nVidia 471.96 and DXGI_DEVICE_HUNG resolved

[Ray Proudfoot 8,980]

@MarkW, I suspect that without the registry fix you would have had one by now. If you're feeling brave remove the TdrDelay fix. 😁

@IanHarrison, hard to say without more evidence. No further instances of it in a 2 hour flight up to Poland.

[JesC 88]

2 hours ago, Ray Proudfoot said:

With TdrLevel=0 now set I'm wondering if that prevented a DXGI problem.

Ray,

Perhaps all we can really say is that the TdrLevel=0 is potentially allowing recovery from the DXGI problem and it appears the underlying problem still exists? So, not so much a "fix" as a workaround?

Jesse

Edited September 27, 2021 by JesC

[IanHarrison 522]

13 minutes ago, JesC said:

Ray,

Perhaps all we can really say is that the TdrLevel=0 is potentially allowing recovery from the DXGI problem and it appears the underlying problem still exists? So, not so much a "fix" as a workaround?

Jesse

Does a puncture repair fix a tyre or is it a workaround until you buy a new one?

[Ray Proudfoot 8,980]

29 minutes ago, JesC said:

Ray,

Perhaps all we can really say is that the TdrLevel=0 is potentially allowing recovery from the DXGI problem and it appears the underlying problem still exists? So, not so much a "fix" as a workaround?

Jesse

Jesse, I suspect the problem will remain unresolved and Registry entries will be the solution for many. These Tdr features are there for good reason. It’s been suggested I uninstall and reinstall P3Dv5 again but given I get no other problems what guarantee is there all that work will resolve the problem? None.

it seems to affect users of complex 3rd party aircraft so LM will never be able to fix it with their simplistic aircraft. And between Microsoft, Nvidia and LM if you raise the issue with one they’ll just point the finger at the other two. Except you won’t hear from LM as we’re not entitled to support.

In over 12 flights now I’ve not had the problem in any bar one instance when I paused the sim. So as far as I’m concerned this is a fix and in all probability, a permanent one.

[JesC 88]

2 hours ago, Ray Proudfoot said:

2 hours ago, JesC said:

 

Jesse, I suspect the problem will remain unresolved...

Please don't misunderstand my post. I don't intend to be critical of what you have found through a considerable effort. And, I certainly am appreciative of your willingness to so freely share your findings.

My point is simply that a true fix is going have to come from LM/Microsoft/Nvidia. And, I say this because my understanding is, like yours, that the TDRLevel=0 setting is really only intended for use in testing and troubleshooting. It obviously is a good workaround for now. But that does not mean LM et al don't need to do more work on this.

Sincerely, thanks for your work and the information you share on this and many, many other topics. You have always been very helpful.

Jesse

Edited September 27, 2021 by JesC

[Ray Proudfoot 8,980]

1 hour ago, JesC said:

Please don't misunderstand my post. I don't intend to be critical of what you have found through a considerable effort. And, I certainly am appreciative of your willingness to so freely share your findings.

My point is simply that a true fix is going have to come from LM/Microsoft/Nvidia. And, I say this because my understanding is, like yours, that the TDRLevel=0 setting is really only intended for use in testing and troubleshooting. It obviously is a good workaround for now. But that does not mean LM et al don't need to do more work on this.

Sincerely, thanks for your work and the information you share on this and many, many other topics. You have always been very helpful.

Jesse

No offence taken Jesse. Who knows who will come up with a solution? This problem has been going on for ages and you wonder if things are getting too complex.

The setting was successful on my second 2 hour flight up to Moscow. Filthy weather on approach and fps were in the high 20s with the GPU hitting 95% from FL200 down to the ground. But the GPU coped. 👍

[Cruachan 705]

1 hour ago, JesC said:

My point is simply that a true fix is going have to come from LM/Microsoft/Nvidia. And, I say this because my understanding is, like yours, that the TDRLevel=0 setting is really only intended for use in testing and troubleshooting. It obviously is a good workaround for now. But that does not mean LM et al don't need to do more work on this.

Methinks MSFS going DX12 will be the stimulus that quickly provides us with a solution. My belief is that the likely ‘culprits’ are Nvidia’s drivers and/or Microsoft’s operating system. What’s the betting that a fix will quietly sneak into our experience via a driver/software update with no acknowledgement of ownership.  Shame on them but, it has to be said, far from being untypical these days.

