Measuring…
The verdict for this game appears once the first full second of packets is in.
Comparing with other tests to this server…
Measuring…
The verdict for this game appears once the first full second of packets is in.
Comparing with other tests to this server…
Are you sure you want to delete all results?
| Date & Time | Duration | Avg Ping | Avg Jitter | Packet Loss | Actions |
|---|
Trolls warping through your base? Every player blaming a different person for the lag?
Valheim's networking works in a way that surprises people who expect a normal client-server game. The world is hosted by one machine - either a dedicated server or one of the players - but the simulation of individual creatures and objects is distributed across whoever happens to be nearest to them. That means the answer to 'who is causing the lag' is genuinely different depending on where in the world you are standing, and it is why a single player with a poor connection can make a raid unplayable for everyone.
What this test sends: 20 packets per second of 300 bytes, matching Valheim rather than a generic ping. estimated
Iron Gate publishes no tick rate; the world is peer- or server-hosted with a low update rate.
Run the test above for at least 60 seconds. Here is what each number means for this game specifically.
The world itself is static and already on your disk, so terrain and buildings never lag. Everything that moves is being simulated by some machine and relayed to yours. When delivery is uneven, that split becomes visible: a perfectly solid world with greylings sliding around in it. That picture is a network symptom, not a performance one.
Combat here is short, decisive and full of one-off events - a parry, a stagger, a troll's swing connecting. Those are discrete messages with nothing to interpolate from. A lost packet does not blur them, it removes them, and the first you know about it is being thrown across your own base by an attack you never saw start.
If your group plays on a dedicated server, your ping is to that machine and it is stable. If one of you hosts, your ping is to their living room, and it changes whenever the host does. Under 80 ms to the host keeps combat honest; the consistency matters more than the number.
What you see in game, and which number on this page explains it.
| What you see in Valheim | Look at | What it means |
|---|---|---|
| Creatures slide, terrain is perfectly solid | jitter | The static world is local; only simulated objects come over the network. |
| Hit by an attack with no wind-up | loss | The packet carrying that discrete event never arrived. |
| Rhythmic hitch for everyone at once | server | The host writing the world to disk. Not a network event. |
| Lag only near one particular player | jitter | That player is the machine simulating the objects around them. |
Most multiplayer games have exactly one authority. A server decides where everything is, clients display a prediction of it, and every player's experience is a function of their own path to that one machine. Valheim spreads the job out instead: the host owns the world, but responsibility for simulating any particular creature or object sits with whichever player is nearest to it, and it hands over as people move around.
That design is what lets a modest home machine host a world containing a great deal of active content - the work is shared rather than concentrated. The cost is that the quality of your experience is not determined by your own connection alone. If the player standing next to a troll has a poor line, that troll behaves badly for everyone, and their connection has become part of yours for as long as they are the nearest one.
This explains a pattern that groups find genuinely confusing: lag that follows a location or a person rather than a session. Everything is fine in the meadows and terrible in the swamp, or fine until a particular friend logs in. Neither observation makes sense in a normal client-server game, and both are the expected behaviour here.
It also changes what a fix looks like. Measuring your own line is still necessary - a bad connection makes you a bad neighbour as well as giving you a bad time - but it is not sufficient, and the group's overall experience is closer to the worst connection in it than the average. For a world that a group intends to keep, moving it onto a dedicated server and getting the regular attendees onto cables does more than any individual tuning.
If the chart above shows packet loss or high jitter, work through these in order.
The Symptom: The world only plays well when one specific person is on.
An in-game host is doing two jobs at once: playing the game and serving it to everyone else. A dedicated server - rented or on a spare machine - separates them, gives the world a stable address, and means nobody has to be online for the others to play. For a group that plays regularly it is the single biggest improvement available.
The Symptom: Everyone lags, but only in one person's world.
The host's connection is the world's connection. Their wireless retries become everyone's rubber-banding, and no amount of tuning on the other four machines compensates for it. If you host, a cable is not a personal upgrade - it is infrastructure for the group.
The Symptom: A brief hitch at regular intervals for everybody.
The host writes the world to disk periodically, and on a large world with a lot of built structures that write is noticeable. It appears at regular intervals rather than randomly, it affects everyone at once, and it is unrelated to your connection. If your hitches have a rhythm, this is what they are.
The Symptom: Lag arrives when the host's household gets busy.
Hosting means uploading world state to every other player continuously. A single large upload elsewhere in the house pushes that traffic into a queue and the whole world stutters for everyone. Router QoS on the host's connection is worth more than any change the other players can make.
Because the simulation of creatures and objects is distributed across the players near them. A player with a poor connection who is closest to a troll is the machine simulating that troll, and everyone else sees the result of their line rather than their own.
For any world you intend to keep, yes. It removes the requirement that a specific person be online, it gives the world a stable address and latency, and it stops one player's household from being the group's bottleneck.
Usually the periodic world save on the host, which grows with the size of the world and the number of structures in it. It is rhythmic rather than random, and it will not appear on this chart because it is not a network event.
Loss and jitter cause more of them than raw ping does. Combat is built on discrete events - parries, staggers, swings - and a lost packet removes one entirely instead of delaying it. That is what being hit by an attack you never saw actually is.