Quick answer
How many CPU cores do I need on a rented dedicated server?
How many CPU cores you need on a rented dedicated server depends on how much work runs at once. Plan 6 to 8 cores for one web stack or a few game servers. Plan 16 for CI runners and busy databases, and 24 to 32 for many VMs. Lumenfer rents 6 cores / 12 threads up to 32 cores / 64 threads, from $91.65; crypto only, no KYC.
How many cores, in short?
| Cores and threads | Every CPU in the Lumenfer range runs 2 threads per core (SMT). Linux counts threads as CPUs: nproc prints 32 on a 16-core chip. |
|---|---|
| Cloud vCPU | On AWS m7i, 1 vCPU is 1 thread: m7i.8xlarge has 32 vCPUs on 16 cores. On AWS r7a, 1 vCPU is 1 whole core (AWS documentation). |
| Hosted CI size | GitHub's standard Linux runner for private repositories has 2 CPUs and 8 GB of RAM (GitHub Docs). |
| Kubernetes | kubeadm asks for 2 CPUs or more on each control-plane machine. |
| PostgreSQL defaults | 8 background worker processes and 2 parallel workers per Gather node, until you raise them. |
| Lumenfer plans | Lumen 6 $91.65, Lumen 8 $109.37, Lumen 16 $205.27, Lumen 24 $356.25, Lumen 32 $749.67, per 30 days. |
What does an extra core actually buy you?
An extra core buys more work running at the same moment, not a faster single task. Ten build jobs, ten game instances or ten busy database connections spread across ten cores. One long compile step or one game tick does not.
The serial part of a job sets the ceiling. If 10% of a task cannot run in parallel, no number of cores makes it more than 10 times faster. That rule is Amdahl's law, and it is why clock speed matters as much as core count.
Count the threads that are busy at your peak hour, add 30% headroom, and rent the plan whose thread count covers the total.
Is a thread the same as a core?
No. A second thread gives a core far less than a second core would. With simultaneous multithreading (SMT), the two threads of one core share its execution units, its caches and its memory bandwidth.
SMT pays off when threads wait on memory, disk or network, which is common in web and database work. It adds little to code that already keeps the core busy, such as video encoding or compression.
Cloud sizes are counted in vCPUs, and the meaning changes from one instance family to the next. AWS documents m7i.8xlarge as 32 vCPUs on 16 cores with 2 threads each, and r7a.4xlarge as 16 vCPUs on 16 cores with 1 thread each. Compare threads with threads and cores with cores.
How many cores do common workloads need?
Use these starting points for one machine, then confirm them against your own load. The last column names the first Lumenfer plan that fits.
| Workload | Cores to start with | What decides it | First plan that fits |
|---|---|---|---|
| One web app with its database | 6–8 | Requests per second and the number of PHP, Node or Python workers | Lumen 6 or Lumen 8 |
| Game servers, several instances | 8–16 | Main-thread speed of each instance first, then the instance count | Lumen 8 or Lumen 16 |
| CI runners | 16 | Parallel jobs × CPUs per job: 16 two-CPU jobs fill 32 threads | Lumen 16 |
| PostgreSQL primary | 8–16 | Active connections plus parallel workers per query | Lumen 8 to Lumen 16 |
| Analytics scans (ClickHouse) | 16–32 | Scan threads and memory bandwidth | Lumen 16 to Lumen 32 |
| Proxmox host with many VMs | 24–32 | Total guest vCPUs you refuse to overcommit | Lumen 24 or Lumen 32 |
| Kubernetes worker node | 16–32 | Sum of pod CPU requests plus a system reserve | Lumen 16 and up |
GitHub's private-repository runner has 2 CPUs and 8 GB, so a machine with 32 threads and 128 GB holds 16 such jobs side by side. The cost side is on is a rented dedicated server cheaper than GitHub-hosted runners.
Does single-thread speed matter more than core count?
For a large share of server work, yes. A game tick, a Node.js event loop, a linker step or one heavy SQL query runs on a single thread. There, clock speed and work per clock beat extra cores.
In the Lumenfer range, more cores come with lower clocks. AMD lists these figures for the CPU behind each plan:
| Plan | CPU | Cores / threads | Max boost (AMD) | Base clock (AMD) |
|---|---|---|---|---|
| Lumen 6 | AMD EPYC 4244P | 6 cores / 12 threads | Up to 5.1 GHz | 3.8 GHz |
| Lumen 8 | AMD Ryzen 7 9700X | 8 cores / 16 threads | Up to 5.5 GHz | 3.8 GHz |
| Lumen 16 | AMD Ryzen 9 9950X | 16 cores / 32 threads | Up to 5.7 GHz | 4.3 GHz |
| Lumen 24 | AMD EPYC 9254 | 24 cores / 48 threads | Up to 4.15 GHz | 2.9 GHz |
| Lumen 32 | AMD EPYC 9354 | 32 cores / 64 threads | Up to 3.8 GHz | 3.25 GHz |
AMD product pages, read 2026-11-01. Max boost is the top frequency of one core on a bursty single-threaded load; all cores together run lower.
The 16-core AMD Ryzen 9 9950X boosts to 5.7 GHz, 1.9 GHz above the 32-core AMD EPYC 9354: for one-thread work, the middle plan is the quick one.
