fixed Screen flickering since 2.5.4

[Flappie]

We've already sent several tracks to devs, but unfortunately the tracks don't always give the same results when played by one user.

I need a short track showing the issue at any time. SignorMagnifico's track is way too long, and I already know the Tarawa at night won't trigger the bug for me.

So if you can create a short track showing the issue right away, without using the Tarawa, that would be great.

[SignorMagnifico]

On 7/31/2022 at 4:23 PM, Flappie said:

We've already sent several tracks to devs, but unfortunately the tracks don't always give the same results when played by one user.

I need a short track showing the issue at any time. SignorMagnifico's track is way too long, and I already know the Tarawa at night won't trigger the bug for me.

So if you can create a short track showing the issue right away, without using the Tarawa, that would be great.

I just recently formatted my drive and installed Windows 11 on a new NVME SSD, and the issue persists on the same track I sent earlier. I'm able to reproduce the issue every time when watching the track by using CTRL + Z to speed up time. It takes me about five minutes of watching the track to get to the issue. (About 10:23AM mission time on final approach right up until landing.) This is on as fresh of a DCS installation as one can get and on the latest version as of today. (DCS 2.7.16.28157) Forgive me, but I don't see how track length is a problem when I can reproduce the issue every time in about five minutes by speeding up the track playback speed.

As a workaround, the devs could try editing the mission file from Mi-8 The Crew Campaign (1.1 Salon) to an air start at 10:23AM and land between the cones. I'm unable to edit the mission file to achieve this because DLC campaigns are copy protected.

 

On 7/8/2022 at 10:25 PM, SignorMagnifico said:

The texture flickering issue appears to have returned in the latest DCS version 2.7.15. 100% reproducible for me when flying the second mission from Mi-8: The Crew Part One Campaign. No texture flickering occurs at the beginning of the mission, but it starts right at the end of the mission when landing at the same spot the mission started. I remember the previous conclusion regarding this bug was that it had something to do with mirrors and the sun's position, and the mission takes place close to sunset. The sun changes position by the end of the mission. I disabled all mods, deleted fxo and metashaders, repaired my DCS installation, and I was still able to reproduce the issue. I hope this helps with resolving this bug. I recorded the issue happening with mods disabled. Track file attached at the drop box link, and the issue can be seen happening in the YouTube video starting at 1:09. https://www.dropbox.com/s/pqb4uw92g1qul8f/textureflickeringbug.rar?dl=0

DCS Version: DCS 2.7.15.26783 Open Beta

Hardware: DX diag attached.

 

 

DxDiag07-08-22.txt 94.33 kB · 5 downloads

 

 

Edited August 4, 2022 by SignorMagnifico

[Nealius]

I started having this issue in 2.7.16.28157 now as well. The times it's worse is when close to the Super Carrier superstructure with MSAA on. With MSAA off it doesn't appear on the Super Carrier, but will still intermittently appear in warbirds on the Channel map. It seems to be worse in VR on my system. Trying to get a short track file.

 

[SignorMagnifico]

Managed to replicate this issue again. This time, the track is much shorter because the issue happens at the beginning of the mission. You can see the flickering starting at about 20 seconds into the YouTube video. I attached a track where I was able to replicate the problem as well as a DXDiag and screenshots of my DCS and Nvidia Control Panel settings. I noticed that the issue could be linked to MSAA settings within DCS. When I toggled off that setting, I wasn't able to reproduce the texture flickering when playing back the track. Running DCS 2.7.16.28157. https://www.dropbox.com/s/s9tzwkv0fg7nxt8/DCSTextureFlickering08-14-22.rar?dl=0

 

DxDiag08-14-22.txt

[Nealius]

On 8/15/2022 at 3:36 AM, SignorMagnifico said:

I noticed that the issue could be linked to MSAA settings within DCS. When I toggled off that setting, I wasn't able to reproduce the texture flickering

I can concur with this, however since upgrading my NVIDIA drivers to 516.94 the flickering seems to have disappeared (for now). What driver version are you currently on?

[SignorMagnifico]

11 hours ago, Nealius said:

I can concur with this, however since upgrading my NVIDIA drivers to 516.94 the flickering seems to have disappeared (for now). What driver version are you currently on?

Same driver version. It’s in the DxDiag I attached.

[Nealius]

Driver solution didn't last long. Reproduced from ramp start, noticeable immediately from jumping in the cockpit.

