Jump to content

Nevada crash


Avernus

Recommended Posts

Hi

 

I have experienced a very persistent crash while flying the Mi-24P on the Nevada terrain.

 

Approximate Coords: PV87 N 35'55'9 W 114'55'47,  construction area north-east of solar panels.

Flight route:  Roughly 245' 7km from Boulder City Airport.

 

edit: Here's a quick and nasty track which shows the issue at slightly different times when near the area each time the track is played.

 

I've noted  a dxgi device removed error driver restart will follow with most but not all attempts.

Setting GPU to -50% power target/underclocking/undervolting had no effect on the crash.

 

Thanks.

server-20210701-235507.trk dcs.log


Edited by Avernus
catn speel
Link to comment
Share on other sites

  • 3 months later...

Hello again,

 

Has this issue been looked at?

 

It appears the Hind is not the cause for this, as I have flown several aircraft over this area with the same results.

 

I have Reinstalled OS clean along with DCSW, file corruption is unlikely in this case.

No mods or alterations, Just a clean DCSW install.

 

Flying over this area will eventually throw DCSW out with a 'stopped working' error.

ATI Radeon driver also restarts in some cases, but not always.

 

I've set the card to -50% power and underclocked it a bit to rule out any hardware issue, This did nothing to supress the crash.

 

 

Link to comment
Share on other sites

  • 2 weeks later...
  • 4 weeks later...

Going to re-open this issue.

I now have a new pc with all new hardware (except Hotas/Trackir/Pedals/Screen), Fresh install of OS and DCSW, This problem is very much still active.

Approximately same place, same exact error.

 

New gpu with power levels throttled to -50%/clocks @ stock and cpu features disabled to be sure of any potential factory OC issues.

DCSW crashes followed by the Radeon driver reset as usual.

 

Tried an old GTX 980 card, windows also reset the gfx driver around the same location.

This is NOT a hardware issue.

 

Other maps Caucusus/PG/Normandy/Marianas seem fine for hours..  Flying in NTTR near the location described results in a crash 8/10, with the Mi-24 having the most problems thus far.

The game will hitch a lot before throwing a crash in this instance.

Track file provided is not 100%, but if i play it enough times the crash is certain.

 

Although i have my doubts as to whether my screen configuration is a potential suspect,

Is it possible Displayport+10Bpp features enabled could cause such a specific problem,  Just for NTTR in THAT spot?

 

 

 

Link to comment
Share on other sites

  • ED Team

I can't repeat the crash. Conditions: Mi-24, Nevada map, flight from Boulder City Airport to a point with coordinates 35'55'9 W 114'55'47.
I flew between two construction cranes and flew over the construction area. There is no crash game.
Please attach a dcs.log after the crash of the game. There are also no crash reports from your account on our server.

Link to comment
Share on other sites

  • ED Team
15 hours ago, Avernus said:

I've set the card to -50% power and underclocked it a bit

The maximum frequency of the GPU at rest is 1250 MHz, and the frequency under load should not exceed 1630 MHz.
What frequency did you set?
Operating System: Windows 10 Pro 64-bit (10.0, Build 18363)
Try to install all Windows updates.
You have a pretty powerful video card, but the diagonal of your monitor is also pretty big. Try lowering your game settings and testing the game in that "cursed" place in Nevada.

Link to comment
Share on other sites

Thanks for the response,

 

The default frequency is set @ max @1630, i manually tweak the pwr to +50% and use a custom fan curve which maxes out at +60'C 3200rpm.

I've tested with the Vega64 set to -10% core underclock @1467 and power to -50%, fan curve maxed out flatline.

Vram is detault @975 stock/auto voltage, tested @900 to be sure also.

For your amusement, I've attempted the test at 1920x1080 instead of 3840x2160, and set ui scaling to 1.

All other settings remain as they are, DCSW crashes following a lot of hitching and small pauses with the mi-24, ending in the same crash.

The position isn't always the exact same, but always within 5km of that area.

 

As noted above, the new card is an ATI 6700xt, same exact problem..  and an older Nvidia gtx980, same result with a different gfx driver reloading affair than Ati.

This is also a completely new pc.

