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

Neawoulf said:
Some of these points sound interesting and i haven't tried them yet. But i can't a lot of information about some of them and to be honest: I'm not sure i want to mess around with any system settings to get just one game to run better. Also i'm afraid those changed may cause issues somewhere else.



Haven't tried this and i can't find any information in google about it. How do i change this? Is it a BIOS setting? And does it work with all mainboard/cpus (i'm using a Ryzen 3600X on a Gigabyte X570 Aorus Elite)? And what exactly does it do? Can it cause issues anywhere else?
Click to expand...
NUC126JPG.JPG

NUC126_on_DeviceManager.JPG


This is not used by CPU control, it is only for user to see what voltages there are, cpu/bios values are handled on other hardware, so safe to disable.
If you see that "NUC126" and values under it on HWINFO64, then it's enabled, and after disabling and restarting HWINFO app there is no NUC line when successfully disabled. On upper picture it is enabled to show you what it shows.
yes it might be a bit hunt to find right USB Input Device, but usually its one of last ones.

Neawoulf said:
Basically the same questions as above.
Click to expand...

In BIOS called AVX offset, change it to 0, if you have overclocked heavily your cpu, then just remember at it will pump heat on synthetic stress testers, my i7-8700k is pumped to 5ghz all core, and AVX 0 will put on those testers thermals pretty high, none of actual real world app do that.

Neawoulf said:
I could do that, but i heard it can cause data loss if something bad happens (system crash, power loss etc.) Is it safe to use on the game SSD only (and does it do anything if not enabled on the system SSD)? Don't really want to risk anything with my system SSD.
Click to expand...
only marginal risk increment, can be activated/disable per disk
disk_cache_flush.JPG

disk_cache_flush2.JPG


Neawoulf said:
Just tried to find this in the config files ... is it in the player.JSON? Only thing i can find there is "Delay Video Swap" and it's on "false".
Click to expand...
player.json
inactiveslöeep.JPG


Neawoulf said:
I read about this some time ago, but i also read that windows doesn't use RAM compression if there's enough free RAM left. I have 32 GB, so i don't think rF2 + Crewchief + VR software will ever use all my RAM. Is it still worth a try or is this something i can ignore with 32 GB of RAM?
Click to expand...
windows RAM compression is always activated as default, how much it will analyse/use you ram content is up to windows.
you can check status of state on PowerShell, and sort of active compression is shown as @Lazza told.
mmagent_stat.JPG


texture compression on player.json:
textures.JPG


my nvidia system info and settings
nvidia_SI.JPG

nvidset1.JPG

nvidset2.JPG
 
@Neawoulf AVX is not a problem to you, just noticed at you are running on Ryzen, different thing on there.. Probably not using NUC either..no NUC used on GB boards, MSI will use..
 
That's a lot of info, thanks! Will test some of that some of that later. I already did those changes to the player.JSON file, but maybe disabling that RAM compression helps.
 
OdeFinn said:
@M Lynn which tricks you have tried for fixing it?
Ok my case stutters originally started alongside with some windows update, after that I have altered several settings and current situation is stutter free.
Kinda list of my changes,
dropped older nvidia driver (439 if recall right) because then newest wasn't working fluently with shadow play and older gives better performance.
Disabled cpu's "external" SA sniffer, chip on mboard, controlled thru USB2 logic (this makes dirt rally 2 stutter a lot)
Lockeed AVX 1:1 with cpu multiplier
Updated ignore list on antivirus
Switched off windows disk cache flush
Switched off replay compression - game
switched off texture compression - game
Switched off background delay timer - game
Switched off windows RAM Compression

What things you have tried?
Click to expand...

Thanks for the suggestions.

I've tried a lot of things, more than I should need to for a game that should work well with default settings. Most of this stuff we are trying should never be needed.

I tried installing new drivers for everything, turning off plugins, different video modes, using third-party apps, various tweaks. Hours spent testing, reading forum posts etc. One thing I found that cured the micro stutters for me was buffering replays to RAM. But, I still get the big freezes that end with the car running into a barrier, often but not always around 33 minutes into a session. So annoying when on a quick lap, but if it happens in a race then your race is over.

A lot of those things you mentioned are platform-specific so don't apply to me. Of the ones that do, I've now added my rF2 folder to AV exclusions, so we'll see on that one (not expecting much), I won't turn off disk cache flush (I shouldn't have to and I won't risk data loss), I've already tried turning off replay compression, I can't see windows RAM compression affecting anything but I'm willing to try it, I'll also try texture compression. Background delay timer is a new one on me, where do I change that please?
 
