CPU cores decide how much work your server can do at the same moment. More cores help when many things happen at once, such as many visitors or several programs. They do not always make a single task faster. This guide explains what a vCPU is, how your workload uses cores, and how to estimate what you need.
Before you start
List the programs that will run on the server and whether they are busy all the time or only now and then. A server that waits for visitors most of the day needs far less CPU than one that encodes video or compiles code non-stop.
What a vCPU is
A VPS plan lists vCPUs, or virtual CPU cores. Each vCPU is a share of a physical processor core (or a hardware thread) on the host server, assigned to your virtual machine by the hypervisor.
To your operating system, a vCPU looks like a normal core. Linux reports it with nproc, and Windows shows it in Task Manager. Two vCPUs means your server can run two threads of work at the same instant.
Because vCPUs run on shared physical hardware, short bursts are fine, but a workload that keeps every core at 100% all day is better served by a larger plan.
Single-thread vs multi-thread work
Single-threaded
Some programs do their main work on one thread. Adding cores does not make them faster; the speed of each core matters more. Examples:
- Many game server main loops, including Minecraft Java's main tick.
- Some trading terminals and indicators that compute on one thread.
- A single long-running script.
Multi-threaded or many processes
Other workloads spread across cores naturally:
- Web servers handling many visitors at once, each request on its own worker.
- Databases answering many queries in parallel.
- Several bots, containers or trading terminals running side by side.
- Builds, compression and video encoding.
For these, more cores means more work done at the same time.
Rough guidance by workload
These are starting points to adjust after you measure.
- A small website, a VPN or a few bots: 1–2 vCPUs is usually enough.
- WordPress or a web app with steady traffic: 2–4 vCPUs, more if you skip caching or run heavy plugins.
- Several sites, Docker stacks or a control panel: 4 vCPUs or more.
- Trading on Windows: 2 vCPUs for one or two terminals; add more as you add terminals or heavy expert advisors.
- Game servers: depends on the game. Fast cores matter for single-threaded engines; extra cores help with plugins, mods and background tasks.
- Windows desktop for remote work: 2 vCPUs for light use, more if you run many browser tabs or heavier apps.
Tip: RAM often runs out before CPU does. Size memory first, then check CPU. See how much RAM a VPS needs.
Check it works
Watch real usage over a normal day.
On Linux
nproc
uptimenproc shows the core count. uptime ends with three load averages for the last 1, 5 and 15 minutes. As a rough rule, a load average that stays above the number of cores means work is waiting for CPU time.
For a live view, run top and press 1 to see each core. Press q to quit.
On Windows
Open Task Manager → Performance → CPU. Watch the graph during busy periods. Short spikes are normal. Long stretches near 100% are the signal.
Troubleshooting
- One core at 100%, others idle: a single-threaded program is the bottleneck. More cores will not help it; optimise the program or choose a plan with faster cores if offered.
- All cores busy and the site is slow: add caching first, then upgrade if load stays high.
- High
stvalue intopon Linux: that is steal time, CPU the hypervisor gave to other work. If it stays high, contact support with the readings. - Windows Update or antivirus scans slow everything: schedule them for quiet hours.
Next steps
- Learn what to watch: monitor Linux server resources.
- On Windows: monitor CPU, RAM and disk on Windows.