@ark, @capceo and @Flysimware:

I apologize if this should possibly be in a PM, but...thought I'd see if the FSW team might be able to assist in tracing a P3Dv5 bug.

Specifically, there are more than a few of us with a v5 bug where with certain aircraft, including default aircraft, we can't change view TO the cockpit.  We can change views from the cockpit, but upon attempting to reenter the cockpit in the same view window, we get a DXGI hung error and CTD.  Faulting module vcruntime140.dll and/or g2d.dll.  I want to be clear that this is a P3D/LM problem, not a FSW problem.

I am asking the FSW team because you happen to have products that both do and don't suffer from these errors.  Specifically, the MU-2 installed with the native v5 installer DOES NOT cause this problem - we are free to change views with no problems.  The Lear 35A installed with the v5 installer DOES cause the error, as does the default F-35, and a bunch of the default aircraft.

So, in the interests of collaborating to solve this v5 bug, which so far from LM's comments in the prepar3d.com forum thread suggests they're stumped, I thought I might ask a couple things:

1) Can you think of what's different in how the MU-2 VC, panel or model differ technologically/programatically from the Lear in such a way that it might be calling a different .NET/runtime module, windows component or P3D component?  Or something in that vein that might help explain/narrow down where the problem might exists?

2) Would you be willing to share those thoughts/findings with LM via your developer/beta/non-public channel to assist them in tracking this down?

One thing I was thinking is that of the aircraft that work - the default Maule, the Majestic Q400, the FSW MU-2 - I don't think any of them incorporate PBR textures, whereas I'm thinking the others (the F-35, the Lear, the Falcon, the AS A330) do use PBR.

But I'm sure there are a myriad of other differences.

Anyhow, again, this isn't your problem, but given that FSW is in the unique position of having actually developed aircraft that both do and don't suffer from this, you may be in a unique position to help LM get pointed in the right direction on a solution.

This is the relevant thread on the prepar3d.com forum:

https://www.prepar3d.com/forum/viewtopic.php?f=6312&t=136854

Thanks, as always, for everything you do and all the wonderful aircraft you've brought into our lives!

👍

p.s. @ark I know you're a volunteer helping with the forum, but wanted to include you anyhow since you may have some ideas for us as well!

Edited April 23, 20206 yr by cwburnett

5 hours ago, cwburnett said:

@ark, @capceo and @Flysimware:

I apologize if this should possibly be in a PM, but...thought I'd see if the FSW team might be able to assist in tracing a P3Dv5 bug.

 

Mark and I got together and tried some things. I don't have P3Dv5 yet but Mark does. Mark did get the failure as you described with the Lear35 when cycling through the views.  One difference between the MU2 and the Lear is the Lear model was exported for 64 bits using the LM SDK, and the MU2 was exported for 32 bits using Microsoft's SDK. We also noticed that the failure occurs when you try to cycle back into the cockpit using the typical 'S' key, but the failure doesn't happen if you use the P3D Views drop down menu.

And strangely, after Mark played with the View drop down menu for a while he restarted the sim and then could not replicate the failure with the Lear. 

Regarding PBR, with FSW only the exterior models have PBR, and in addition the FSW Falcon50 worked OK and has PBR. So this seems to suggest PBR may not directly be the issue.  However, it was after the Lear started to 'pass the test' that we tried the Falcon so it is not clear if the Falcon would have failed the TO cockpit test initially.  I will ask Mark to try the Falcon 'first' tomorrow.

Thanks for including me in this. And BTW, just FYI capceo is also a volunteer. FSW is a one man operation -- Mark.

Al

 

 

Edited April 24, 20206 yr by ark

It seems that this crash primarily affects users with AMD Radeon 56xx or 57xx GPUs? Curious what GPU Mark uses, since he was able to replicate the crash.

FWIW, I have not had any issues with getting a device hung error even when repeatedly going to external view and back into the VC of either the Lear or Falcon in P3DV5. I have an Nvidia GTX1080Ti GPU.

Agreed, but there is at least one GTX1080 user on the LM forums having the same problem.  https://www.prepar3d.com/forum/viewtopic.php?t=137279

1 hour ago, ark said:

Mark and I got together and tried some things. I don't have P3Dv5 yet but Mark does. Mark did get the failure as you described with the Lear35 when cycling through the views.  One difference between the MU2 and the Lear is the Lear model was exported for 64 bits using the LM SDK, and the MU2 was exported for 32 bits using Microsoft's SDK. We also noticed that the failure occurs when you try to cycle back into the cockpit using the typical 'S' key, but the failure doesn't happen if you use the P3D Views drop down menu.