(Track larger than 5mb due to other AI units present I guess): Google Drive

[Roastie]

On 7/22/2022 at 6:39 PM, Flappie said:

What I meant is:

I can reproduce the black square bug occasionally.

The track you posted does not trigger the bug on my PC.

Someone else posted a very similar track before, and it didn't trigger the bug on my PC either.

So it seems some black squares are kind of universal, and others are only affecting some users.

Are you using an AMD card? Because some people think the black square is an AMD thing. I'm using an AMD card, and can reproduce 100%.

[Flappie]

No, I have an NVIDIA GPU. And as I said before:

 

[RED]

Just adding this as it is related:

I see this with all picture in picture stuff and running out of memory (with e.g. high object counts in the mission). My guess is that it is a type of memory leak where there is not enough memory available to render picture in picture things. How it looks like in the TGP for me:

 

[Roastie]

On 8/19/2022 at 2:10 PM, Flappie said:

No, I have an NVIDIA GPU. And as I said before:

 

Eh, but the black square problem could be an amd only problem. Shouldnt you also test with amd cards? Or is dcs the only game on earth only supported by nvidia?

[MAXsenna]

35 minutes ago, Roastie said:

Eh, but the black square problem could be an amd only problem. Shouldnt you also test with amd cards? Or is dcs the only game on earth only supported by nvidia?

Flappie is not employed by ED, so I don't think we should expect him to get another GPU.  

Cheers! 

[Flappie]

True that.

Yet, I keep looking for a cheap and old AMD laptop in order to help more. 

Some professionnal ED testers do have AMD GPU by the way. I'll ask one of them to test the latest tracks, as soon as I get back from holiday.

[Roastie]

3 hours ago, MAXsenna said:

Flappie is not employed by ED, so I don't think we should expect him to get another GPU.  

Cheers! 

ED Beta Testers

Made me think Flappie gets payed. Sorry bout that.

1 hour ago, Flappie said:

True that.

Yet, I keep looking for a cheap and old AMD laptop in order to help more. 

Some professionnal ED testers do have AMD GPU by the way. I'll ask one of them to test the latest tracks, as soon as I get back from holiday.

Sounds good m8, dankjewel.

[Flappie]

37 minutes ago, Roastie said:

ED Beta Testers

Made me think Flappie gets payed. Sorry bout that.

I understand. It can be confusing.

The simplest way to know is by the nickname color: dark blue = ED crew, light blue = third parties.

[Heroman30]

I can confirm that I also have the screen flicker! Thankfully after reading this thread it confirms it’s not a hardware GPU issue as other people can replicate.

I have an Nvidia GPU (3080) and I think it’s to do with the angle of the sun from the horizon in certain lighting conditions.

Do we know if the devs can replicate this issue often and if it’s planning on being resolved in the next beta update? Thanks.

[Flappie]

13 hours ago, Heroman30 said:

I can confirm that I also have the screen flicker! Thankfully after reading this thread it confirms it’s not a hardware GPU issue as other people can replicate.

I have an Nvidia GPU (3080) and I think it’s to do with the angle of the sun from the horizon in certain lighting conditions.

Do we know if the devs can replicate this issue often and if it’s planning on being resolved in the next beta update? Thanks.

The issue is reported but the tracks often reproduce the bug at random, which isn't good for debugging. A first fix was brought some months ago but it was solely about mirror-induced flickers. The main issue is still here (it started with initial 2.7 release, I think).

Apparently some of you, especially AMD GPU owners can reproduce this issue on purpose. I'll send over the latest tracks from here to the team when I get back home (in one week).

Edited August 24, 2022 by Flappie
typo

[Nealius]

8 hours ago, Flappie said:

The issue is reported but the tracks often reproduce the bug at random

In my case it seems to happen when the fxo folder gets to a certain size. One of those many fxo/shader bugs again. 

[Cycolonic]

Was directed here by Flappie from another post. It's definitely not an AMD graphics card only issue as I'm running a geforce1080.

So considering that the track files which are relatively small and unlikely to capture what's happening at the rendering level (where the bug is occurring - actually, I read somewhere that someone who has this bug could play someone else's perfectly clean track file and see the bug when he plays it back on his system!), what can we do to help you guys figure out what's going on?

If the track file is not going to be of diagnostic utility, is there a debug mode or something that we can turn on that can capture a deep dive on the state of the sim at a particular frame that we can run for you? I really want to be able to play without constantly having to scrub a mission because I got blacked out again.