M Lynn said:
I won't turn off disk cache flush (I shouldn't have to and I won't risk data loss),
Click to expand...
Risk of losing current replay or other just waiting to be writed stuff isn't big deal, it just sounds scary thing, but actually it isn't, if you want to make sure at that is not a case, then do it a safe way, switch off entire caching for test, then you don't have any risks. That is the upper checkbox.
M Lynn said:
Background delay timer is a new one on me, where do I change that please?
Click to expand...
Player.json around 207 row.
 
Ah, you mean Inactive Sleep Time? Can't see that helping, it only does anything when rF2 isn't the active application.
 
New log. I turned off texture compression, went into a session, 33 minutes later, surprise surprise, big freeze and crashed into a barrier. So, texture compression does nothing to fix this problem for me. The full log and replay are here https://www.dropbox.com/s/1hfz4fu4rcv38qi/Zandvoort Freeze at 33 Minutes.7z?dl=0

The freeze is at the end of the replay.

Note the remaining time on the dash of a 2 hour session.
 

Attachments

  • Clipboard Image.jpg
    Clipboard Image.jpg
    239.2 KB · Views: 289
Last edited:
M Lynn said:
New log. I turned off texture compression, went into a session, 33 minutes later, surprise surprise, big freeze and crashed into a barrier. So, texture compression does nothing to fix this problem for me. The full log and replay are here https://www.dropbox.com/home/Public?preview=Zandvoort+Freeze+at+33+Minutes.7z

The freeze is at the end of the replay.

Note the remaining time on the dash of a 2 hour session.
Click to expand...
It would be very good if you could share the link to the file. Not sure if this is your local link and it works. But we cant download it.
 
@M Lynn even some of those doesn't feel to make sense, please test all, then you know and us at those aren't your case.
I had freezing problems, not anymore, those are things what I did and get it end.
That repeatable ~30min sounds to me lot of data io problem, even disk writing, memory dumping, or even schedule memory scan, but definitely data io problem.
There is two windows services which are acting bit on that way, for testing those could be disabled, full memory scan, run by AT scheduler, and superprefetcher service, names aren't probably correct, but I can check those later and put here. Those aren't suggested by me, only for testing purposes, because those deactivate bit more windows functionality, nothing drastically.
 
@OdeFinn Thanks, I'll try your suggestions. I'm aware of superfetch and I have that enabled, not sure about full memory scan though. I agree that it could be IO related.

@Jimmi Allison I'm on the current non-beta branch at the moment, old UI.

@lagg Thanks, I've just looked at that thread. Sounds like the same problem, although it also happens offline.
 
OdeFinn said:
Windows+rf2=mismatch, props to microsoft.
Still going to data io, memory io, disk io, all are data io :), windows wants to do something what makes cpu hang on simultaneous processes, fighting cpu time.
Click to expand...

You could be right. It can't all be down to Microsoft though, other games aren't doing it, there must be some sort of workaround.
 
M Lynn said:
You could be right. It can't all be down to Microsoft though, other games aren't doing it, there must be some sort of workaround.
Click to expand...
Rf2 base is older than any current other sims, older than win10 and tweaks inside windows memory management. Probably some urgent 0day vulnerability patch broke something.. Who knows, but I'm willing to believe at main reason is on windows and they broke something old what rf2 is still actively using.

lagg said:
Yes but why at the same time?
It's strange.
Click to expand...

There's ton of different timed triggers on AT scheduler tasks, huge amount of different triggers waiting to break when we're gaming.
Is it counting "idle" time wrong end launches something.. Who knows.
 
But why would Windows cause a freeze that always repeated at the exact same time while in the game? If it was something in Windows, you'd think it would trigger sporadically. This sounds like something that can only be fixed inside the game, actually there was an attempted fix at this here. I wouldn't waste time trying to turn Windows services off. Btw, there are lots of old games out there, iRacing has used more or less the same base for 10 years as well, so I don't think it's much of a problem.
 
@stonec and that occurs only few persons and few recovered from by rollbacking nvidia driver.. Well, dunno, just trying to help, mine is currently working fine, no stutters.. Maybe DDU?
 
You must log in or register to reply here.
Back
Top