bandit648

Can one of you guys post a screenshot of what you think is pagefile usage? Task Manager does NOT show pagefile usage. If you're looking at the "Commited" memory in task manager that is not your pagefile usage. The only way to see pagefile usage is by using Performance Monitor and enabling the pagefile usage counter.

What are you using to monitor your pagefile usage that makes you think it's using 30 GB?

Correct, the pagefile isn't used unless you run out of available system memory. In my testing with page file enabled, DCS doesn't use it unless I force it to use all available memory. You can do that by setting "preloadRadius = 0" in the options.lua. I don't recommend doing that for anything other than testing though. Even with hundreds of objects loaded in a mission it still doesn't use all 32 GB that I have.

C:\Program Files\Eagle Dynamics\DCS World OpenBeta\Doc\DCS User Manual EN.pdf Page 19: PRELOAD RADIUS. Radius of the preloaded terrain around user position in meters. A large setting can decrease hard disk swapping but it requires more system RAM.

According to the manual, the Preload radius setting is supposed to load terrain objects into memory with more memory being used the larger the radius. In my testing I'm not seeing any difference in memory usage with Preload Radius setting at 100 vs 150000. Does this setting actually do anything or is currently disabled? Logging system RAM usage in RTSS shows little to no difference between Preload min and max settings while floating around the PG Map.

Are you guys running driver 418.81? I'm installing that tonight to give it a try. The last couple drivers were reported to be worse for Pascal cards and better for Turing. This latest one is supposed to be better for Pascal so I'm going to try it.

Unfortunately I don't think many people use dynamic weather due to the difficulty in configuring it so I don't expect this to get much support. It probably won't get fixed... :(

I opened a support ticket with Natural Point several weeks ago detailing the issue. They responded right away and said they were eager to help and find a solution.... and that was last I heard from them.

Actually, I think whatever sticks in memory and reduces stuttering with this, survives restarting the game. If you restart your PC it will be gone of course. I did forget to mention, I tested preload radius and I see little to no difference between the lowest setting of 100 and highest setting of 150000. The manual states this setting is supposed to preload objects into system RAM so you should see increase in RAM usage with a higher preload radius. This does NOT happen. I'm currently running preload radius set to 100 and have my pagefile off in Windows. With 32GB system / 11 GB GPU this is getting the best results for me currently. If you force preload radius to zero in options.lua, then it uses all available memory. In my case it would go up to 32 GB of system RAM and 11 GB of GPU memory. If it was empty mission it would run fine like that but with a lot of AI units the game would crash unless I enable the page file. Maybe I just need 128 GB of system RAM and 32 GB of GPU memory and the game will run great!? :lol::lol::megalol:

We need as many replies to this thread as possible to get attention. Otherwise I suspect this won't be fixed anytime soon by ED. If you have this issue, please reply to the thread with details! Thanks

A nice thing I discovered to use the F11 trick and not muck up your static weather- Create an empty mission where you cycle through the F11 views on the desired map. Then save a replay track of that mission. The next time you start DCS you can just run the replay track prior to loading your real mission. Then you get the affect of having things loaded into memory to reduce stuttering but you don't lose your static weather clouds immediately. I also found that setting preloadRadius = 0 in the options.lua will allow DCS to use all of your system RAM and GPU memory.. this is interesting for testing. It eliminates stuttering on an empty mission. As soon as you load a mission with lots of AI units it maxes out and will crash if you have your pagefile disabled. With pagefile enabled, it will continue to run but doesn't work as well as F11 cycling + pagefile disabled in my experience.

I'm seeing water flickering when using dynamic weather under heavy cloud cover. The issue doesn't occur with static weather clouds. The flicker usually starts a few minutes after starting a flight and is only visible from inside the cockpit. If you flip to external view there is no flickering. I've tested this with a variety of aircraft and they all do it. The issue is present in both open beta and stable release. Is anyone else seeing this? I want to use dynamic weather but I can't stand it with this flickering. I'm running 1440p resolution with G-Sync, no VR. I've tested it with a variety of DCS settings and system config and nothing seems to get rid of it. Reducing water detail level down to medium or low helps but it still flickers where the water is reflecting the sun. Here's a video I made of it... https://youtu.be/Xgye29k3IFU

It loads objects and textures into physical memory. I'm going to post a new thread shortly with testing results on this. I found some new tricks to avoid the disappearing clouds after cycling F11.

