Logging Out-of-Realtime Physics Freezes [public-test]

It never ceases to amaze me how software can behave so differently from one computer to the next. I had horrible stutters and freezes until the latest update and now it's been smooth as butter. :rolleyes:
 
If there is nothing mentioned in the build notes, then more likely nothing was changed and it's just a matter of luck or some changing factor on the local PC. This is what makes it hard to solve for the developer I guess if the issue cannot be reproduced consistently even when running on the same PC. As far as I can see, the only freeze related thing that was altered in the previous update was the stutters in online when someone joins/disconnects.
 
The freeze appears to happen for me every time I'm ~33 mins into the game. After that, no more freezes (although I probably never played for more than 2 hours successively). Sometimes I probably just don't hit or notice the freeze when I'm in the pits rather than on-track.

I never have the entire game/OS completely crashing though. The graphics always come back after a few seconds.

I checked MSI Afterburner for what happens during the freeze. I'll post a screenshot this evening, but basically the GPU activity drops to zero (and it clocks itself to its idle clockspeeds), the CPU does likewise and (V)RAM usage stays stable. I didn't check the disk activity yet.

The join/disconnect stutters have definitely disappeared after the recent patch, but for that Marcel also confirmed it was a regression and would be fixed again.
 
If there is nothing mentioned in the build notes, then more likely nothing was changed and it's just a matter of luck or some changing factor on the local PC. This is what makes it hard to solve for the developer I guess if the issue cannot be reproduced consistently even when running on the same PC. As far as I can see, the only freeze related thing that was altered in the previous update was the stutters in online when someone joins/disconnects.

Exactly. No mention of a fix for this in the notes. All I did was opt back in for the new UI again and (so far) have had no issues. Could be like Tygernoot says, maybe they are happening when I'm in the garage and don't notice. Though before the latest build I was getting stuttering pretty much every lap and freezes were frequent. Bryan Birtwell was on the same server and had no issues. I made no changes whatsoever to my system.
 
I'm hearing you there. Having the same freeze issue in most sessions now, can be driving, then graphics stops but car, sound, FFB, motion platform, keeps going, then (if I'm lucky) the graphics will return but I'm in a wall somewhere. It's extremely irritating when you are 2 laps from the end of the race and doing OK and then you're in the wall with no engine or one of your wheels hanging off (it nearly always happens in a friggin corner) or, worst case, it doesn't come back up and you have to do a reboot to get out of the sim and my expensive equipment is bashing itself to pieces because the sim still thinks it's halfway through a paddock somewhere. (I'll send the bill for that later o_O)
I've been pretty tolerant of the bugs and lack of communication with regard to most of them, but it's wearing pretty thin. It's like S397 don't want to acknowledge that they have this freeze / lag issue, kind of like
"well we saw we had a problem so we got some logs and changed something so it should be all good now, thanks"
That is starting to hurt a bit at this stage.

My 2cents worth. Thanks for listening (if you did)

Totally agree with all of that. I get the graphics ones like you, where the screen freezes but the car runs into a wall, but I also get the (physics?) ones where the car freezes in place then continues (this one shows on replays). I've still got a few more things to test, so I'll see it through until I've run out of options. I'll probably post some more logs when it happens, although it's anyone's guess if the logs are still being investigated.
 
Totally agree with all of that. I get the graphics ones like you, where the screen freezes but the car runs into a wall, but I also get the (physics?) ones where the car freezes in place then continues (this one shows on replays). I've still got a few more things to test, so I'll see it through until I've run out of options. I'll probably post some more logs when it happens, although it's anyone's guess if the logs are still being investigated.
Yep get both types of stutter / freeze.
 
I would also like to report this same issue too. I too get screen freezes for a second or 2 then find myself crashed in a wall when the freeze stops.

I also get small pauses if someone crashes and loses a part of their car like a wing. It does not matter if they crash on the other side of the track to you, I still get a small pause. These pauses are not as bad as the pauses reported above, but I thought I would still report this.
 
I have the same screen freezing problem that you are commenting on, the image is frozen and in 3 or 4 seconds it comes back and I am against the wall. It only happens to me once per session, two weeks ago I ran a 6-hour resistance, we made a pilot change every hour, I did not close session and in my second hour it happened in all the rest of the race and it did not happen again. I have been trying to solve the problem for more than two months thinking that it was my pc, and now I see that no, that there are many people with this ploblem and they have abandoned us. That way you can't run.
 