Mike

 

[Cruachan 705]

30 minutes ago, Ray Proudfoot said:

The setting was successful on my second 2 hour flight up to Moscow.

TdrDelay=0 or TdrLevel=0 or both? Is TdrDelay=0 still in place?

Mike

Edited September 27, 2021 by Cruachan

[Ray Proudfoot 8,980]

13 minutes ago, Cruachan said:

TdrDelay=0 or TdrLevel=0 or both? Is TdrDelay=0 still in place?

Mike

Only TdrLevel=0. TdrDelay has been removed.

I’ll test for a paused flight tomorrow. That was the one scenario that TdrDelay had no effect with.

[SteveW 1,792]

On 9/16/2021 at 5:57 PM, kevinfirth said:

Have you tried reseating the GPU and cleaning the contacts? I recall @SteveW recommending that after a technical explanation that I don't remember and probably went right over my head! 😜

The post referred to was related to unexpected shutdowns, rather than GPU errors. That particular problem was solved by reseating the GPU and power cables.

On 9/16/2021 at 6:02 PM, Ray Proudfoot said:

Nope. Seems a long shot and not one I’ve seen with a DXGI_DEVICE_HUNG search on google. Why would it start the day after P3Dv5 was installed if it was a hardware issue? Huge coincidence.

I’ve never cleaned the contacts on any PC in 30 years.

When PC cards or power connectors are pushed on, or reseated, that cleans the contacts. There's usually no need to manually clean the contacts but if the contacts on a card have been handled, perhaps fingerprints left, these can in time create a poor contact and cause problems.

 

[Ray Proudfoot 8,980]

1 hour ago, SteveW said:

When PC cards or power connectors are pushed on, or reseated, that cleans the contacts. There's usually no need to manually clean the contacts but if the contacts on a card have been handled, perhaps fingerprints left, these can in time create a poor contact and cause problems.

Steve, I’ve had the case opened once in three years to check for dust etc. There was a minimal amount. I’ve never touched the graphics card.

Your suggestion about removing and reseating to clean contacts is valid and I’ve done it with my hi-fi loudspeaker cables. But not computer components.

Oxidisation seems very low risk especially when contacts appear to have gold-plated connections. Not 24-carat I imagine. 😁

[ha5mvo 255]

4 hours ago, Cruachan said:

 My belief is that the likely ‘culprits’ are Nvidia’s drivers and/or Microsoft’s operating system. What’s the betting that a fix will quietly sneak into our experience via a driver/software update with no acknowledgement of ownership.  Shame on them but, it has to be said, far from being untypical these days.

Mike

 

Mike,

I am using an AMD card (6900) with AMD drivers and I can't say my experience is any different than yours. I'm getting those DXGI hung error just as frequently, if not more.

Its most likely to do with the software as no other application is yet to trigger a similar error, at least not for me  - though I hear it's not unique to p3d.

I can predict the risk factors ( PBR, heavy or old scenery and so on) although its definitely not something I can reproduce  with any consistency.

[Ray Proudfoot 8,980]

23 minutes ago, ha5mvo said:

I'm getting those DXGI hung error just as frequently, if not more.

Have you tried the Registry hack?

[ha5mvo 255]

No Ray, no registry hacks or tweaks.

I try to avoid those as this *may* open a can of worms that's difficult to track down later. Besides, if this was really the silver bullet for the hangs, it would have probably been a common knowledge by now I reckon.

[RXP 2,338]

22 minutes ago, ha5mvo said:

. Besides, if this was really the silver bullet for the hangs, it would have probably been a common knowledge by now I reckon.

It will change from experimental to common knowledge only when more people try out and validate this works for them too!

I was having a hardware total lock up problem with the new test PC I got, and there was no solution that I could find as "common knowledge", until I find the solution myself which involves the same Graphics TDR related settings.
 

Quote

From time to time, my Win10 system would just halt with a black screen with no other recourse than powering off. It is very frustrating and you never know what damage to your files will result out of this [...] When you’re using a HDD and you enable sleep mode, it can take more than 2 sec. to wake up and during that time, at least on my test system...

 

More details here: PSA: Solution to HDD and NVidia causing Win10 System Halt with Black Screen

Edited September 28, 2021 by RXP