DCS Causing SteamVR Fail | Description, evidence and how to replicate

[Stal2k]

FWIW I posted a link to this thread on hoggit and within 3 hours, there are already 4 people saying this is happening to them.

 

One guy even RMA'd his headset, this might be more widespread then is being reported, because frankly it doesn't look like a DCS issue at first.

[RealDCSpilot]

@Stal2k

 

 

Just to make sure, i already stated that i had the same problem. More VRAM solved the issue for me without changing any other setting (instead i can go higher now). I know this isn't the answer you want to hear. I basically followed the same procedure with setting additional SS to invest more GPU usage because my 2080ti was bored at around 60% usage. But for a long time i didn't mind that this also takes a higher impact on VRAM usage and lead to video memory overflow and VRAM to disk caching (which looks like to be the source of the problem).

Edited October 6, 2020 by Alec Delorean

[divinee]

I also have this and few guys in our community also has this same problem. Crash mostly happens when opening F10 map in Syria and started occuring couple patches ago.. Also, i'm not sure if its connected to this but if i'm staying F10 map for too long and return back to F1 view, my fps goes down from 45-50 to 35 and stays there. Same fps drop occurs when i change slot or die in MP. After that fps drop it's about 90% probability that pressing F10 will crash steamvr. If i dont' change slot, die or press F10 i can run with 45-60fps even very hevy multiplayer missions without steamvr crashing. I've run with same settings since spring without any problems before this and tried all possible repairs, cleanings, resetting shaders, shadermods, drivers etc..

Edited October 6, 2020 by divinee

[andyn]

I'm affected by this, too. Currently hitting F10 in multiplayer is basically playing Russian roulette with SteamVR.

 

At no point does DCS or SteamVR crash; SteamVR just apparently throws in the towel when the game can't draw frames fast enough and stops rendering to the HMD, which goes black. WMR headtracking stays on, the desktop mirror stays on and DCS continues to run at a slideshow-like frame rates until it is shut down manually. I actually finished one desperate JDAM strike in multiplayer with goggles on my forehead when VR stopped working at the ingress point

 

Frame rates tend to go down over longer multiplayer sessions with multiple respawns and the game has to be restarted at regular intervals to prevent this issue from happening. Usually the Windows task manager shows VRServer.exe eating some 70 % of GPU while DCS only gets 30ish %, when the normal ratio at game startup is like 10:90.

 

Relevant specs are 2070 Super with 8 GB VRAM, 32 GB system RAM, i10600K, HP Reverb, 1.0 PD, 150 % SS. Only the builtin SteamVR left eye mirror that can't be disabled is on.

[FinalGlide]

I'm affected by this, too. Currently hitting F10 in multiplayer is basically playing Russian roulette with SteamVR.

 

At no point does DCS or SteamVR crash; SteamVR just apparently throws in the towel when the game can't draw frames fast enough and stops rendering to the HMD, which goes black. WMR headtracking stays on, the desktop mirror stays on and DCS continues to run at a slideshow-like frame rates until it is shut down manually. I actually finished one desperate JDAM strike in multiplayer with goggles on my forehead when VR stopped working at the ingress point

 

Frame rates tend to go down over longer multiplayer sessions with multiple respawns and the game has to be restarted at regular intervals to prevent this issue from happening. Usually the Windows task manager shows VRServer.exe eating some 70 % of GPU while DCS only gets 30ish %, when the normal ratio at game startup is like 10:90.

 

Relevant specs are 2070 Super with 8 GB VRAM, 32 GB system RAM, i10600K, HP Reverb, 1.0 PD, 150 % SS. Only the builtin SteamVR left eye mirror that can't be disabled is on.