With the only remain components being Displayport 4k monitor, Creative ZxR, TMWH, Trackir, Pedals, and the main SSD for OS/DCSW on the same drive, oh and the tower and cables :)

SSD was 'cleaned' using diskpart and repartitioned etc before the OS install and a clean install of DCSW.

SSD was set as GPT instead of the prior MBR.

 

I have pinned Win10 to 1909 as the newer build has messed something up with game controller hid devices and system standby/sleep.

However during previous tests on the newer build 21h2, the outcome is the same. DCSW crashes, then gfx driver reset.

 

To rule out and power/heat related potentials, I've run a 24 hour prime95 torture on CPU, and the MSI Kombustor program to try set fire to the gfx card.

both ran for hours with no issues, while having the gfx maxed out @+50% power.

 

with the problem following a new system, and persisting over 3 different gfx cards, it seems unlikely to be a hardware issue.  or can i be just THAT unlucky?

Only this map in that area, Surely not.

 

Regards

Link to comment
Share on other sites

  • ED Team

On the one hand I see that you have a problem not only with our game.
From the latest DXDiag:
Problem signature:
P1: DOOMx64vk.exe
Problem signature:
P1: glquake.exe

On the other hand, I would like to help you solve this issue. I will ask some more testers to check this section of Nevada.
Also ask you to reinstall the video card driver from scratch on the official ATI website.
And I want to clarify the flight conditions again:
1. Map of Nevada, coordinates PV87 N 35'55'9 W 114'55'47, construction area north-east of solar panels.
2. any module (not just Hind)

And we are waiting for new reports

Link to comment
Share on other sites

41 minutes ago, maximov said:

I will ask some more testers to check this section of Nevada.
And I want to clarify the flight conditions again:
1. Map of Nevada, coordinates PV87 N 35'55'9 W 114'55'47, construction area north-east of solar panels.
2. any module (not just Hind)

And we are waiting for new reports

Report:

Flew with various aircraft (F-16C etc....) + the UH-1H (i don't have the Hind):
Take-off from Boulder City and overflew the coördinates given, no crash, not in any module.

Ran the track provided by Avernus in the first post, DCS installed Mi-24P module before running because that was missing to run the track.
Track runs fine, tried a few times, track plays through to the end and asks then to "refly", no crash at all.

Win11 Pro 64-bit, Ryzen 5800X3D, Corsair H115i, Gigabyte X570S UD, EVGA 3080Ti XC3 Ultra 12GB, 64 GB DDR4 G.Skill 3600. Monitors: LG 27GL850-B27 2560x1440 + Samsung SyncMaster 2443 1920x1200, HOTAS: Warthog with Virpil WarBRD base, MFG Crosswind combat pedals, TrackIR4, Rift-S.

Personal Wish List: A6 Intruder, Vietnam theater, decent ATC module, better VR performance!

Link to comment
Share on other sites

Ah, the 'crashes' of doomvk/glquake are intentional, steam glquake is beyond out of date and crashes always, had to manually update.

Doomvk, was terminated by user on startup.

 

As for DCSW NTTR, there needs not be an aircraft used in my case, just a freecam fly over the area, give it time, perhaps push the resolution to 3840x2160 via VSR/DSR.

Fly the camera low enough around the area, a simple once over may not be enough.

 

This crash reminds me of when DCSW would throw me an error and crash when the FOV was set to  or exceeded 150' with multiple screens, early 2.x versions.

ED 'fixed' it by limiting the FOV to 140 which was impractical, thus i switched to a single screen.

 

I'll try running the test again at 1920x1080 resolution with the vega card and report back.

Edit: same result, attempting to extend win10 TDR delay via registry.

 

 


Edited by Avernus
catn speel
Link to comment
Share on other sites

Extending the TDR resulted in the same driver crash, and this time DCSW froze with no error on exit.

From the HW logs of all 3 cards, I have noticed one thing in common.

The moment before the crash, the GPU clock will exceed the maximum set frequency.

For example the V64 card is set to 1630(max) and the log and LCD screen on the keyboard will display 1650mhz, just as the program crashes.

I have a feeling this may be related to the power target value.

Thoughts?