This only works both ways. Not much point in posting logs if S397 aren't even investigating the issue. Several people have posted their logs already in the past weeks, but currently there's zero indication this problem is even being looked at. No mention of it either in the weekly updates: https://www.studio-397.com/2020/02/build-1117-weekly-updates-2/

Guess it's not a high priority, despite this thread being 22 pages long. Hopefully after the GUI updates they have the resources to fix some more issues, including this one.
 
I also get small pauses if someone crashes and loses a part of their car like a wing. It does not matter if they crash on the other side of the track to you, I still get a small pause. These pauses are not as bad as the pauses reported above, but I thought I would still report this.

This is an interesting development which should be looked into more if you ask me. A car that I was modding had 'wheel' shapes on the collision box and as a result the collision box was making contact with the floor under heavy cornering and braking. This seemed to cause a 'memory leak' type problem. It didn't affect the in game performance that I was aware of (I'm not usually on track for longer than 10-20 minutes when testing), but definitely caused the replays to max the CPU and run in slow motion. It would get slower and slower the longer the replay went on. Fixing the collision box appears to have fixed the problem.

I know this is slightly unrelated to the screen freezing issue, but if I was a gambling man, I'd say they were related and there's something weird going on maybe with the way damage is handled by the game.

I did report this to @Marcel Offermans but whether it's been investigated by any of the devs I don't know...
 
https://1drv.ms/u/s!AiFdgPxJnvUlqCNBSLDnTpmdwnx7?e=szQVJP
Freeze screen of 6 seconds, the game continues but my screen frozen, when it comes back I am touching the wall. In the repetition the image does not stop. The freezing occurs in the last 10-15 seconds last session, lap 15, lap time 14 seconds in the repetition. RealTimeLogging in 1. I am using version 1117 of rF2, all plugins disabled. Please look at it, it can't be that in every session it happens to me
 
Well after a LOT more testing, I'm thinking it's down to having replays enabled. For me it is anyway. I tried setting replay compression to 0, and it had no effect. I turned replays off and it was noticeably better, hardly any stutters after a one hour session, so then I added 20 AI and it was still smooth all the way to the end of a 2 hour session. Usually stutter got worse after a long session, but with replays turned off it seems much better. So the question now is why? Is it hardware related? Is there something that can be done to help with replays on? I don't know, I'm still testing, but with replays off it's much better. I'll update when I've tested more.
 
Unfortunately, turning the replays off, which I've done for a month now, didn't help things for me.
In fact this was a first time I ever got both types of freezes (graphics/physics) during an event.
What's interesting is that the "graphics" type of freeze, I never get twice.
So when it happened during the warm up lap for my quali attempt I was kind of relieved.
But the rubber banding will happen as many times as you like.
The join/leave lag proved to be there and very much alive, now joined by the graphics freeze for a better and more diverse experience.
 
And I don't know but I have posted another incident I have posted above a few days ago, that was again Sebring, and the total of 4 freezes were in the last sector.
Now that may not sound like anything but it is quite the coincidence.
 
Last edited:
Well after a LOT more testing, I'm thinking it's down to having replays enabled. For me it is anyway. I tried setting replay compression to 0, and it had no effect. I turned replays off and it was noticeably better, hardly any stutters after a one hour session, so then I added 20 AI and it was still smooth all the way to the end of a 2 hour session. Usually stutter got worse after a long session, but with replays turned off it seems much better. So the question now is why? Is it hardware related? Is there something that can be done to help with replays on? I don't know, I'm still testing, but with replays off it's much better. I'll update when I've tested more.
Do you buffer replays to RAM (player.json "Record To Memory":true) or default write it from time to time?
 
But the rubber banding will happen as many times as you like.

Rubber banding is due to your CPU hitting 100% usage on physics thread. The other issues are more difficult to diagnose, but rubber banding basically never happens unless CPU is fully loaded. Not much of help probably, but at least it should be possible to figure out in theory why CPU usage gets to 100% and what app is responsible for that.
 
Back
Top