 

This is the exact issue I’ve been having since the last few patches As well. If I use F10 or respawn Vrserver takes up over 50% of my GPU usage and destroys my frame rate. Same as you I’ve had similar settings without issue for a while but this recent issue has made me pull my hair out in frustration. I’ve noticed more recently that using helicopters at low level taxing the GPU on multiplayer can recreate this issue fairly constantly without dying or pressing F10. Happy to test for any of the dev team to recreate things as it’s fairly common place for me atm.

[Stal2k]

So Bignewy, between this thread and the one I started on hoggit we are now in the double digits (13 as of this post) of people directly reporting this issue.

 

Are we yet at the point where someone will (seriously) look at this, or what further action/proof is required? I would hope we are at least past the point of considering this just a "me" problem.

 

**edit: Inadvertantly ended up chatting to BN on discord about this and would like to tone down my statement here. It's been acknowledge and now it seems like the next step is for ED to replicate on their end**

Edited October 6, 2020 by Stal2k
spoke to BN on discord

[Grontar]

Absolutely +1 to OP's message: I have been experiencing the exact same issues he describes, and he describes them accurately (I mean, you learn how to "force" the issues to happen, and how to "cope with them").

 

As a DCS enthusiast, and owner of several of its modules, I feel it is really sad and dissapointing to see how this poor performance / FPS drops / crashes are not a high priority matter to address and fix, but having a new version of something already existing (A-10C II as an example) is. You can argue getting money from a product is important for a company, but having a dwindling playerbase in latest years due to poor optimization should be a big concern too I guess.

 

 

Renewing the old graphic engine is a promise still unfulfilled since years, and optimization to better take advantage of newer technologies should be a must. I agree the solution to the problem should never be "toss more RAM to it", that is just lazy. If you have a leak in your gas tank, you do not buy a bigger leaked one, you just fix the leak. Pure logic there.

 

I hope ED realizes this can have a big impact in their existent and future playerbase / customers, especially when there are other simulators available showing this is posible to achieve, not just a crazy dream.

 

 

Thank you in advance for the time and attention, and kind regards.

[CobaltUK]

The OP observations were valuable and all the dilution by example Alec Delorean is frustrating. Its not changing of hardware or settings that have caused a visible and reported increase of vr crashes.

[Motomouse]

Yes I can confirm that i also get crashes when I use 500% Supersampling in complex multiplayer missions.

 

What I am really astonished about is, that it starts with insanse settings in the first place :-)

 

I prefer it that way. I can get a glimpse of the future. I would not like if DCS would just stop with a boring vram warning. ;-)

[Taz1004]

Yes I can confirm that i also get crashes when I use 500% Supersampling in complex multiplayer missions.

 

What I am really astonished about is, that it starts with insanse settings in the first place :-)

 

I prefer it that way. I can get a glimpse of the future. I would not like if DCS would just stop with a boring vram warning. ;-)

 

No one's running 500% SS or "insane" settings. Everyone's playing as they did in previous versions and are reporting crash "specifically" related to F10 key usage.

 

OP just said to change modules if you want to "reproduce" the bug as when it happens for "normal usage" is inconsistent.

[Motomouse]

I am for fun and exploration ;-) And I get crashes when I do :-)))

Edited October 7, 2020 by Motomouse

[Majik]

Can confirm, also experiencing this.

 

I've had the F10 crash happen a few times now, including once tonight. Also had two crashes mid air tonight. All on Syria in busy MP servers.

[Grontar]

Can confirm, also experiencing this.

 

I've had the F10 crash happen a few times now, including once tonight. Also had two crashes mid air tonight. All on Syria in busy MP servers.

 

 

 

I experience the issues in different maps as well, such as Caucasus / PG / Normandy

[RealDCSpilot]

@CobaltUK

 

I'm only pointing at VRAM usage, DCS seems to need to allocate more ressources since the introduction of the syria map and the latest patches seem to add even more VRAM usage on top of it. I also had the same problem before with my old 2080ti. To test this on your end you can simply lower every setting that needs plenty of VRAM (render resolution with PD setting, visibility range, low textures, lower or 100% SS settings). If the problem still persists than the reason is something else.

