Palworld Server CPU Requirements: What Your PC Can Handle
Palworld recommends 4 CPU cores but no model or clock speed. Check your PC under load, avoid outdated flags and compare hosting specs.
You will need
- A Palworld server and world you can observe
- Access to the server computer's CPU monitor
- Optional secured access to the server REST API
Pocketpair recommends 4 or more CPU cores for a Palworld dedicated server, but publishes no required clock speed, processor model, or single-thread score. That gives you a baseline, not a guarantee that any particular PC will keep your world smooth. Check the machine while it runs the world your group actually plays, compare the CPU reading with the server’s own metrics, and avoid copying old launch flags as universal fixes.
Start with Pocketpair’s CPU recommendation#
The current Palworld server requirements list 4 Core + as recommended. Pocketpair does not label fewer cores an absolute startup failure, and the page does not say that four cores will handle every group or world. Treat it as the published point to compare against, not a pass/fail test for smooth play.
| Official item | What Pocketpair says | What it does not tell you |
|---|---|---|
| CPU | 4 or more cores recommended | No model, clock speed, benchmark score, or player-to-CPU formula |
| Memory | 16 GB listed; 8 GB can boot with increased out-of-memory risk | Whether your world will fit at its busiest |
| Storage | Faster SSD recommended | A CPU performance target |
| Operating system | 64-bit Windows or Linux | Whether the server will remain responsive under your own workload |
The requirements page also says its Docker Desktop deployment is deprecated from an I/O perspective because it increases the risk of save corruption or malfunction. That is a storage and deployment caveat, not a reason to buy more CPU. Keep each resource in its lane: if memory is nearly full, changing processor settings will not create memory headroom. For memory sizing by group, use the Palworld server RAM guide.
For the rest of this guide, “CPU pressure” means that CPU activity rises at the same time that the server’s own frame readings worsen during a repeatable session. A high whole-machine percentage by itself is not enough to diagnose the Palworld process. If the symptom is lag with several possible causes, use the broader server lag diagnosis before shopping for a processor.
Treat CPU models and thread counts as clues#
A core count, GHz figure, or “vCPU” label does not tell you how your particular Palworld world will perform. Pocketpair gives a core recommendation but no model or clock-speed cutoff. Hosting plans may publish a CPU model, a thread limit, a vCPU count, or only a broad label; those units describe what the provider says it allocates, not a measured server result.
When two plans show different processor names, don’t assume the larger model number wins for your game. Ask whether the listed CPU is the model used on your node, what the plan’s thread or vCPU limit means, and whether the provider has a way for you to see instance-level usage. “Six threads” and “four threads” are not a Palworld benchmark. More threads can matter to other work, but no official Palworld source sets a thread-count threshold that predicts your world’s frame rate.
The same caution applies to home computers. A newer desktop processor may have more headroom than an older server chip, but the product name alone cannot account for background apps, cooling, power limits, the game client running beside the server, or your world’s activity. Don’t replace “four cores recommended” with a made-up GHz minimum or a rule such as “one fast core is always enough.”
Check your Palworld server CPU requirements under load#
An empty server after a restart is an easy test and a poor stand-in for a busy evening. Check the PC with the server and the other programs you expect to run. If you plan to play on the same machine, include the game client; if several friends will build at active bases, test during that kind of session. A five-minute check can reveal a constraint, but it does not certify the server for every future world state.
Compare your CPU with the published baseline
On Windows, open Task Manager → Performance → CPU and note the processor model, core count, and logical processor count. On Linux, use the system monitor or a process monitor that shows CPU activity by core or thread. Compare the physical core count with Pocketpair’s 4-or-more-core recommendation. A lower count means you are below the recommendation; it does not prove that the server cannot start.
Run the server in a representative session
Load the world you intend to use and reproduce normal group activity. Include the game client, voice chat, or streaming software only if you plan to run them on that PC. Write down the player count, base activity, uptime, and whether the symptom happens to everyone or only one player.
Watch the CPU while the symptom happens
In the CPU monitor, view individual logical processors where your operating system supports it. Note whether one or more graphs stay busy during the same period that the server becomes less responsive. Also note overall host CPU use; another application can be the process consuming the time.
Repeat under similar conditions
Check again with roughly the same players and activity. If CPU activity and server frame readings move together each time, CPU may be part of the limit. If only one client stutters, or the server metrics stay similar, investigate the client and network before buying hardware.
