Resolved Questions about the Prediction Loop on Client and Server

I have a couple question regarding the prediction loop. The loop is used by the client and server. I am struggling a bit to understand how the loop works. If there are some resources with detailed explanation I appreciate some links.

I use some terms in the questions so I want to explain the beforehand what I do understand for these terms. If there are more adequate terms please feel free to state them here.

Terms:

• Player Loop - This is the execution of all systems which are requiring an uppdate. The player loop is split in three major groups: "Initialization", "Update" and "PreLateUpdate". Systems may still get their "OnUpdate" called multiple times during one distinctively execution of a player loop.
• Player Loop Execution - One distinctively execution of a Player Loop including all systems.
• Prediction Loop - As far as I am aware of this are all systems running inside the PredictedSimulationSystemGroup.

My questions are:

1a
While the client runs the Prediction Loop many times for adjusting the component data incrementally based on the server data, is it correct that the server runs the Prediction Loop always only one time during a Player Loop Execution?

1b
In case the Player Loop may also run multiple times on the server during a Player Loop execution, what are the reasons for this case?

2a
Is the Prediction Loop (respectively PredictedSimulationSystemGroup) on the client guaranteed to be executed exactly once for every NetworkTime.ServerTick having the NetworkTime.IsFirstPredictionTick == true? This would mean that each predicted server tick like 23, 24, 25, 26, etc. gets always exact one execution time where NetworkTime.IsFirstPredictionTick == true. Otherwise the value would be false when the specified tick is processed within the Prediction Loop.

2b
How about the server regarding the question 2a. Are the server systems within PredictedSimulationSystemGroup are guaranteed to be executed exactly once for every NetworkTime.ServerTick having the NetworkTime.IsFirstPredictionTick == true?

2c
Is the behavior of the NetworkTime.IsFinalPredictionTick the same way as NetworkTime.IsFirstPredictionTick regarding question 2a and 2b?

3
Before fully running the Prediction Loop on the client, all values like the fields of LocalTransform are containing non-predicted/interpolated values. After fully running the Prediction Loop for the remaining Player Loop Execution certain data is predicted then. Is there some possibility to get the predicted values (like LocalTransform or other component data) of the last Player Loop Execution _before_ the Prediction Loop runs? For example within the GhostInputSystemGroup.

 

Thank you for taking the time answering my questions. Helped me understanding the prediction loop. There is still one thing I don't undersand fully yet.

Especially you stated that the IsFirstPredictionTick can be used for predicted spawning or similar one-time code execution in the prediction loop. I made a test using the following system:

Spoiler: System

Code (CSharp):

1.   [BurstCompile]
2.   [WorldSystemFilter(WorldSystemFilterFlags.ClientSimulation)]
3.   [UpdateInGroup(typeof(PredictedFixedStepSimulationSystemGroup))]
4.   public partial struct AnotherTestSystem : ISystem
5.   {
6.     private LoggerHandle _logger;
7.  
8.     public void OnCreate(ref SystemState state)
9.     {
10.       this._logger = new LoggerConfig()
11.         .MinimumLevel.Debug()
12.         .WriteTo.UnityEditorConsole()
13.         .CreateLogger().Handle;
14.  
15.       state.RequireForUpdate<NetworkTime>();
16.     }
17.  
18.     [BurstCompile]
19.     public void OnUpdate(ref SystemState state)
20.     {
21.       var networkTime = SystemAPI.GetSingleton<NetworkTime>();
22.  
23.       if (!networkTime.ServerTick.IsValid)
24.       {
25.         Log.To(this._logger).Info("CLIENT: Prediction tick skipped", networkTime.ServerTick.TickIndexForValidTick);
26.         return;
27.       }
28.    
29.       if (networkTime.IsFirstPredictionTick)
30.       {
31.         Log.To(this._logger).Info("CLIENT: First prediction tick on {TickValue}", networkTime.ServerTick.TickIndexForValidTick);
32.       }
33.       if (networkTime.IsFinalPredictionTick)
34.       {
35.         Log.To(this._logger).Info("CLIENT: Final prediction tick on {TickValue}", networkTime.ServerTick.TickIndexForValidTick);
36.       }
37.       if (networkTime.IsFinalFullPredictionTick)
38.       {
39.         Log.To(this._logger).Info("CLIENT: Final full prediction tick on {TickValue}", networkTime.ServerTick.TickIndexForValidTick);
40.       }
41.       if (!networkTime.IsFirstPredictionTick && !networkTime.IsFinalPredictionTick && !networkTime.IsFinalFullPredictionTick)
42.       {
43.         Log.To(this._logger).Info("CLIENT: Casual prediction tick on {TickValue}", networkTime.ServerTick.TickIndexForValidTick);
44.       }
45.  
46.     }
47.   }
48. }

This is an excerpt from the logs of this system printed in the Unity Editor console:

I don't know if it is of any importance here, but I used the Network Simulator for the Client with an RTT Delay of 500 ms. Also the system was burst compiled for the test.

You can see that certain predicted server ticks like 242, 244 and 246 are not getting the "First prediction tick" log message. But when something important happens in these ticks which requires a one-time logic to be executed within the prediction loop, how this can be achieved then when these ticks don't get the IsFirstPredictionTick == true?

 

Am I correctly understand that PredictedSimulationSystemGroup systems will never be executed for InterpolatedGhosts?
And another question related Simulate tag. Does this tag is set and enabled for interpolated ghosts? The only case when it is not set is when predicted ghost does not need to be predicted on client?