Game Server Using 100% of One CPU Core: Normal or a Problem?

A game server pinning one CPU core at 100 percent while the others idle is usually normal. Almost every game runs its world simulation on a single main thread, so that thread uses one core and cannot spread to the rest. It only becomes a problem when the server falls behind its tick rate: players rubberband, Minecraft drops below 20 TPS, Rust's server FPS sinks. Some servers also run their loop flat out even when empty, which shows as 100 percent on one core with no harm done. Check the game's own performance numbers before you change anything.

Why only one core does the work

A game tick has to happen in order: read player input, move everything, run AI and physics, resolve combat, send the results. Each step depends on the one before it, and the next tick cannot start until the current one finishes. That chain is very hard to split across cores, so the main loop lives on one thread. Extra threads handle side jobs like saving, networking, chunk or terrain generation and pathfinding, but the main thread sets the pace.

That is why you see something like this in htop on Linux:

  1  [||||||||||||||||||||||||||||||||100.0%]
  2  [||                               4.0%]
  3  [|                                2.1%]
  4  [|||                              6.3%]

Or on Windows, Task Manager shows the server at 13 percent total CPU on an 8 core machine. That 13 percent is one core flat out. Total CPU percentages hide this, which is why a server can lag "at only 15 percent CPU". If you are choosing hardware, our guide to the best CPU for a game server explains why single core speed matters more than core count.

Case 1: busy waiting (harmless)

Some servers run their main loop as fast as they can instead of sleeping between ticks, or poll the network in a tight loop. They show 100 percent of one core even with nobody online. This is a design choice, not a fault, and it does not mean the server is struggling.

How to tell: the load is the same with 0 players and with 10, and the game's own performance numbers are healthy.

You can often cap it with a frame limit:

Capping saves power and leaves room for other servers on the same machine. Do not cap below the game's normal tick rate, or you will cause the lag you are trying to avoid.

Case 2: real overload (a problem)

If the core is maxed and the game is falling behind, the server cannot finish its ticks in time. Symptoms:

• Players rubberband and actions land late.
• AI freezes or teleports.
• The console warns it "can't keep up" (Minecraft prints Can't keep up! Is the server overloaded?).
• Built in stats show the drop: Minecraft TPS under 20, Rust's server FPS falling hard in the fps command output, Valheim or 7 Days to Die logs complaining about long frames.

Then the question is what the main thread is spending its time on.

Finding what is eating the core

• Minecraft: install the spark plugin or mod and run /spark profiler. It shows exactly which entities, chunks or plugins cost the most.
• Rust: compare the fps output with player count. Plugins can be timed with Oxide's o.plugins (it shows hook time) or Carbon's equivalent.
• ARK, Conan Exiles, other Unreal games: check the log for long frame warnings, and count dinos or NPCs. Large numbers of tamed creatures in one area are a classic cause.
• 7 Days to Die: big hordes, lots of active vehicles and large player bases all load the main thread. The server log prints FPS and entity counts.
• Factorio: the in game debug view (F4, show-time-usage) breaks down update time by system.

Ways to bring the load down

1. Reduce simulation work. Lower entity caps, mob spawns, view or simulation distance. In Minecraft, simulation-distance in server.properties matters more than view-distance for CPU.
2. Pre generate the world. Generating new terrain is expensive. Minecraft's Chunky plugin, for example, can pre generate a map while nobody plays.
3. Clean up builds and creatures. Decay settings, creature limits and cleaning abandoned bases all cut main thread work.
4. Trim plugins and mods. Profile, then remove or replace the worst offenders.
5. Restart on a schedule. Some servers slow down over long uptimes. See how often you should restart a game server.
6. Move to a faster core. If you have done the above and it still maxes out at peak, the core is just too slow for your player count.

On a VPS: check steal time

On shared virtual machines, your "core" can be shared with other customers. In top, the st value is steal time: CPU time the hypervisor gave to someone else. A few percent is fine. If it sits high while your server lags, you are not getting the core you think you are paying for. Our lag guide covers this and other causes.

Running several servers on one machine

Since each server leans on one core, a machine with eight cores can run several small servers side by side, as long as each main thread gets its own core. On Linux you can pin servers to cores with taskset, for example taskset -c 2 ./server to keep a server on core 2. On Windows, Task Manager's "Set affinity" does the same. Pinning helps most when servers compete at peak times.

When the easy fix is a different host

If your home machine or VPS is maxing its main core at peak and you have already trimmed entities and plugins, better hardware is the answer. A GamezServers server gives you a live console to watch performance, every setting from the game's server template in the panel, and plan changes at any time with the price prorated. Games like Rust and Minecraft are a quick switch.

FAQ

Is 100% CPU on one core bad for a game server?

Not by itself. It is normal for the main thread to use a whole core. It is only a problem if the server falls behind its tick rate and players feel lag.

Why does my server use CPU when nobody is online?

Many servers keep simulating the world or run their loop without sleeping. A frame cap like Rust's fps.limit or Arma Reforger's -maxFPS reduces idle load.

Will more cores make my server faster?

Rarely. Extra cores help background work and running more servers, but the main tick still runs on one core. A faster core helps more.

Why does Task Manager show only 12% CPU while my server lags?

On an eight core machine, one maxed core is 12.5 percent of the total. Look at per core usage instead of the total.