How can I help you help me?

 

[Flappie]

We both have the same GPU. If you have some very short tracks showing this issue, please send them over.

[Cycolonic]

Hi,

Ok, I played until the black block showed up and completely obscured my front view.

Take off was ok but a few minutes into the mission, a big black block takes out the entire front view. You can see me move my head from side to side and you can see how I can kind of see the head rest looking back left and right.

One thing to note is that even with black block, I can see the plane ID's through the black block. So that might be a clue as to what is futzing up here. Some kind of lens flare effect gone bad? Fog?

If you are able to see the black block that I am seeing, that is...

After you take a look at this, can you confirm whether you can or cannot see the black block? And to be clear, the last parts of this clip is all black block.

Thanks.

CCBlackBlock.trk

Edited August 25, 2022 by Cycolonic

[Cycolonic]

AHA!!!!!!

I confirmed it for myself - you will not see the black block in the track file!

I just tried playing back that track file that I sent you and the black block is not there even for me! When the black block showed up for me during play, I start looking around wildly left and right because I was surprised by the black and seeing if I can see around it or get it to disappear before settling in on looking forward so I can show that the plane ID labels are still visible through the black block. But in the track file playback just now, everything's fine. I just start erratically looking around for no apparent reason until the track file ends.

Further - this means that for someone who does not suffer from the black block problem, no matter who sends you a track file, you will never see a black block. And for someone like me, who does suffer from the black block problem, I will probably never see the black block manifest in the same way twice. Sometimes, my playback will look just fine. No black block. Other times, I'll get a black block, but it will be spontaneously generated in real time at different points in the trackfile play back and in different ways - for any given playback session.

I suspect that the reason why this is is because the black block is not a predictable phenomenon in the graphics processing path. Sometimes it shows up, sometimes it doesn't. And that is true even down to the track file recording. Correct me if I'm wrong but the things that the track file is recording is the kind of things that would be sent from my computer to a multiplayer server - it's just the data necessary to update the sim. You want the track files to be as small as possible so why would the devs include debug level information being sent to and received from the graphics card?

Can you confirm that you do not see the black block in the track file?

I'm not a game dev but I'm a pretty informed PC gamer with decades of experience. I suspected the track file would not replicate the black block. Especially if you don't see the black block at the end of my track file, please believe me that a track file is the wrong way to pursue this problem. And I want you to believe me because if you pursue this problem down this path, you will never be able to fix it. And I really want you to fix it.

 

---------------

Now then,

what can I do to help you help me? What OTHER diagnostics can I run on DCS running on my system to help debug what is happening at the deeper levels where this bug is occurring? Is there a debug or diagnostic version of the executable that I can run or something? It can be slow as molasses, that's fine. I just want to help you help me.

EDIT: oh, I just read that you are not an ED employee? Am I talking to the wrong party? If so, sorry for pestering you. Can you direct me to the right party?

Thanks,

CC

 

Edited August 25, 2022 by Cycolonic

[Flappie]

It's OK. That's what I do: collecting Intel for devs to dig.

I see you have now understood the whole problem with this bug. And it gets even more complicated when testing with different builds. By the way, I think everyone has this bug happening. Some users, especially AMD users, might have it happening in some additional cases. We know low sunlight effect has a role in this, and it seems the Tarawa deck lights also has a role - this time for AMD users only I think.

I'll test the latest tracks from this thread when I get back home.

And as I said before, I'll ask a dev equiped with AMD GPU to test these as well.

[Cycolonic]

Cool. And thanks much.

But I've been saying from the beginning that the track files will not contain the error that we're seeing and are therefore a diagnostic dead end. That's why I questioned the gathering of track files in the first place. IMO, it's a waste of time.

I have a feeling you don't quite agree with that sentiment quite yet and that's fine. I only ask that you keep that possibility in your mind when you go through the track files and hopefully sooner rather than later, you can get the devs to concoct somekind of debug executable that would provide a deeper dive into the render pipeline that will be useful in hunting down the problem.

I didn't record a video cap of my session but this is pretty much what it was like except instead of looking down, i was looking at my head rest looking right and left: 

 

Edited August 26, 2022 by Cycolonic

[Flappie]

Let's agree to disagree, then.

Track files never 'contain' bugs, but they represent a situation which led someone to seeing a bug. They contain time of day, weather settings, units, context, actions... which led a user to witness a bug.

So, they're often helpful, even when we, testers, can't see the bug for ourselves.