i really hope you get this sorted Ray, it must be driving you potty !

Hello to all, 

I expose you what happened to me after about two months ago I finished implementing the hardware of my Pc with a CPU I9-9900k 5 GHz - GPU RTX 3080 Ti 12 Gb - Asus Prime 390 - RAM 32 Gb 4000 Mhz - 2 x Samsung 980 PRO 1T - Power 850 W gold - liquid CPU cooler.

After a fortnight where everything had gone O.K., while I was flying I had to pause and when after about ten minutes I returned to the PC there was the ... infamous DXGI_DEVICE_HUNG.

After about a week, where I had made a dozen flights without problems but I never put paused flights while I was landing after a flight of about three hours, another ... infamous manifested itself and that is... blue screen and loud buzz.

This blue screen and buzz then manifested itself at other times but not in the flights taken.

Leaving aside the various unsuccessful attempts made to find the culprit, I remembered that after replacing the old GPU with the RTX 3080 I had made a variation in the BIOS of the Asus Prime Z390 where on AI Tweaker - Power-Saving & Performance Mode I had put Performance Mode.

After putting AUTO, more than a month has passed by now where I have made about fifty flights and paused some flights for about ten minutes but no more DXGI_DEVICE_HUNG nor blue screen and buzz.

Obviously this, for at least a month, seems only the solution to my described problems but the doubt it leaves me is that for the blue screen and buzz I think the BIOS configuration also comes into play. 

I've had many of those errors with earlier V5 version but, since re-installing 5.2hf1 about a month ago, I haven't experienced the problem. Been flying mostly the PMDG 777/747 in mid to long-haul routes.

9 hours ago, RobPol471 said:

Hello to all,  I remembered that after replacing the old GPU with the RTX 3080 I had made a variation in the BIOS of the Asus Prime Z390 where on AI Tweaker - Power-Saving & Performance Mode I had put Performance Mode.

After putting AUTO, more than a month has passed by now where I have made about fifty flights and paused some flights for about ten minutes but no more DXGI_DEVICE_HUNG nor blue screen and buzz.

Obviously this, for at least a month, seems only the solution to my described problems but the doubt it leaves me is that for the blue screen and buzz I think the BIOS configuration also comes into play. 

Thank you. I’ll check my BIOS and see what setting I have.

1 hour ago, GCBraun said:

I've had many of those errors with earlier V5 version but, since re-installing 5.2hf1 about a month ago, I haven't experienced the problem. Been flying mostly the PMDG 777/747 in mid to long-haul routes.

Putting my logical head on if you were getting the DXGI problem and a reinstall of v5.2HF1 solved it that would suggest a bad file or files was replaced. But why would only the DXGI files become corrupted? If you were getting other problems I could understand a reinstall would fix those too.

It just seems like a coincidence. How did you reinstall? Just uninstalling and reinstalling the client or a complete uninstall and reinstall of P3Dv5? What about add-on.cfg, scenery.cfg and other important files?

1 minute ago, Ray Proudfoot said:

Putting my logical head on if you were getting the DXGI problem and a reinstall of v5.2HF1 solved it that would suggest a bad file or files was replaced. But why would only the DXGI files become corrupted? If you were getting other problems I could understand a reinstall would fix those too.

It just seems like a coincidence. How did you reinstall? Just uninstalling and reinstalling the client or a complete uninstall and reinstall of P3Dv5? What about add-on.cfg, scenery.cfg and other important files?

It was a fresh installation. Haven´t made any reg or cfg tweaks since installing it but I must say that I am not using P3D quite extensively these days, so it may be just luck that I have not bumped into this issue yet.

22 minutes ago, GCBraun said:

It was a fresh installation. Haven´t made any reg or cfg tweaks since installing it but I must say that I am not using P3D quite extensively these days, so it may be just luck that I have not bumped into this issue yet.

I went from 25 September until yesterday without a DXGI message appearing. But then I had inhibited that appearing with the Registry fix. It should not have appeared but did. I don’t understand that at all.

The thing about doing a full install is that you don’t know what the problem was. It’s the nuclear option. I really wish LM would say if they are still able to reproduce it and are working on a fix but their silence is deafening.

We still have no idea what triggers it because it cannot possibly be scenery related as it would be reproducible. It seems to happen when the system is far from being stressed by too high settings.

I could go another two weeks with things being fine or it could happen again today. It is the most infuriating bug I have come across in 30 years of simming.

Not sure where you got the TdrDelay=0 from Ray, that seems to be the reverse of what Adobe suggest.

Setting zero would mean not waiting at all and terminate the process immediately.

I have a TdrDelay of 60 set and have not seen the error occur in probably two years now since setting the entry. Graphics drivers and an GPU appear to be key, especially on any OC GPU whether by user or manufacturer.

 

From Adobe:
 

Quote

 

In order to prevent any rendering or GPU computation from locking up the system, the Windows operating system kills the GPU driver whenever a rendering takes more than a few seconds. When the driver is killed, the application using it crashes automatically. It is not possible to know how long a rendering task or a computation may take (it depends on the GPU, the drivers, the OS, the mesh size, the texture size, etc.), therefore it is not possible to put a limit on how much the computer should process and avoid the crash from the application level.

On Windows there is a registry key specifying how long the OS should wait before killing the GPU driver. Application are not authorized to modify this setting directly, this procedure has to be done manually (see below).

 

 

8 minutes ago, srcooke said:

Not sure where you got the TdrDelay=0 from Ray, that seems to be the reverse of what Adobe suggest.

Setting zero would mean not waiting at all and terminate the process immediately.

I have a TdrDelay of 60 set and have not seen the error occur in probably two years now since setting the entry. Graphics drivers and an GPU appear to be key, especially on any OC GPU whether by user or manufacturer.

I got TdrLevel=0 from the FS Labs forum where it appeared to solve the problem with their Airbus fleet. A value of zero should turn the warning off but clearly not under all circumstances.

I’ll try 60 in TdrDelay and remove TdrLevel. I believe you have a 1080Ti too Stephen. What drivers are you using? AI SUITE does not touch the GPU so it will be the same as it was when I bought the system from Chillblast 3 years ago.

The card is a GTX1080 with 471.11 drivers.

Some GPU's are factory overclocked, I have seen some users having the dxgi error getting a good result my backing the OC off a few Mghz.

 

1 hour ago, srcooke said:

The card is a GTX1080 with 471.11 drivers.

Some GPU's are factory overclocked, I have seen some users having the dxgi error getting a good result my backing the OC off a few Mghz.

 

Thanks. Different card and drivers. I'm pretty sure Chillblast won't have o/c the graphics card. It never happened with P3Dv4 - just v5 and purely at random. No pattern whatsoever. It has never happened at heavy airports like EGLL and KLAX even with 100+ AIG Ai aircraft.

2 hours ago, srcooke said:

I have a TdrDelay of 60 set

Stephen, is that 60 decimal or hexadecimal? I've just set 60 decimal.

It is decimal Ray.

In case you missed the link: https://substance3d.adobe.com/documentation/spdoc/gpu-drivers-crash-with-long-computations-128745489.html

@srcooke, thanks Stephen. Did you also set TdrDdiDelay as advised in that link?

Currently enroute Palma to Manchester. Fingers crossed.

Yes I have both entries set Ray.