And strangely, after Mark played with the View drop down menu for a while he restarted the sim and then could not replicate the failure with the Lear. 

Regarding PBR, with FSW only the exterior models have PBR, and in addition the FSW Falcon50 worked OK and has PBR. So this seems to suggest PBR may not directly be the issue.  However, it was after the Lear started to 'pass the test' that we tried the Falcon so it is not clear if the Falcon would have failed the TO cockpit test initially.  I will ask Mark to try the Falcon 'first' tomorrow.

Thanks for including me in this. And BTW, just FYI capceo is also a volunteer. FSW is a one man operation -- Mark.

Al

Thanks, I am so appreciative that you guys jumped on this. I never cease to be amazed with FSW, Mark and the dedicated volunteers. Count me in if I can ever be of service.  Hopefully we can help LM track this thing down.

8 hours ago, cwburnett said:

Thanks, I am so appreciative that you guys jumped on this. I never cease to be amazed with FSW, Mark and the dedicated volunteers. Count me in if I can ever be of service.  Hopefully we can help LM track this thing down.

Mark built a new computer recently and as I recall it has an Intel processor and Nvidia GPU.

I wonder if 3rd party add-ons that deal with 'views', like ChasePlane , TrackIR, etc., might be playing a role in this strange problem with moving from an outside view to the cockpit view.

9 hours ago, JRBarrett said:

FWIW, I have not had any issues with getting a device hung error even when repeatedly going to external view and back into the VC of either the Lear or Falcon in P3DV5. I have an Nvidia GTX1080Ti GPU.

Jim, thanks very much for your input on this. I think your experience has likely ruled out PBR as the culprit.

Al

59 minutes ago, ark said:

I wonder if 3rd party add-ons that deal with 'views', like ChasePlane , TrackIR, etc., might be playing a role in this strange problem with moving from an outside view to the cockpit view.

So after we started troubleshooting over on the LM forum, I built an entirely fresh OS with just native P3Dv5 on it on a second hard drive just to test this one issue and the problem occurs repeatably every time. Yesterday, after receiving both the MU-2 and Lear updates, I installed just those two aircraft in combination with native v5 and it has never happened on the MU-2 and happens every time with the Lear, as well as the F35 and some others.  I have tested almost every permutation of settings; I attempted turning "transition=on" for the cockpit cameras, to see if I could 'slide' from the left to right seat, and still the same problem. I've tried each version of the AMD drivers since 19.12.1 using DDU between attempts, I've tried with Windows 10 1906 and 2004 review build, I've tried with various resolutions, HDR on/off, and a bunch of other tests to try and nail this down.  I've even tried underclocking and undervolting my GPU.  AMD pushed a new driver yesterday - 20.4.2 - and I had hope that would solve it, but that also didn't work.  I'm completely and totally stumped.  I have actually never used any of those tools - I may seem like a neophyte, but I don't use EZDOK, ChasePlane or anything like that.

Edited April 24, 20206 yr by cwburnett

Just tested this sitting on the tarmac using amd 580 graphics card...I could not duplicate. repeatedly cycled through the " S "  key for 2 minutes 

which was about 30 cycles..  p3dv5, learjet

2 hours ago, rickgo said:

Just tested this sitting on the tarmac using amd 580 graphics card...I could not duplicate. repeatedly cycled through the " S "  key for 2 minutes 

which was about 30 cycles..  p3dv5, learjet

The problem seems to be limited to RX 5xxx (5700/5600) cards and the Radeon VII (though some with the VII have said it works). I bought and put a VII in last weekend to test and I still had the CTD on a fresh install of the OS, driver and P3D - so I returned the VII and went back to my 5700XT.  A number of people have said the 5xx series work fine, and only one or two Nvidia users have reported a similar problem.

I guess your in the same boat as far as no Nvidia waterworks. My card is no powerhouse, but the lack of good water is ugh. 

I will be replacing it later on this year. Waiting for next Nvidia Gen.

Just wanted to report, I'm getting the g2d.dll CTDs on view change to internal after v5-HF2 as described above with my RTX2070.