AMD gives the AMD EPYC 9254 an all-core boost of 3.9 GHz, a useful figure for loads that keep every core busy. The full model-by-model comparison is on what is the best CPU for a rented dedicated server.
When are 6 or 8 cores enough?
Six or eight cores are enough when one service dominates and rarely runs more than a dozen busy threads. Typical cases are a WordPress or Laravel site with its database, a Node.js API or a reverse proxy. A few game servers or a staging copy of production fit too.
Lumen 6 pairs its AMD EPYC 4244P with 32 GB DDR5 ECC. Lumen 8 carries 64 GB DDR5 and a higher boost clock. For game hosting, the CPU choice matters more than the count: see what is the best CPU for a Minecraft server host.
When do you need 24 or 32 cores?
You need 24 or 32 cores when one machine carries many tenants or long parallel jobs. Examples: Proxmox with a dozen VMs, a Kubernetes worker, a large monorepo build, or ClickHouse scans over billions of rows.
Core count is only half of the reason. AMD EPYC 9254 and AMD EPYC 9354 sit on AMD's SP5 socket. AMD lists it with 12 DDR5-4800 channels, up to 460.8 GB/s of memory bandwidth and 128 PCIe 5.0 lanes. The AM5 chips below have 2 memory channels and 28 lanes.
On a VM host, add up the vCPUs you plan to give your guests. Keep that total under the thread count unless most guests sit idle. The cluster side is covered in how to build a Proxmox cluster with rented dedicated servers.
How do you measure the cores you use today?
Measure at your busiest hour before you pick a plan. These commands work on any Linux server or VM (mpstat and pidstat come with the sysstat package):
nproc # CPUs Linux sees (threads)
lscpu | grep -E 'Model name|Thread|Core|Socket'
mpstat -P ALL 5 12 # load per CPU for one minute
vmstat 5 12 # column r = threads waiting to run
pidstat -u 5 1 # which processes use the CPUIf the r column of vmstat stays above the CPU count, work is queuing and more cores will help. If %steal in mpstat is above zero on a cloud VM, another tenant is using your host's cores. A bare-metal server has no such neighbour: No hypervisor and no neighbours: all cores, all memory and every NVMe drive of the machine work for you alone.
Do newer cores cost more per month?
Yes, by 22% at the median of the hosts Lumenfer checked.
In Lumenfer's check of 30 October 2026, no-KYC crypto hosts priced servers with 8+ cores, 64 GB RAM and 1 TB disk at a median $291.52 monthly. With a 2022+ CPU and NVMe only, the median was $356.07.
Lumenfer prices Lumen 8 at $109.37 for 8 cores / 16 threads and 64 GB DDR5. Each plan, with its terms and cities, is listed on how much a Lumenfer bare-metal server costs.
CPU cores on a rented server: FAQ
Are 8 cores enough for a dedicated server?
Eight cores are enough for one web application with its database, a few game servers or a small CI setup. Lumen 8 gives 16 threads and 64 GB DDR5. Move to 16 cores when jobs queue in vmstat or memory runs short.
How many cores does a Minecraft server need?
One Minecraft server leans on a single main thread, so clock speed counts more than cores. Extra cores pay off when you run several servers on one machine or heavy plugins. The 16-core AMD Ryzen 9 9950X also has the highest boost clock of the range.
Are 32 cloud vCPUs the same as a 32-core server?
Usually not. On AWS m7i, 32 vCPUs are 16 cores with 2 threads each. Lumen 32 has 32 cores / 64 threads, so twice those cores. Prices side by side are on is bare metal cheaper than the cloud.
How many cores do I need for Proxmox?
Proxmox itself needs little: its documentation asks for a 64-bit CPU with Intel VT or AMD-V. Size the host by the vCPUs of your guests. A dozen mid-size VMs fit well on 24 or 32 cores.
Can I add cores to a rented server later?
Not to the same machine: a bare-metal server's cores are fixed by its CPU. You order the larger plan, move your data and let the old server lapse. Setup time: Up to 72 hours from the moment your payment is confirmed, so order before you need it.
Rent the core count your load needs
Five plans from 6 cores / 12 threads to 32 cores / 64 threads, from $91.65 per 30 days. No KYC, paid in crypto.
- Operator: DomaineGoat LLC, Panama
- Prices checked against 13 hosts, 30 October 2026
- No KYC: no ID, no card, no phone
- Built by hand within 72 h of payment
Sources (11)
- AMD EPYC 4244P: specifications — AMD (read 2026-11-01)
- AMD Ryzen 7 9700X: specifications — AMD (read 2026-11-01)
- AMD Ryzen 9 9950X: specifications — AMD (read 2026-11-01)
- AMD EPYC 9254: specifications — AMD (read 2026-11-01)
- AMD EPYC 9354: specifications — AMD (read 2026-11-01)
- Amazon EC2 instance types: general purpose specifications — Amazon Web Services (read 2026-11-01)
- Amazon EC2 instance types: memory optimized specifications — Amazon Web Services (read 2026-11-01)
- GitHub-hosted runners reference — GitHub Docs (read 2026-11-01)
- Installing kubeadm — Kubernetes (read 2026-11-01)
- PostgreSQL documentation: Resource Consumption — PostgreSQL Global Development Group (read 2026-11-01)
- Proxmox VE: System Requirements pve.proxmox.com/wiki/System_Requirements — Proxmox Server Solutions (read 2026-11-01)