I keep having the issue but currently mostly with PMDG products. I get the crash on 4k and also 2k resolution, but sim runs for a bit longer on 2k before the crash it seems.
I am in contact with support and hope for a solution but otherwise tried all the suggestions I found on the internet. Even reset windows and wiped everything from scratch but problem returned…
I push ALT and F3 both during loading and once the game has loaded, but nothing happens? (both L + R ALTs btw)
get the error now since SU15,some days it work,then it happen,try all fixes…nothing help. only msfs were this happen 
I had this issue for many weeks (after a SU update, don’t remember which one)
Solution:
AMD Driver - tuning option: Set all to standard, but not GPU Frequency - set it -2%
No CTDs anymore…
Yes, my Radeon was undervolted and overclocked a little bit. No Game has any troubles with it, nor the stress tests failed. Just MSFS is very picky with that …
Why -2% GPU core frequency? Because I got the same issue with all settings standard.
For those of you making adjustments in the “NVIDIA App” - is it this?
I am not currently running any additional NVIDIA software, so am not sure if there’s another app, or another way to do it.
TIA
Hello everyone! Have any of you tried changing the setting in the NVIDIA Control Panel “Vulkan/OpenGL present method” from “Auto” to “Prefer Native?”
Also, a good idea to update your FSUIPC. Mine was outdated. Correlation != Causation, but updating seems to have fixed my DXGI crashes so far. It’s been almost a whole couple weeks, and crashes went from nearly every flight (making me rage quit MSFS nightly) to never.
Same here. The irregular pattern here points to a solution that is not related to your local computer. For me it can works for weeks until an update and than it doesnt work for more than max 5 minutes until some weeks after update, and so it goes…
Hello everyone. I thought it’s just me. CTD is back for me after the recent update. I’m using ASUS RTX ROG 2080 Super, i9-10900K, 64 GB ram and sim is on steam located at 500GB NV Me.
I can only play for 4 hours before CTD with error that my GPU is hot, overclock etc.
I play Wukong, the first descendant, Star Wars outlaw for more than 12 hours for each game and no crashing.
Update 9/14/2024: I turn off live weather and the SIM is stable now for me.
Are you using a fixed swap file size?
SUMMARY : 4090 TRINITY OC ZOTAC →
if watchdog error → RMA
if error s_ok (0x0) msfs OR dxgi_error_device_hung …0x887a0006 → switch to DX11, set physix on GPU not auto and let application decide for quality (maybe install windowstop and place mfs2020 always on top → no more crash but FYI note that some GPU (even exactly same brand and model) will not trigger an error when some others will in my experience (as i have one that works fine in DX12 and the other doesn’t but they are the same). I’ll try DX12 again since i got to over 40 hours with no crash
I tried 4 zotac 4090 trinity oc.
One had bad vram, it was an amazon return as you cant find them new anymore.
First one was new from a shop and is back at zotac for repair. It could go for weeks working fine and then give me random watchdog.
Wasted so much time trying everything but i was just a bad gpu.
While waiting back for it, i grabbed 2 other same gpu.
One gives me the hung issue. Edit regiztry set nvidia to performance… doesnt help. I ll return it.
I meam, it should not happen aince same gpu has no issue.
The second one works perfectly in same computer. No error what so ever. A keeper.
I ll build another setup with the rma gpu when it returns.
I chose to stay with zotac dispithe issue as i soent so much time troubleshooting the 1st one i know when it is bad and not.
There is something going on with zotac trinity oc 4090
OK, here are the results of my research.
I tried 4 ZOTAC 4090 TRINITY OC in the same PC (same hardware, same software and driver up to date). I was so tired of the issues i kept having.
First i ever bought was unstable, it would run for days even weeks without issue and then cause all sort of crashes but mainly [DPC_WATCHDOG_VIOLATION]. I returned it after a couple years when i just had enough of this BS.
Got one on online preowned on amazon
→ returned it right away OCCT vram becnhmark showed it faulty, had artifacts on screens as well so clearly Amazon didn’t check this return
Purchased a new one that works perfectly with nvidia detault setting (no need to set to maximum performance the power managment)
Got another new one at same time as the one above.
