High ping/Bad latency with multiple game instances on the same network

Summary

Playing with 2 game instances each on their own device. The player that joins second always has a high latency (for us around 140ms Europe).
The other player on the same Network has around 24ms.

No matter what I did I couldn’t get a lower latency on my second connection.
I noticed that even if I manually select Server “Australia” with a latency of more than 350 it seems like it uses the server that was used for the first connection (had a latency of 140ms ingame).

To make sure it is not the devices fault we tried it the other way around.
Disconnected both devices → connected device 2 first → then connected device 1
This resulted in the same bug but for the other device. So I am very sure it is not some weird bug on my devices.

I did not add any pictures or videos since all I could show is a high latency :slight_smile:


How to replicate?

Join a Realm with game instance 1 on device 1 → Join any Realm with game instance 2 on device 2


Workarounds

Currently there aren’t any.
Using a VPN works fine. Both instances run with around 30ms.

I tried

  • Creating new characters
  • Creating a new realm
  • Joining different realms with different characters
  • Changing mutliple settings in my game settings

Player.log


Player-prev.log


DxDiag


Datastore.zip


Additional Info

I am not sure if this has something to do with my IP Address. I will try joining the realm with an active VPN but for now this is all that I got


Has it been reported before?

Doesn’t seem like it was. Atleast I couldn’t find any topic or reddit thread about it.

1 Like

Are you using the same steam account simultaneously for testing, I have to ask, both logs have same steam id.

I believe that the save operation is not async, even though its udp which means that the host or provider is being forced to save first and since the session logic rolls in chunks, could be causing a race to finish session logic atomically. Secondary clients might be using some smart logic to mess up packet delivery [looking at you asus routers]. If this is something deeper, you might need to change router to forward ports, and determine if that Natted ip is a problem for your router and its packet priority. I cant imagine the net code put in wouldn’t have been testing for something more basic like this since moon would have had to run and use their servers in a pretty similar default architecture.

You might need to clear dns cache as well, and make sure there aren’t any static ip conflicts. Are both machines normally on same network, or is one like a friend who brought the laptop over? finally tracert/curl/ping their backend api endpoints that in the log files.

I use 2 different accounts.
Those are only logs from the first account.
I can provide the logs for account 2 later if that helps.

Will try to delete the dns cache for both devices and see if that helps.
The devices are 2 Windows PCs one from my wife and one from me. They are desktop pcs so I can’t really imagine dns cache being a problem. But it is worth a shot.

Im just throwing some shotgun solutions out there for ya from what I have information wise.

1 Like

Are the specs for both machines similar? Do they both have a nvme or solid state drive? I noticed that you have a slightly older ryzen and only 16gb ram as listed which might push the multi threading and virtualization into the error zone. The game seems to be both logging and unloading, with timestamps that indicate at piloints 300ms of locked save operations. Since quantum is constantly syncing save state from both clients, the last client to synchronize the game state could be the common speed. Even with multi threading, the order of operations is probably locking.

Total: 283.753000 ms (FindLiveObjects: 67.402000 ms CreateObjectMapping: 26.908400 ms MarkObjects: 171.557700 ms DeleteObjects: 17.884200 ms)

I have a decompiler I can run to dig deeper this afternoon to test/debugmultiple instances in virtualization to simulate if need be, but I dont think I would beat the moon team in effort by myself :stuck_out_tongue:

I still assume network, or bad player state and session sync. There are extra debug command that can be run at startup as well, not sure if moon publishes those for steam.

Player.txt Log Analysis - Network & Latency Issues

Session Duration: 72 minutes (19:17 - 20:30)


CRITICAL LATENCY INDICATORS

Issue Category Pattern Count Frequency Impact on Latency
Multiplayer State Sync Player:2 reference resolution failure 4 0.06/min Cannot resolve other player’s state - forces constant resync/corrections
Session Authority Spectating mode errors 2 0.03/min Session demoted to observer - all inputs require server validation
Quantum Frame Sync NullReferenceException in EntityView 16 0.22/min Frame update failures force rollbacks and re-simulation
Save Blocking Save cycle wait-before-disconnect 4 0.06/min Blocks network operations during save (0.5-1s freeze)

SAVE SYSTEM AUTHORITY PATTERN

Metric Value Implication
Total Save Cycles 79 Normal operation
Designated Provider (Authority) 25 (32%) Minority authority role
Not Designated Provider 54 (68%) Majority follower role
Blocking Events 4 (5%) Occasional network blocking

Analysis: This client is NOT the authoritative host for most operations, indicating secondary/follower status.


MULTIPLAYER STATE TRACKING

Player Reference Failure Count Status
Player:0 (self) 2 Occasional self-reference issues
Player:2 (other) 4 Cannot track other player consistently

Analysis: Failures to resolve Player:2 indicate this client cannot properly track another player’s character state in the multiplayer session.


SESSION DISCONNECTION EVENTS

Event Type Count Significance
Normal disconnects (DisconnectByClientLogic) 4 User-initiated or routine
Spectator mode errors 2 Forced session demotion

Analysis: Spectator mode errors indicate the game detected this session as secondary/invalid and demoted it to observer status.


ROOT CAUSE ANALYSIS

Primary Latency Sources (Ranked by Likelihood):

  1. Multiplayer State Desynchronization

    • Player:2 reference failures indicate inability to track other players
    • Quantum deterministic networking requires perfect state sync
    • Desync forces rollback → visible latency
  2. Secondary Session Status

    • Only designated provider 32% of the time
    • Spectator mode errors confirm session demotion
    • Client treated as secondary by backend
  3. Quantum Frame Update Failures

    • 16 NullReference exceptions in EntityView.Tick
    • Failures occur in Quantum frame update loop
    • Failed updates require rollback and re-simulation

LATENCY MECHANISM

When this client connects second to the network:

Primary Client:  Local Input → Prediction → Display
Secondary Client: Local Input → Server → Validation → Correction → Rollback → Display

The secondary client experiences additional latency from:

  • Server validation round-trip
  • State correction processing
  • Rollback/re-simulation overhead
  • Player reference resolution delays
1 Like

Thank you for your effort! :100:
Device 2 has a better CPU (I think) but runs on a GTX 1660.
Device 2 has no nvme (just an SSD)
(I will add the logs of device 2 later this day)

Even though the other pc is slightly under the minimum requirement it runs fine (around 30ms) if the device 1 is not connected to any realm first.

Whatever device is connected first runs completely fine. It is only the second device that experiences this latency issue.

Let me know if I can do anything to help.

These are the log files for device 2.

Not sure if this helps but last session this was the device that joined last.

Edit: And since I didn’t send them in the beginning, there are report files of device 1

Hi Pubble,

I see you have some troubles. I’ve read this topic and have basically nothing to add on the technical side.

You could still try a VPN for the person with the most severe latence.

There is also a person in Discord who is a strong advocate for tunneling software, and reported good gains from it. Find me in Discord if you want some more advice on that.

Other than that, the right people at Moon Studios are aware this is an issue. Let’s hope they can deliver improvements!

Silas

1 Like