A friend of mine just ran into the same problem yesterday on a 1070, we just tested the unlock all liveries mod on our syria server with all modules available and his VRAM just filled up too much, FPS were going down, F10 map troubles. He had to restart the game without the mod to fix it.

Edited October 9, 2020 by Alec Delorean

[Cytarabine]

Having the exact same problem here, was working fine up until the latest patch, now get this intermittent error as described.

 

Something is going on here!

[Stal2k]

@CobaltUK

 

I'm only pointing at VRAM usage, DCS seems to need to allocate more ressources since the introduction of the syria map and the latest patches seem to add even more VRAM usage on top of it. I also had the same problem before with my old 2080ti. To test this on your end you can simply lower every setting that needs plenty of VRAM (render resolution with PD setting, visibility range, low textures, lower or 100% SS settings). If the problem still persists than the reason is something else.

A friend of mine just ran into the same problem yesterday on a 1070, we just tested the unlock all liveries mod on our syria server with all modules available and his VRAM just filled up too much, FPS were going down, F10 map troubles. He had to restart the game without the mod to fix it.

 

To be fair, I kind of get where @CobaltUK was coming from. I know you had the best intentions but as you can kind of see from even this thread, it can be a challenge to get ED to start taking something seriously. I don't blame them either, as they have a pretty full plate with all the other issues.

 

It takes a good deal of work to push past the point of it just being assuming it's something the end-user is causing themselves through either mods or some other bastardization of what you'd call reasonable use of the software. That is the main reason why, despite you trying to offer "solutions," for me anyway, was missing the mark. It wasn't about solving my problem, it was about solving the problem for me and everyone else. That needs to happen on the ED side, not my personal configuration.

 

Crashing a VR HMD in such a severe manner some people are (erroneously) RMA'ing their devices isn't even close to being accepting of, or tweaking settings to accomadate poor (read: unoptimized) hardware utilization.

 

Believe it or not, this isn't the first time I've had to report in great detail a VR issue that others impacted were either unable or failed to articulate in a way that wasn't dismissed. One of the toughest parts of crossing that hump is getting buy-in from ED that it isn't just a "you problem." I'll admit there was a lot less pushback the last time, but it was also easier to reproduce.

 

I just got home and was actually going to make a note about asking BIGNEWY for a status update on this and suggest maybe not pushing a stable update with a known bug of this degree - but that ship kind of sailed :)

[Taz1004]

May not be a VR only issue.

 

https://forums.eagle.ru/showthread.php?t=288445

[Stal2k]

May not be a VR only issue.

 

https://forums.eagle.ru/showthread.php?t=288445

 

While I'm certainly a fan of letting no good crisis go to waste it does sound a bit different. For VR. That thread describes a specific consequence of zooming into a bunch of units in F10. For this thread, it's not zooming in, merely opening F10 will prompt it. DCS doesn't actually crash per say, just your HMD/overlay.

[RealDCSpilot]

@Stal2k

 

Five pages later and you still misunderstanding what i'm trying to say. I'm not backing up ED and i'm definitely not trying to say it's your fault. You know how they work, we are the last line of beta testers.

How about that, please try my suggested steps which may lead to a more precise bug report. Something like "Increased VRAM usage leads to Steam VR fail since the latest patches". Be very conservative with the first step, go pretty low on settings. Watch VRAM usage. Then go higher and try to find the sweetspot. After that, post the settings that worked before and where you are now. I can't do this anymore by myself. I can be totally wrong, it's just my theory from what i've seen on my end. It will help a lot of people to understand whats going on.

Edited October 10, 2020 by Alec Delorean

[BIGNEWY]

To be fair, I kind of get where @CobaltUK was coming from. I know you had the best intentions but as you can kind of see from even this thread, it can be a challenge to get ED to start taking something seriously. I don't blame them either, as they have a pretty full plate with all the other issues.