It was important to me to let eveything on default so even non agressive fan settings as it worked with the exact same other GPU.
I had 2 errors with this one but not as bad as the DPC_WATCHDOG_VIOLATION
This would happen mid flight, it would happen for sure on long flight (like i would launch a transatlantic overnight for testing. This was weird cause stress test like furmark, engine heaven, OCCT,… showed no issue. I had not tried MSI KOMBUSTOR.
(This paragraph is kindergarden stuff for overlockers and knowledgable people but not for others) I did lots of reading about the silicon lotery and GPU peromances being not the same for a same line of GPU. I read that having to tweak a bit your GPU wasn’t necessarly a RMA reason and that the GPU wasn’t necessarly defect. I also read that the main longebity factor of a GPU was low temperature and that running with slightly higher voltage and allow, when necessary the GPU to use more power wasn’t a big deal as well as it was not maxed out at all time. in other words, more voltage (that is applied all the time i think but again, not sure, maybe allowed when needed as i had conflicting info here) and more power draw allowance (110%) allows the GPU to stabilize when needed.
So I decided it was worth seeing if i could stabilize this GPU.
I used ZOTAC FIRESTORM (but you could do so with MSI AFTERBURNER) and deactivated AUTO FAN MODE and set it to MANUAL 100%, i increased the voltage +30mv allowed temperature to go as high as 88C and 110% power. I let frequency on default. Started a flight overnight and it didn’t crash. I even started a hour OCCT vram test and no issue. I noticed that in zotax firestom the automode for the fan uses the curve you can enter manally. indeed, i enter differnt curves and selecting AUTO or CURVE ended up using the curve i entered in CURVE mode (there is CURVE, AUTO and MANUAL mode). This is weird but so you know.
For your information, I had tried to apply the TdrDelay and TdrLevel registry hacks here but when the HUNG fault was not allowed, i had that other error mentioned above.
As this GPU worked without crash (that would happen 100% of the time before), i consider it is stabilized, i will now use lower values and see what is the culprit/saviour in all of these settings and will report back.
Now, i could be all wrong and the error could be cause (device hung) by a bad file the GPU cannot render but i doubt it and feel temperature is the main culprit. Again, will report back
EDIT 1.
The game just crashed. MFS window is frozen and the game won’t disappear until i click ok on that window so I brought up Nvidia control panel, GPU driver is set to maximum performance for power → so that didn’t solve the issue.RIVETUINER shows that the temperature was at 41 degrees celcius at time of crash which is a 36% GPU load, frequency was 2618Mhz and fans were running at 36% which looks more than fine to me. No other task running in background.
So i thought i had solved this issue but I have not.
As you can see the registry entries were made for the TdlLevel hack set to 0 but it didn’t help nor prevent this issue from happening (when it technically should based on a video i saw on youtube). At least, compared to my 1st ZOTAC 4090, I’m not getting a WATCHDOG violation and a computer freeze impositing a complete restart. I feel this is not as bad but if temp is low, what is causing this. As I kept overvoltage set to 30Mv and allowed max power to 110% and max temp to 88C, th eonly change i made compared to last night where it flew all night long without crashing is the FREQUENCIES, I had set everything to minus 200. Also, this flight is above germany (not sure where it was when it crashed and not from EUROPE to KJFK as last night (thinking of bad files the GPU is trying to lload and that would cause this crash). So 2 options left 1. increasing voltage allowed and see what happens or decreasing the frequencies (I’ll start by the 3 BOOST GPU MEMORY then we will do one at a time if it works). I’ll do the same flight.
If issues persist, it’s either a bad GPU or an issue with MFS2020 code.
I’ll try another game after but all stress test i tried didn’t report issue.
