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

[blaunarwal 433]

3 hours ago, Ray Proudfoot said:

Thoughts on this texture problem? Related to the freezes?

They are just missing or a problem in the config terrain.cfg but because of this I don't expect a freeze. It just remains black, as you wrote.

 

3 hours ago, Ray Proudfoot said:

That’s crazy with a 4090 limited to 30

The gpu is way to good for the cpu i think. The sim still needs a lot of cpu power. My RTX 3090 is almost idling when i fly.

 

3 hours ago, Ray Proudfoot said:

I noticed HWInfo recording 100°C for the CPU on occasions

You're close to limits and this could make your system freeze. I remember when I used wrong AM settings, my system got hot too. Since I went for Steve's recommended AM settings and leave the cpu on its default state, I have no problem with heat. Ok, it's winter now and not so hot in my office.

[Ray Proudfoot 9,002]

8 minutes ago, blaunarwal said:

The gpu is way to good for the cpu i think. The sim still needs a lot of cpu power. My RTX 3090 is almost idling when i fly.

It’s the fastest Intel chip. How can it not be a good match? You don’t mention what resolution you use. I’m running 4K. All cloud sliders in ASCA are set to max. The 4090 certainly isn’t idling. Do you have EA enabled?

10 minutes ago, blaunarwal said:

They are just missing or a problem in the config terrain.cfg but because of this I don't expect a freeze. It just remains black, as you wrote.

But why would the fps drop to single figures? The sim was really struggling to draw the visible scenery.

11 minutes ago, blaunarwal said:

You're close to limits and this could make your system freeze.

Well yes, that’s the whole point of this topic. But what is causing the high temps? No one has explained that one yet.

12 minutes ago, blaunarwal said:

Since I went for Steve's recommended AM settings and leave the cpu on its default state, I have no problem with heat.

You don’t say what Steve’s recommendations are. I’m using the default one that was generated by P3D and only changed the primary thread from 0 to 6 today which is also recommended by LM SDK.

I had no problems with heat on my old Chillblast. And I notice you’re running v5.3HF2. Why not 5.4?

[Luke 782]

7 hours ago, Ray Proudfoot said:

Impossible to say as task manager was unresponsive. I looked in the hardware log and nothing there.

Sorry, where specifically did you look? I am thinking the System log in Event Viewer.

Cheers

[dave2013 1,138]

P3D is still a CPU intensive application.  Having a super powerful video card will help with some aspects, but the CPU is still the bottleneck IMO.

I use a 4070 and even with complex scenery and heavy clouds it isn't nearly fully utilized by P3D.

I suspect that ORBX Australia V2 has something to do with this.  ORBX's scenery can be very taxing on the system, plus they do all kinds of hacks with P3D, some of which it may not like and which can cause problems.

Dave

[blaunarwal 433]

8 hours ago, Ray Proudfoot said:

It’s the fastest Intel chip. How can it not be a good match? You don’t mention what resolution you use. I’m running 4K. All cloud sliders in ASCA are set to max. The 4090 certainly isn’t idling. Do you have EA enabled?

That's something Lockheed Martin has to answer. They moved some workload to the gpu but obviously not enough. I run on 4K too in the sim. In ASCA i have the two first sliders on 2048 and the 3rd on 1024. Models are balanced. I think it is optically sufficient.

I have EA not enabled, don't like the look. EA was faster on my system.

8 hours ago, Ray Proudfoot said:

But why would the fps drop to single figures? The sim was really struggling to draw the visible scenery.

That I think must be something different. Maybe you should just reset to default values and start again loading the system with more demands.

 

8 hours ago, Ray Proudfoot said:

Well yes, that’s the whole point of this topic. But what is causing the high temps? No one has explained that one yet.

Maybe just one setting in the prepar3d.cfg you did some time ago. Therefore delete it and start from default values. That's at least what I was trying.