 

It takes a good deal of work to push past the point of it just being assuming it's something the end-user is causing themselves through either mods or some other bastardization of what you'd call reasonable use of the software. That is the main reason why, despite you trying to offer "solutions," for me anyway, was missing the mark. It wasn't about solving my problem, it was about solving the problem for me and everyone else. That needs to happen on the ED side, not my personal configuration.

 

Crashing a VR HMD in such a severe manner some people are (erroneously) RMA'ing their devices isn't even close to being accepting of, or tweaking settings to accomadate poor (read: unoptimized) hardware utilization.

 

Believe it or not, this isn't the first time I've had to report in great detail a VR issue that others impacted were either unable or failed to articulate in a way that wasn't dismissed. One of the toughest parts of crossing that hump is getting buy-in from ED that it isn't just a "you problem." I'll admit there was a lot less pushback the last time, but it was also easier to reproduce.

 

I just got home and was actually going to make a note about asking BIGNEWY for a status update on this and suggest maybe not pushing a stable update with a known bug of this degree - but that ship kind of sailed :)

 

 

Hi

 

you seem to think we are not taking it seriously, we are, but if we can not reproduce how can we fix it?

 

I will ask the team to take another look, but at this time not being able to reproduce the issue is the biggest problem.

 

Thanks

[Stal2k]

Hi

 

you seem to think we are not taking it seriously, we are, but if we can not reproduce how can we fix it?

 

I will ask the team to take another look, but at this time not being able to reproduce the issue is the biggest problem.

 

Thanks

 

I don't think that currently, I said initially. However, you only said YOU could not reproduce it, and at that point said you were bringing in a tester. There was no follow up since then, so I'm unclear on your progress in reproducing the issue. If I'm to understand based on this response you and the tester you've enlisted were unable to reproduce it then the next step would be to communicate that in this thread. The last note from you on this was you were getting a tester to help, then nothing so how would we know you haven't been able to reproduce?

 

As you've seen, roughly ~24 people or so between here and the hoggit thread are reporting it, so if I were you I'd ask for one of them to help in a more direct manner - much like what was done in the past with regards to the "50% vr performance" troubleshooting. That to me would say you were taking it seriously. Also, not pushing a stable update with an issue like this present would tell me you were taking it seriously.

 

I'm not trying to be overly harsh here, but since you asked how to fix and state reproduction being the biggest issue; then the next steps would be to outline an action plan on how you either 1) plan to scale out testing/hardware configuration of the testers TO REPRODUCE or 2) steps or even intent to look into a possible causes i.e. relation to the recent overhaul of how the F10 menu/memory is handled from about two months ago and most importantly 3) an idea of when you or someone would give a response on the results of those actions.

 

• 9/13 I reached out to you privately about this and supplied a mission file to help encourage this crash as a follow up from someone elses reddit thread on the problem, you said you'd take a look.
  
• 9/30 I checked back in about this via discord on 9/30 asking if you were tracking this internally yet, and you responded "if you have the steps to make it crash please post, I use stream vr everyday and not seen an issue myself"
  
• 10/2 I posted this thread and give you a nudge about it, you again responded that YOU personally couldn't reproduce this and suggested the typical repair/unmod steps.
  
• Several others piled into the this thread, and the one I made on hoggit saying they were having the same issue, you said you were getting a tester to look at it and that was it until now
  

 

So why would I think that you all aren't taking this seriously? Your initial response(s) to me from 9/13 up through page one of this thread indicates to me that despite spurious reports about this issue from other people you were still treating it like an individual user problem which is understandable.

 

After what I would consider an undeniable amount of folks began substantiating the issue beyond you saying you would have someone else look at it, it's been radio silence. For me personally, I've been doing this w/ you since mid September. I understand you must have a lot of things reported, and you can't be expected to follow up with everything but after being told you'd look into it without any follow up 3x and doing what you asked me to do it's frustrating to hear basically the same thing you told me on the 9/13 in terms of some variant of "someone will look into this." without any follow up - even if it's bad news it just seems like the solution is currently to hope it just goes away/fixes itself.