Another thing I’ll change is turning off GPU control settings in MSI FIRESTORM to make sure Zotac and MSI utilities are not fighting one another (i doubt it, FIRESTORM updates automatically its settings as soon as ZOTAC firestorm is edited). EDIT : I had turned off MSI AFTERBRUNER monitoring right after last crash and turning it back on (at least ticking the monitoring checkbox in settings crashed mfs * was also trying to activate rivatuner layer at time of MFS2020 crash)
EDIT : Deactivated msi afterburner and rivatuner (both closed and msi has no moniforing access to gpu). Dont think it has anything to do with it but we can always turn back on and see.
Lowered all frequencies by 200. Took off from zueich with tbm930 with unlimited fuel, i am now ovwr syria on eay to dubai and no crash. Lets see what happens tomorrow morning.
If if works, i will nog change anything. I will go fly again befween brussels and cologne ahere i heavy aerosoft sceneries which could have files gpu cant compile (but i doubt it since no such issue with other zame gpu). If works tomorrow, we can most likely conclude zotac gpu are not stable at default oc settings.
One thing i have not tried is to run OC SCANNER maybe it sets the frequencies so fhere is no crash. I ll keep editing and reporting
EDIT running at -200mhz. No more error, a 7 hours flight is undergoing… (zurich to phuket in the “slow” tbm with unlimited fuel). I have taken back up the frequency to 2535mhz so I have set -125mhz in zotac firestom for my 4090 trinity oc. It seems ok as well so far. I set back to 0 the memory. It looked as if it was running at 27% of the load so it felt normal to set memory back to 0.
EDIT : MFS2020 crashed not too long after applying these settings. I’ll restart a long flight at - 200 mhz and will leave the memory at 0 and see what happens (0 being default zotac frequency for the memory frequency). If it crashes, we will try to set the core and boost frequency to 0 and set the memory to -200. I’ll report
EDIT : It crashed again
Now let’s try at -200 on CORE+ BOOST +MEMORY (even though usage is at 29% or so and GPU temp is low) so far so good see screenshot below
MY CONFIG
(please, keep on mind i have another exact same GPU so ZOTAC 4090 trinity OC with no crash so these tests are meant to find the cose of these crashes on one same GPU)
4090 TRINITY OC ZOTAC
DARK HERO Z790 ASUS MB*
14900K INTEL CPU
CORSAIR RAM
EDIT With all at -200… including MEMORY… so far so good no crash and very long flight already. Now, i need to add that i made another change, i uninstalled GEFORCE experience and also in nvidia control panel, i put all antialisaing settings i could see in general and program specific tabs to OFF since the games handle this themselves. It is to be noted I used directX12. If the device hung error comes back, I’ll try to set to Directx11. Once again, i want to say that another same GPU doesn’t have this issue so i’m still convinced this error is not software related 100% of the time (not everyone can test 4 same 4090) I’ll get to the bottom of this
EDIT 5H34 AND NO CRASH (all clocks minus 200mhz + AA tunred off in nvidia control panel + GEFORCE experience uninstalled) so only changes from last flight with a crash is the fact i also downgraded the MEMORY CLOCK (not only CORE AND BOOST) by 200mhz + AA OFF + GEFORCE UNINSTALLED (The windows registry was already edited with the entry mentionned in this forum thread.
EDIT so 7 hours… i deem this to be stable.
LEt’s see of resetting the clock to 0 make it crash (first i’ll keep the memory clock at -200 and reset the CORE and BOOST to 0
EDIT issue came back as soon as i turned the CORE and BOOST frequency to 0. So it looks the GPU is stable when they are set to minus 200. Just prior this, i was able to have 8 hours + stable (i stopped the flight at that point). Now i only got 3 hours 35 minutes at time of crash. I see 2 possible solutions here.
Either i just need to set this GPU at -200 for the 3 frequencies. Or, on previous flight, i was coming and going during the day to the computer to check on the flight, so maybe activity prevents MFS to crash. I’ll restart a session and will leave the flight unattended. we’ll see if it crashes. If it doesnt, this will prove that this bug may not be coming from MFS2020 but from the GPU. to be noted, i had SPAD.NEXT running during this last test and not the one thart lasted 8 hours so i’ll try to discard this as well in future tests. I’ll aslo set VSYNC on for this next test flight…
EDIT it crashed after a few hours at -200.
