issues with a 12900k at locked frames and I doubt a 13900k would as well.

Good luck and I hope you get it sorted.

[Bob Scott 7,740]

5.4 is nice and stable for me with a similar config to Ray's...30Hz monitor with unlimited frame rate in sim and hardware VSync limiting frame rate to the hardware refresh rate--that actually relieves the CPU of a lot of work generating unneeded frames, because with a limit set in sim it will still try to generate frames in excess of the vsync rate.

[Afterburner 374]

Ray, just an idea:  Have you tried the default AM setting in prepar3d.cfg (or deleting and generating a new file)?  Just to rule things out.

[Ray Proudfoot 9,025]

2 hours ago, Afterburner said:

Ray, just an idea:  Have you tried the default AM setting in prepar3d.cfg (or deleting and generating a new file)?  Just to rule things out.

Yes. That was the setting until i changed it a few days ago. Freeing up core 0 for the OS makes accessing HWINFO easier.

[Ray Proudfoot 9,025]

Here's today's lockup report. Same route. Lockup occurred later than yesterday and rather than wait for it to recover I left it locked for 5-10 secs to gather info and then terminated P3D.

PCH Temps remained between 50-70. Here's the CPU Package Power chart. No real increase in power on the lockup.

SPD Hub temps for DIMMs varied between 36.3 to 38.8 for DIMM1 and 34.3 to 36 for DIMM2.

I hope that helps but I see nothing untoward.

I also monitored FSUIPC Autosave (every 3 mins) and other than a very small stutter it executed without incident. It was not executing when the freeze occurred.

[Ray Proudfoot 9,025]

Interesting... very interesting. Given I've been able to reliably reproduce this freeze / hang on one particular flight I decided to try a flight of similar distance somewhere else in the world.

Manchester to Heathrow is one I had a problem with before updating my graphics drivers 10 days ago. So what better short route than that to try this test again.

I started the logging soon after takeoff and then waited for the hang. And I waited... and I waited... and I waited. At Westcott NDB I expected a hang but it never came. I completed the flight into a very busy Heathrow with 180 Ai in the 80nm bubble. Far more demanding than the Australian bush.

One flight is no guarantee but since I updated the drivers 10 days ago the only place I've had these hangs is on the YSCB-YSSY route at various places along that route.

In what must be 10 attempts on that route I've not had one successful flight. But here I am flying in a far more testing area and the computer is fine.

I shall refly YSCB-YSSY tomorrow but with AUS v2 disabled. If that flight is successful then everything points to that scenery being the culprit.

@Luke, your theory might be right.

[blaunarwal 433]

1 hour ago, Ray Proudfoot said:

In what must be 10 attempts on that route I've not had one successful flight. But here I am flying in a far more testing area and the computer is fine.

This sounds like a corrupt file somewhere and then Windows error reporting uses a lot of cpu power. On my system, sometimes this let the radiators spool up, sometimes i only get the hourglass and then CTD.

About the heat development i don't understand why you don't try what you pinned and discussed with us a few months ago.

So SteveW gave us some samples of settings for the i13900k cpu, first for full load and then others for adjusted performance and heat development. That's the one i use and showed you some posts before. It is easy, just put HT on in the BIOS, use these settings in Prepar3D.cfg and look if it keeps temperatures lower.

As i understood SteveW, that you may pull all out of the i13900k, if you can dissipate the heat it generates. Since my aio cooler does not work this good, i had to step down. I hope the industry will bring out better cooling systems one day or at least shops that list comparable cooling rates.

Also limiting fps to 30 is not the worst nightmare of a simmer and if you don't like it, you can always unlock it again. I run the sim at this fps. because without, my textures become blurry during flight, because of too little resources.

Dan

[Ray Proudfoot 9,025]

Dan,

the possibility of a corrupt file may become clearer once i fly the route with Aus 2 deactivated or uninstalled. I have previously verified the files and it made no difference. Disabling FT YSSY Landclass solved the textures around Sydney.

Regarding the pinned topic I don't see the need to run with HT enabled with 24 VPs available and of course lower temps.

With a monitor capable of 30Hz why play around with settings in P3D? Bob Scott explained very clearly why the method he and I use is the best possible solution.

Lockheed-Martin also recommended in the SDK that utilising a core away from 0 if the user prefers to leave the OS to use it. Even with HT enabled Steve always suggests only using 1 of the first few pairs shared with the OS.

With your kit you shouldn't be having blurries. Is that in P3Dv5?