8 hours ago, Ray Proudfoot said:

You don’t say what Steve’s recommendations are. I’m using the default one that was generated by P3D and only changed the primary thread from 0 to 6 today which is also recommended by LM SDK.

SteveW's recommendations are a pinned thread in this forum. In this very long thread is a setting for the i13900k as well for many other cpus. 

In my prepar3d.cfg is

[JobScheduler]
AffinityMask=268391765
P3DCoreAffinityMask=268391765
MainThreadScheduler=0
RenderThreadScheduler=1
FrameWorkerThreadScheduler=2

The cpu has all cores enabled, HT on.

 

8 hours ago, Ray Proudfoot said:

And I notice you’re running v5.3HF2. Why not 5.4?

I run on 5.4. Just forgot to change my signature.

[vjeanmarc 23]

2 hours ago, blaunarwal said:

SteveW's recommendations are a pinned thread in this forum. In this very long thread is a setting for the i13900k as well for many other cpus. 

In my prepar3d.cfg is

[JobScheduler]
AffinityMask=268391765
P3DCoreAffinityMask=268391765
MainThreadScheduler=0
RenderThreadScheduler=1
FrameWorkerThreadScheduler=2

The cpu has all cores enabled, HT on.

I also use these settings, I don't encounter any temperature problems either.

[Ray Proudfoot 9,002]

7 hours ago, dave2013 said:

P3D is still a CPU intensive application.  Having a super powerful video card will help with some aspects, but the CPU is still the bottleneck IMO.

Agreed which is why my sliders are not fully right. Plus with a 30Hz monitor and VSync the system is not being hammered. I still get 30 At Simwings Heathrow with 160 Ai.

7 hours ago, dave2013 said:

I suspect that ORBX Australia V2 has something to do with this.  ORBX's scenery can be very taxing on the system, plus they do all kinds of hacks with P3D, some of which it may not like and which can cause problems.

Totally agree. They definitely do things differently to other developers and don’t always follow the SDK. P3D never seems to close cleanly when at an ORBX airport.

[Ray Proudfoot 9,002]

3 hours ago, blaunarwal said:

I have EA not enabled, don't like the look. EA was faster on my system.

That’s the difference. I do and it definitely places more workload on the GPU but the 4090 can cope easily.

3 hours ago, blaunarwal said:

Maybe just one setting in the prepar3d.cfg you did some time ago. Therefore delete it and start from default values. That's at least what I was trying.

I did that a couple of weeks ago. Still the problem persisted. But updating graphics drivers appeared to fix things in Europe so maybe this problem is related to ORBX AUS 2 scenery. I’ll do the fix they recommend and then test both in Oz and Europe.

3 hours ago, blaunarwal said:

SteveW's recommendations are a pinned thread in this forum. In this very long thread is a setting for the i13900k as well for many other cpus

That’s with HT enabled. I turned it off as 24 LPs is plenty and it helps with heat.

[Ray Proudfoot 9,002]

8 hours ago, Luke said:

Sorry, where specifically did you look? I am thinking the System log in Event Viewer.

I looked in several places probably including that one. Couldn’t see anything.

[Ray Proudfoot 9,002]

11 hours ago, Luke said:

Sorry, where specifically did you look? I am thinking the System log in Event Viewer.

Cheers

I had two lockups this morning the first 3 minutes out of YSCB, the earliest so far. It recovered then a few minutes later happened again and this time I terminated P3D after several minutes.

But look at this.. The 3 cores assigned to P3D are still running and there's sound from the sim. All the rest are locked. HWInfo logging shows CPU Temp locked at 100C. It was only because I had assigned P3D cores away from 0 I could use Paint etc.

Here is my JobScheduler entry...

[JobScheduler]
AffinityMask=16777215
P3DCoreAffinityMask=16777215
MainThreadScheduler=6
RenderThreadScheduler=2
FrameWorkerThreadScheduler=4

In System logs I have several warnings to ACPI. The text reads ": The embedded controller (EC) returned data when none was requested. The BIOS might be trying to access the EC without synchronizing with the operating system. This data will be ignored. No further action is necessary; however, you should check with your computer manufacturer for an upgraded BIOS."