This is a reader-run check, not a result we measured on your PC. A short quiet session may miss what happens after the group has built up its bases or when the host runs another game. If you cannot reproduce the problem, keep the result as “not tested under load” instead of treating a clean startup as proof.
Read server FPS without inventing a target#
Palworld’s REST metrics endpoint returns serverfps as an integer and serverframetime in milliseconds. The same response includes current players, maximum players, uptime, base camps, and in-game days. These values help you compare two observations from your own server; they do not reveal CPU use by themselves.
| Reading | What it represents | How to use it |
|---|---|---|
serverfps |
Server FPS, returned as an integer | Compare your own reading during similar activity |
serverframetime |
Server frame time in milliseconds | See whether it changes when the same group action slows down |
currentplayernum / basecampnum |
Current players / base camps | Record context for the CPU and frame readings |
Pocketpair’s example shows serverfps: 57. That is sample response data, not a minimum, maximum, or “healthy” score. The docs publish no universal cutoff. Take several readings during the symptom and compare them with a quieter period that has similar uptime and settings. If your REST API is not already enabled and protected, skip the API part rather than exposing its basic-auth credentials to the public internet; the operating-system monitor is enough for the CPU observation.
This guide uses metrics to help you check a CPU hypothesis, not to replace a full lag diagnosis. For causes and follow-up checks, use the Palworld server lag guide. The wider list of hardware, storage, and network baselines is in the dedicated server requirements guide.
Leave the old multithread flags alone until you test them#
Older Palworld guides often recommend adding -useperfthreads -NoAsyncLoadingThread -UseMultithreadForDS to every server. Pocketpair’s current server argument guide says that in v1.0 and later, leaving those parameters unset may improve performance. It does not publish an optimal flag combination or promise a specific improvement.
Do not add that bundle just because an old checklist calls it a universal optimization. If you want to compare startup arguments, change one thing at a time on a backed-up test world and compare the same activity before and after. Keep other launch options intact. Pocketpair says -NumberOfWorkerThreadsServer=X must be used with the multithread performance arguments; that dependency is not a recommendation to enable either one on a current server.
Ask what a host’s CPU number means#
Before choosing a managed plan, ask for the CPU model used on the server, the thread or vCPU limit for your plan, whether that limit is shared, and what usage information you can see after setup. Ask separately about memory: a plan can have a clear CPU label and still be the wrong memory size for your world. The broader dedicated server requirements guide covers memory, storage, operating system, and network checks.
Compare like with like. A model “up to” a named processor may not mean each node runs that exact model; a vCPU count does not disclose the host’s CPU generation; and a thread limit does not show whether those threads are dedicated physical cores. Unless two providers measure the same Palworld build, settings, world, activity, and time period, their product labels cannot establish which plan has better in-game performance.
For example, Game Hosting’s Palworld plan cards identify an AMD Ryzen 9 9900X and a 4-thread CPU limit per plan. That makes the published allocation visible while you compare providers. It is not a claim that the server is faster than another host or that a given world will meet a particular FPS target.
Rent a server when CPU upkeep no longer fits your group#
If your PC stays responsive during the sessions you expect and you are happy to leave it on, self-hosting remains a reasonable free route. Don’t upgrade the processor just because the server uses more CPU than it does while idle; observe a real group session first. CPU is only one part of the setup, alongside memory, storage, upload capacity, and whether the host machine is also running the game.
If the home computer is the constraint, compare the cost and time of replacing it with moving the world to a managed server. We recommend Game Hosting for Palworld when you would rather move the always-on machine and server settings out of your household chores. The plan cards show the CPU model and thread limit, while the integrated game panel handles settings and scheduled restarts; three free backup slots give you copies to restore, download, or lock. Those features address management and visibility, not measured performance. Check the plan’s memory recommendation for your group and keep a copy of the save before moving it.
The useful answer to “what CPU does Palworld need?” is therefore a starting baseline and a test, not a part number: compare with Pocketpair’s 4-core recommendation, observe your own server during a representative session, and ask a provider what its published allocation actually means.
Quick answers
How many CPU cores does a Palworld dedicated server use?
Pocketpair recommends 4 or more CPU cores. It does not publish a processor model, clock speed, or performance cutoff.
Is a Palworld server CPU or RAM heavy?
The official requirements do not rank CPU against memory. Observe both under the same real workload before deciding what is limiting your server.
Can an old PC run a Palworld server?
It may start below the published recommendation, but startup alone does not show how it will handle your world. Test it with the group and activity you expect.