Even if you have 32 GB DCS doesn't use it efficiently. If I fire up a mission and don't do anything my RAM usage sits at around 12 GB. Try cycling through all of the airbases on whatever map you're on with the F11 view. That will load tons of stuff into memory and your RAM usage will go up significantly. This reduces stuttering. See my thread in this sub-forum "Static weather causes stuttering". I proved it with testing. F11 actually works. Why the game doesn't do this automatically I don't know. Preload radius doesn't appear to do anything with physical memory. I believe it loads into virtual memory (pagefile) ?? Either way you still get stuttering with it. Cycling F11 view actually works though.

:doh: One thing I had forgot to try apparently. Honestly I kinda thought of it as another DCS performance myth but this one turns out to actually work. If anyone doesn't believe it proof is here. I think this really highlights the continued issues with DCS memory management.

After days of testing and developing a mild obsession I finally found a workaround for this issue. It turns out the problem is related to the dreaded DCS memory management and seems to get worse with heavy cloud levels set in static weather settings. In my testing I tried everything from changing DCS settings (shadows, view distance, MSAA level, etc.) to 9 different Nvidia drivers going back to Nov. 2017 to page file vs no page file. Nothing seemed to make a difference in the stuttering. Then I remember reading on the DCS forums once that hitting F11 a bunch of times helps with stuttering... With preload radius set to max, I cycled through F11 views for all 27 air bases and then watched the RAM perf counter in RTTS go up to nearly 19 GB. Then I started my test which is just a camera view floating above the PG map, 400 knots @ 20,000 ft. for 7 minutes with cloud level 8. Here are the results.. nearly a perfect 60 fps all the way though. Without doing the F11 cycle I get dips of fps into the 30's as noted in my previous graphs. Now here's the interesting part. After restarting DCS I still get good results in my test scenario even without doing the F11 cycle. Notice the RAM usage is half of what it is in the first test. Whatever the F11 cycling does sticks through a restart of the game. I suspect it goes away after restarting your PC though.

Yes, I totally understand and agree with this. All you have to do read through some of the posts in this thread and see that the majority of them are user created issues. However, when someone raises an issue that is reproducible under certain circumstances and provides evidence of the issue then I think that support should take the time to look into that or at least acknowledge it. In my case I found that static weather clouds set to level 5 - 8 causes stuttering. This is easily reproducible and I've described exactly how I reproduced and provided fps logging showing the dips. I would think this would be classified different than the usual "I'm getting low fps" threads we normally see on here. I hope so! It would nice if they were a little more transparent in regard to updates and fixes for performance.

It's on the map and visible in the satellite textures, but it's just outside the detailed area.

The 10th tactical airbase at Konarak/Chahbahar would make a nice addition to this map. Mirage F1's and F-4D's are based there and would likely be the first Iranian aircraft a US Carrier group would encounter upon approaching the SOH. https://www.scramble.nl/orbats/iran/airforce Hope this is being considered!

Something I noticed, ED seems to be much more responsive to issues regarding modules and other aspects of the game but when it comes to performance issues they are mostly ignored. If you look at this sub-forum there are little to no issues acknowledged as being a bug. Even in the case of static weather clouds causing stuttering, which is easily reproducible in both open beta and stable release. I opened a ticket on that issue as well and it has no response. So, should we expect that ED will not look into any reported performance issues and not report them? I do understand that performance issues can have many different causes but sometimes it's the game engine itself that needs to be corrected.

I wish it was as simple as editing the dynamic weather lua file. If you configure dynamic weather you'll notice the cloud base altitude is the same every time. It's set to a default level of a rather high 8000 ft. There must be a static/default value for this located somewhere in the config files but I've been unable to find it. Cirrus cloud levels are set in the cirrus.fx file but I can't find a similar value for the dynamic weather clouds. Unfortunately it doesn't appear there is any way to set the cloud level or thickness the way you can for static weather clouds.

It appears with dynamic weather systems that the clouds base altitude is set to some static value of around 8000 ft. Does anyone know if there's any way to change this through the config files (.lua, .fx etc.). I've looked everywhere for where this value is set and can't seem to find anything.

I suspected no issue with overcast so thanks for confirming. For now I'll be using dynamic weather. I just learned that CombatFlite can also generate dynamic weather conditions and it's super easy to use.. great work-around until ED fixes this. https://www.combatflite.com/ Does anyone know if there's a way to control the cloud base level with dynamic weather? The default seems to be around 8000 ft and I haven't found any way of changing that yet.

Adding some more test results. The issue is reproducible in the "Stable Release" version 2.5.3.24984 also so I suspect this has existed for some time. Here is my test flight in F/A-18C over SOH map. Static weather clouds set to level 8. Flight time is 7 minutes.

I tested this as well and didn't get any stuttering until I added static weather clouds at level 5 to 8. I didn't test full overcast. So, the problem appears to be with the clouds. For some reason the same clouds added through dynamic weather run perfectly fine.