[Eldur]

I will have another tester check, but at this time I am unable to reproduce on my system.

 

thanks

 

To be frank, your system is way too overpowered to make this happen.

EDIT: Didn't see the specs of the OP in the first place. Seems not to be directly hardware related, but my take on this is it happens way easier on lesser hardware.

 

I get these VR crashes (HMD screen freezing up, going black, DCS instaclose, no crash log, no crash message whatsoever, sometimes the Oculusware crashing up at the same time) a lot and I always thought it was just the other random DCS crash that we always had occasionally. I even set up a pagefile to make sure my rig doesn't run out of mem, but it still happens. And it does happen a lot on my stone age rig. And since I always thought it was my limited hardware, I never cared to investigate or report, just tried not to load up too much within a session to some success. Going through F2/6/7/etc and F10, but especially chaning the airframe and even more so exiting the session and loading up another map without exiting DCS is a warranty for this to happen eventually. I always had that since 2.0 Alpha, but it literally went away completely when upping to 32GB, only to come back a lot harder than in ever was when going into the CV1.

Edited October 11, 2020 by Eldur

[GrapeFruiT]

Same issue here.

[GrapeFruiT]

Got the same issues, especially in scenes with heavy loads. I can start and fly mp, when i switch to external views or to F10 map, HMD goes black, then Steam VR crashes, and dcs still runs with the mirror window in a kind of low fps style.

 

I tried Steam VR repair, deinstallation, re-installation could not fix the issue.

Also enable\disabling VR shader mod, DCS repair etc.

 

 

I also updated my nvidia graphic drivers twice. Latest WHQL Builds. Still the problem occures very often in DCS. Before i had never problems like now. Is there any solution. I also googled this issue, and it seems to be a steam VR issue.

 

The only one work around is to avoid, to switch views on heavy load scenes. But still this error / bug annoys and can be frustrating, with steam vr crashes, after restarting the HDM (Steam VR) DCS also quits.

 

 

And this suddenly occured with the release or update since the syria map release. Could also be a steam VR update...Before i never had these issues.

DCS had been running "well" in VR. Still ED pls put mor effort in VR users to optimize performance etc.

 

 

 

My PC specs: Intel i7 7700K, 32 GB DDR4, Zotac RTX 2080 OC

Valve Index VR, Thrustmaster HOTAS (Joystick & Throttle), 2x Cougar MFDs, Wheelstand Pro + Monstertech Table Mount Combination

Edited October 12, 2020 by GrapeFruiT

[Stal2k]

I also updated my nvidia graphic drivers twice. Latest WHQL Builds. Still the problem occures very often in DCS. Before i had never problems like now. Is there any solution. I also googled this issue, and it seems to be a steam VR issue.

 

 

Initially, I thought the same thing, but it happens with the Oculus software as well (tested myself) and according to others reports WMR. Do you have anything you could share on the issue being central to SteamVR, i.e. something from the SteamVR bug forum?

 

In terms of getting this fixed, I guess all we can do is continue to pile on here until it becomes even more overwhelming. You can see from a few posts up the last word is they can't reproduce, my hope is BN answers some of the questions I posed in regards to helping them reproduce this issue or what else we could possibly do to help them narrow it down.

 

So far the big common factors are

 

• Issue is common during heavy memory I/O, i.e. F10 map or external views/module switching
  
• This popped up around the same time Syria launched, though not just a Syria issue
  
• Folks with more than 11gb VRAM seem unimpacted
  
• Happens to Steam/WMR/Rift HMDs
  

 

I personally intend to continue to follow up on this until I'm either told they won't fix it and we all need 3090s now for MP VR or they share some meaningful progress on getting this resolved.