The texture issues around Sydney were resolved by disabling FlyTampa YSSY Landclass.

I'm wondering if the CPU cooling is working properly.

[blaunarwal 433]

What are the (6) sub processes within Prepar3D.exe. Could you open please. I have at least two less.

The cpu load is definitely overexcessive. I have only core0 at full, the rest is like 2 here. Also noteable the Anti Malware Executable at 1.9%. Maybe something still not excluded. Not a big problem but missing performance anyway. 

P.S. I have to correct. New flight, different values. Flying Carenado Do228 over Uluru in Australia. Some room left for the OS etc. but also heavy load. No active weather at the moment (server down).

P.S.2 Did you see on the ASUS page is a new BIOS 1801 for your board, increasing performance. 

Edited December 24, 2023 by blaunarwal

[Ray Proudfoot 9,002]

1 hour ago, blaunarwal said:

What are the (6) sub processes within Prepar3D.exe. Could you open please. I have at least two less.

I'm running Aivlasoft EFB, LittleNavConnect, ChasePlane, PollyPot GIT, RAAS.

Here are some interesting stats. I monitored AIO Pump as well as CPU Temp. Here's the results. Note how the pump is working close to max all the time and is limited to 2700rpm.

So I set AIO Pump to Full Speed in the BIOS and tried the same flight again. It locked again a little later than before. But this proves to me that cooling is insufficient. By leaving the AIO Pump on the setting it was delivered with I wanted to see if that cured the problem. It didn't.

Here's two shots of performance a couple of minutes before the  lockup and soon after it. That looks okay to me around 40nm NE of Canberra over the bush.

I can't do much more now until new year when I'll talk to Chillblast and provide this evidence.

I'm wary of updating BIOS before talking to Chillblast. What are your thoughts on the problem? I'm now connvinced it's either insufficient or inadequate cooling, not P3D which I've had for years and also this particular scenery. Plus the lockups are at random places.

Thanks for your help, much appreciated.

[blaunarwal 433]

11 minutes ago, Ray Proudfoot said:

Here are some interesting stats. I monitored AIO Pump as well as CPU Temp. Here's the results. Note how the pump is working close to max all the time and is limited to 2700rpm.

You remember I started with the cooling control software... My pump is at 5315 rpm. The pump runs always at this rat. I have the radiator steered over temperature. I think your pump should be on the same value as mine. Did you plug the pump to the right connector on the board?

I tried a takeoff from Canberra, ORBX airport and city, and Australia v2. You remember my last screenshot I had much more load suddenly. The system is changing a lot. Now only 13% load but no weather at the moment. 

Now we soon will celebrate Christmas, friends are coming. Merry Christmas Ray

Edited December 24, 2023 by blaunarwal

[Ray Proudfoot 9,002]

5 minutes ago, blaunarwal said:

Did you plug the pump to the right connector on the board?

The computer was built professionally. Maybe they've messed up. The cooler was changed as the one I wanted was unavailable and holding up the build. I have the Fractal Design Celsius+ S24 Prisma 240mm All-in-One Liquid CPU Cooler.

What setting do you have in the BIOS for the AIO Pump. It was Full Speed on delivery so I changed it to Turbo thinking they made an error. My older Chillblast (i7-8086K @ 4.9GHz) never had this problem even though it was overclocked.

Case is Fractal Design Meshify 2 Compact Gaming Case and just long enough to hold the 4090. No problems at all with the 4090 heat.

[Ray Proudfoot 9,002]

Just found this on the Fractal site for my AIO Pump...

"Rotational speed.

800-2800 RPM in normal use, 3500 RPM during thermal protection mode that activates if liquid temperature is over 60°C"

It's never getting close to 3500rpm. Maybe if it did the thermal protection would kick in and prevent the locking.

Is this thermal protection mode something in the BIOS?