The Current DCS World Performance

[FierceLV]

Hi,

Since I was the creator of one of the topics addressing the issue of the "Frame Time Spikes" from way back, the topic contained walls of text, and eventually got solved-and-closed, and the Ultimate Solution was the Park Control Software that disabled the CPU Core Parking and disabled the CPU Core Frequency Scaling (the Solution has improved the entire performance of the PC (I guess it was the Windows 10 thing)), so I feel obliged to give an opinion regarding the Current DCS World Performance.

Flying the latest version of the DCS World, I can surely conclude that the Current DCS World Performance is Good.

However, there is that one moment, but that moment is like a general-rule in the way some games perform, and that moment can be described as follows:

1. So the game contains lots of data to be processed by the system.

2. You launch the game first time after booting up the system.

3. The system loads the essential game data (but not all the data), and the mission starts.

4. So you fly the aircraft of your choice.

5. After some G-maneuvers, change of scenery, adjustments to the afterburner (the sound data here is crucial) you get that One-Time-Lag.

One-Time-Lag - less-than-a-second lag (a hitch), but still noticeable.

One-Time because it appears ONCE during random G-maneuvers, random change of scenery, random adjustments to the afterburner.

The One-Time-Lag does not appear during the second time the Player experiences the same G-maneuvers, same change of scenery, same adjustments to the afterburner (because the extra game data was already recalled and processed, and is stored somewhere in the system).

Personally, the One-Time-Lag that I sometimes encounter (mostly full fidelity aircraft) is not game breaking.

The One-Time-Lag is a problem in cases where it is occurring often while the Player's actions are recalling the processing of additional game data.

Important to note that I fly the simple missions, not the all-out-war scenarios.

If only there was something that could one time process all the game data that can be triggered by a mission, save that data somewhere, and recall it in the future without the One-Time-Lag.

The One-Time-Lag is bad because it pulls the Player from the immersion of the flying, even if it happens rarely and once.

But then again, the One-Time-Lag is a normal phenomena for some types of games, and can not be considered a performance problem.

Personally, I think that the current DCS World Performance is Good.

 

 

 

 

[scommander2]

On 8/6/2025 at 12:13 PM, FierceLV said:

3. The system loads the essential game data (but not all the data), and the mission starts.

There is two cents according to my experience so far:

After DCS login, wait for 5 to 10 mins (I also launch task manager for monitoring all core activities till they are kind of stable that tells by graphs).

Then, proceed step 4. 

[dmatsch]

16 hours ago, scommander2 said:

 

After DCS login, wait for 5 to 10 mins (I also launch task manager for monitoring all core activities till they are kind of stable that tells by graphs).

 

what?!  please tell us you're being sarcastic here.

[scommander2]

Well, as @FierceLV mentioned, DCS is loading stuffs (either signed or unsigned mods) in the background (unless disable background loading) after logging in.

Due to the "better" hardware, it may not need 5 to 10 mins.

 

[FierceLV]

Hi.

What is described in the first post is how some games generally perform on PC.

The idea is that at some point not long into the game a specific combination of events (not always the same) triggers a millisecond freeze. During that freeze the hardware processes that specific combination of events, after the hardware has done it's job, experiencing the same combination of events (and similar combinations) does not trigger the millisecond freeze.

After closing a game, and then launching it again, that same combination of events does not trigger the millisecond freeze (I guess the result of processing during the first launch of the game after a system startup is stored in some form on the hardware (a prefetch file, a temp file, cache or something like that)).

However, after turning off the PC for the night, and then launching that same game again, during first minutes of play time a random combination of event triggers a millisecond freeze again. And this is not unique to the DCS World, many other games perform like this. I don't know about other Players, but it seems that it is how some games perform on PC by default. Maybe this is only a case with Intel + NVidia hardware combination.

It's all not scientifically accurate information, but it seems that if there was a way to get that Cached-For-The-DCS-World file being present before launching the game after night could circumvent the millisecond freeze.

And if that theoretical Cached-For-The-DCS-World file could take gigabytes of space (even tens), that would be no problem in the case where the DCS World has zero freezes at all stages of the Game (because it is totally worth it), because the DCS World is a delight that allows itself to have that kind of privilege (the extra gigabytes).

Also it is important to note that in the scenario of a serious performance load on the hardware due to average PC, the Event-Processed-For-The-First-Time microfreeze is occurring with a constant interval, and that leads to a conclusion that:

1. The result of game being processed is stored somewhere on the system in some form (maybe on the RAM?).

2. At some point during game play the result of game being processed exceeds the allocated limit of resource in question, and the Event-Processed-For-The-First-Time microfreeze occurs again. In case with the RAM, the Event-Processed-For-The-First-Time microfreeze is able to occur in an empty mission. The Avionics Scripts exceed the 32 gigabytes of RAM?

3. If there was a way to save the result of a game being processed for the next play sessions, and to seamlessly recall the data during gameplay, that would in theory get rid of the "every time the PC needs to process some data for the first time after rebooting of the system.

This phenomena (Event-Processed-For-The-First-Time) is occurring in the scenario where the hardware is able to run the game flawlessly, and where there is no performance load on the system.

The DCS World completely without stutters, freezes, lags is a very good cause. And I am really happy to see that things like "Issue with high CPU performance load even in main menu" are getting addressed.

**FOLLOW UP INFORMATION**

It seems that the Event-Processed-For-The-First-Time microfreeze is CPU related. There appears to be some kind of hold-up in the CPU processing some data for the first time.

Managed to video-capture the Event-Processed-For-The-First-Time mircofreeze during the first-launch-of-the-day of the DCS World. 

WARNING: VIDEO STARTS WITH ABRUPT LOUD SOUND

Notice how the Event-Processed-For-The-First-Time mircofreeze occurs during the camera movement at 00:02. The prominent microfreeze occurs at 01:33.

Also, it appears that the Event-Processed-For-The-First-Time mircofreeze occurs during the GPU-CPU Bound swaps.

Additional two screenshots depict the nature of the Event-Processed-For-The-First-Time mircofreeze: same aerobatic maneuver, first time there are microfreezes, the second time there are no microfreezes.

Adding the DCS World log and the DxDiag just in case.

I can provide more information if required.

I understand that the phenomena described can be an individual case, so this is in no way a claim.

 

WARNING: VIDEO STARTS WITH ABRUPT LOUD SOUND

 

DCS World Debrief 11.09.2025.log DCS World Log 11.09.2025.log DxDiag 11.09.2025.txt

Edited 10 hours ago by FierceLV