Link to comment
Share on other sites

Ok here's an update on the situation.

I've reverted driver for both Radeon and GeForce cards to 19.2.2 and 362.00 respectively.

For the Radeon V64 card, the Max GPU frequency has been set to 1632 as default.

For the GeForce 980, the cards base clock is set to 1203, and boost clock varies with a approximate max of 1304.

 

For both cards, using driver Radeon 21.10.1 and Nvidia 496.13, These values are slightly increased and display higher variance clock speeds during positive throttling.

Potentially resulting in this type of crash, of which DCSW seems to be very sensitive to, while other games such as Metro Exodus have no problem for hours.

 

At the time of writing this, The Radeon V64 card has been hovering around the described area for close to 30 minutes now, which is potentially better than the usual random but small timeframe window as previously experienced.

Both cards appear to be a lot more time sensitive to power and clock throttling with these earlier driver sets.

The Radeon card doing it's best to not exceed 1632Mhz at +50% power and full fan speed, Core and Vram clocks set to 'Dynamic'.

The Gtx 980 is also stable, no matter what i do, even overclocked +200mhz offset for core and +700 offset for Vram @+25% power target and maxed out fan speed.

 

I have yet to test the Rx 6700xt, but have my doubts as the driver version is likely to predate the hardware.

Preliminary conclusions:

Newer isn't always better.

Both companies Ati/Nvidia may be attempting to push the cards harder than expected.

DCSW is attempting to use more power and clock value variance than the driver will allow in very specific cases, which the other maps do not exhibit.

 

DCSW NTTR on V64 has been running hovering at low speed low alt freecam for over 30 mins now.

This problem appears to be related to drivers and NTTR gpu/power usage.

Newer Nv and Ati card users may have to hunt for earliest possible driver versions which predate the hysteris update, which in turn appeared to have lowered sampling rates from the Hardware, causing potential overshoots and subsequent crashes, regardless of underclocking attempts, 

Or request the Gfx driver teams to reinstating the faster HW monitoring update rate.

Regards.

No crashes this time. Not a hardware issue x3 :)

Link to comment
Share on other sites

  • ED Team
6 minutes ago, Avernus said:

Crash, sent report via dcs.

No dx11gideviceremoved error, no gpu reset this time.

2021-11-12 08:03:10.732 WARNING LOG: 258 duplicate message(s) skipped.
2021-11-12 08:03:10.732 ERROR   DX11BACKEND: failed to create structed buffer. Reason: DXGI_ERROR_DEVICE_REMOVED
2021-11-12 08:03:10.732 ERROR   DX11BACKEND: CreateShaderResourceView FAILED. Reason: E_INVALIDARG

Link to comment
Share on other sites

  • ED Team

I had a case in my practice where a user replaced an old 1080 graphics card with a persistent driver error.
He bought a 3090 and got the same error. The user was very offended with us. In the end, this situation was resolved by replacing the video card with a new 3090.

Perhaps this is not your case. And there is indeed an error.

Link to comment
Share on other sites

Ya, i saw that.

These are 3 different cards, using ddu to clear drivers on a completely new system, with only NTTR in that area affected.

Very strange problem, Could displayport be a potential cause perhaps?

There was a wierd issue i had with an MSI card that refused to boot the system at all, no POST if the DP was connected at power on, HDMI was fine.

RMA was not approved due to no actual fault of the hardware.

I guess no NTTR for me anymore?

This is unlikely to be resolved by anything less than a wildcard lucky dip, as buying new gfx cards over and over is not a practical solution 🙂

 

Maximov Humour me if you will please.

Set the graphics driver to the maximum power target allowable and retest this issue.

Do not limit framerate target, set texture quality in driver to 'high quality' and disable 'surface format optimisation'

Use freecam mode to fly around the area and allow the gfx to plateu for thermals and power/clock levels.

Use high speed at low alt for as long as you can tolerate. use various camera angles to throw the gfx power/clock values around a bit.

The time of crash for me would appear to be random.

I'm going to attempt this test again on a friends PC later when i get the opportunity.

 

Thanks again.

 


Edited by Avernus
catn speel
Link to comment
Share on other sites

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...