Bottom line is
I will now place the working GPU in the same PC without changing anything and confirm it works
EDIT ; UPDATED TO LAST DRIVER, RESET SETTINGS TO DEFAULT so no minus 200. I clicked on debug mode in nvidia control panel so it is checked. I ran the game at x16 mode and it made the flight from Zurich to bangkok (don’t ask me why this flight, i have no idea)
EDIT ; had another test with eveything set at -200 and it crashed.
To me it looks like you can’t leave the game unattended too long with DX12
It is once again weird since it works with the other CARD (I’ll put it back in again to make sure).
FIX ; After all of this, I switched back to DX12 which solves my issue. I have the same flight ongoing right now, I’m at 14Hours of flight with no crash. (all settings on auto and no lowered GPU frequencies.
STABLE 16H21 NOW… GOOD TIMING ZURICH PHUKET IN TBM 930 ON TOP HIMALAYAS… (UNLIMITED FUEL)
it poped up at 19H
OK HAVE SOME NEWS THE TBM 930 WITH ITS UNLIMTED FUEL TANK HAS BEEN FLYING FOR OVER 40 HOURS WITH NO CRASH.
STILL DIRECTX11 BUT THIS THIS TIME I MADE 2 EXTRA CHANGES. I LET NVIDIA CONTROL PANEL AA SETTINGS ON DEFAULT BUT SELECTED LET APPLICATION DECIDE SETTINGS AND ALSO CHOSE THE GPU FOR PHYSICS WHICH WAS SET TO AUTO. I BELIEVE THIS IS WHAT MADE THE DIFFERENCE BUT I COULD BE WRONG. I ALSO INSTALLED APPLICATION WINDOWSTOP AND PAID 10BUCKS FOR IT. WORTH EVERY CENT. CHOSE MFS2020 TO STAY ON TOP. FIRESTOM ZOTAC WAS SET TO DEFAULT AND AUTO (SO NO DOWNCLOCKING FOR THIS TO WORK)
OVER 53 HOURS 28 AS YOU CAN READ ON TOP RIGHT LEFT INFO - LET’s TRY TO GO TO DX12 TO SEE IF THE PHYSICS BOX WAS ALL THAT WAS NEEDED
SWITCHING BACK TO DIRECTX12 CAUSED THE ERROR AFTER 32 minuters.
CONCLUSIONS : NO ERROR WHEN USING DIRECTX11 AND WHEN I SET NVIDIA PHYSICS TO MY 4090 and not auto.
WELL… CAUSE OF THE CRASH…I RESET MFS2020 TO DX11 but it CRASH AFTER 50 minutes…SAME CONFIG that just worked for over 40 hours. makses no sense… i didn’t restart cvomputer after setting MFS2020 to DX11. don,t feel i had to…
I also have an RTX 4090 but mine is from Gigabyte - RTX 4090 Gaming OC. I had a similar issue before and in my case motherboard BIOS settings causing the problem. I’m curious, what’s your exact hardware configuration?
edit
The “Reset this PC” function in Windows also helped me.
I updated my posts with config.
Once again, if you read my long post, you’ll see that the config and windows are not causing the CTD device hung error in mfs2020 as the same computer has no CTD, device hung with another same GPU
Disabling the XMP RAM profile did the trick for me as well, thank you for your input !
Been struggling with this issue over the whole weekend and finally found a fix for my system.
I enabled Debug Mode in nVIDIA Control Panel, which basically turns off any overclocking settings on the graphics card. Even if I haven’t overclocked my graphics card in any way it did the trick for me. Maybe it’s because my graphics card has an OC profile by default (MSI RTX 2080 Ti GAMING X Trio).
None of the other solutions provided in this thread helped me. Debug Mode did, and I was finally able to fly the whole Sunday without any crashes.
If this solutions works for you keep in mind that you have to re-enable Debug Mode after every reboot of your system.
Cheers!
MSFS is known to be quite fussy with an overclock.
Some folks have had luck with an O\C, it seems most have not.
Thanx, this help for me
I can’t believe this is what got it to work for me after everything I’ve tried.
Thank you 
