What Hardware Does Your Modded Minecraft Server Need? (CPU, RAM, Storage)

You found a ThinkCentre with an i7-7700 and 16 GB of RAM on the local classifieds for a hundred euros. It would make a decent Minecraft server, right? For vanilla with a few friends, absolutely. But you want to run 50 mods. Maybe 100. Maybe an entire kitchen-sink modpack. And now you are staring at spec sheets wondering if that four-core chip from 2017 will burst into flames during server startup.

The answer, like most things in Minecraft server administration, is "it depends." But this guide will give you the actual decision framework — not vague advice like "get more RAM."

We already have a guide on how much RAM your server needs, but it focuses on vanilla through light modpacks. Modded servers have different bottlenecks. RAM matters, but CPU single-thread performance and storage I/O often hit their limits first.

Why Modded Servers Are Different

Vanilla Minecraft is already surprisingly demanding for what looks like a game about blocks. The server has to simulate every loaded chunk, process entity AI, handle player interactions, and do it all 20 times per second.

Mods amplify every one of those workloads:

• More blocks to process: Tech mods add machines, pipes, cables, and automation that tick every server cycle. A single Create contraption can involve hundreds of moving parts.
• More entities: Mobs from mods like Alex's Mobs or Ice and Fire. Item entities from sorting systems. Spawned vehicles, trains, or airships.
• More data per chunk: Modded blocks store more NBT data than vanilla blocks. Chunk loading and saving becomes heavier.
• Startup overhead: The server has to load and initialize every mod at boot. Some modpacks take 5+ minutes to start on slow hardware.

Vanilla Minecraft is a single-threaded application. Modded Minecraft is also single-threaded — but now that single thread has to do significantly more work.

CPU: Clock Speed Beats Core Count

Here is the uncomfortable truth about Minecraft server performance: your expensive 16-core Ryzen Threadripper is not twice as good as an 8-core chip. It is not even 10% better. The main server loop runs on one thread. More cores do not help the main loop run faster.

What helps is:

1. Higher clock speed — A CPU running at 4.5 GHz processes instructions faster than one running at 3.0 GHz.
2. Better IPC (Instructions Per Clock) — Modern architectures do more work per cycle. A 4.0 GHz chip from 2024 outperforms a 4.0 GHz chip from 2017.
3. Large caches — More L3 cache means fewer expensive trips to RAM.

The i7-7700 Question

The i7-7700 (and its unlocked 7700K sibling) were excellent gaming chips in 2017. They boost to 4.2–4.5 GHz and have decent IPC for their era. For a modded server with 50–60 mods and around 10 players, they are adequate. Not great, not terrible — adequate.

Here is what the 7700 handles well:

• Light to medium modpacks (under 100 mods)
• Small player counts (10 or fewer simultaneous)
• Modpacks without heavy automation (less Create, less Applied Energistics, less factory building)

Here is where it struggles:

• Kitchen-sink packs with 300+ mods
• Modpacks with heavy worldgen (Terralith + many biome mods)
• Player bases with large automation setups
• Servers with more than 15 active players

If your modpack is something like RLCraft, Valhelsia 3, or ATM6 Lite, the 7700 can work. If you are running All The Mods 9 at full scale, you will feel the limits.

What to Look For Instead

If you are buying hardware specifically for a modded server, prioritize:

You do not need more than 6–8 cores. The extra cores sit idle. Spend that budget on clock speed and generation instead.

Good options in the current market:

• Intel Core i5-13400/14400 (budget)
• Intel Core i5-13600K/14600K (midrange)
• AMD Ryzen 5 7600/7600X (midrange)
• Intel Core i7-13700K/14700K (overkill for most servers)
• AMD Ryzen 9 7950X3D / 9950X3D (top of the stack — see below)

An i5-13400 at 4.6 GHz turbo will outperform an i7-7700 in single-threaded workloads despite being a "lesser" i5. Architecture generation matters.

Why X3D Chips Are Different

AMD's X3D variants — the Ryzen 9 7950X3D (Zen 4) and the newer Ryzen 9 9950X3D (Zen 5) — deserve a separate mention because the workload they were designed for is essentially the workload a Minecraft server puts on a CPU. They stack an extra slab of cache directly on top of the die, giving them 128 MB of total L3 cache versus the 32–64 MB you get on most other chips in the same tier.

Why does that matter for a server? The main tick loop reads from a relatively small "hot set" every tick — the loaded chunks, the entity list, the tick queue, the active block-entity list. On most CPUs that data spills out of L3 and into RAM, which costs hundreds of nanoseconds per access and adds up fast at 20 ticks per second. On an X3D chip, almost the entire hot set fits in L3, so the same tick simply finishes faster. Heavy modpacks under load see meaningfully better TPS-under-pressure on an X3D part than on a same-clock non-X3D one.

Combine that with 5.7 GHz single-thread boost and 16 cores for parallel work (mod-loader startup, autosaves, view-distance worker threads, multiple servers on one box) and they are about as well-suited to modded Minecraft as currently-shipping silicon gets.

This is, not incidentally, what we run on the Swelis Hosting fleet — current nodes are built around the Ryzen 9 7950X3D, with a 9950X3D rollout in progress.

RAM: Baseline Allocations by Modpack Tier

RAM requirements for modded servers scale with:

1. Number of mods loaded — Each mod adds classes to memory
2. World complexity — More chunks, more entities, more data
3. Player count — Each player loads more chunks and runs more entity AI
4. Modpack type — Tech mods with inventories use more RAM than content mods

RAM by Modpack Tier

These are server allocations, not total system RAM. Your server machine needs additional RAM for the operating system, Java overhead outside the heap, and any other services running.

Common RAM Mistakes

Over-allocating: Giving Java 16 GB when you only need 8 GB causes longer garbage collection pauses. The JVM has to scan more memory during GC cycles, which creates lag spikes. Allocate what you need, not the maximum available.

Under-allocating: Starving the JVM causes constant garbage collection. You will see TPS drops as the server repeatedly pauses to free memory. If your server log shows "Can't keep up!" messages alongside high memory usage, you need more RAM.

Ignoring player scaling: Your empty server might run fine with 6 GB. Add 20 players building separate bases and you could need 10 GB. Each player loads their own view distance worth of chunks.

For more on RAM allocation and the "Can't keep up" message, see our complete lag troubleshooting guide.

Storage: Why NVMe Changes Everything

Storage is the sleeper bottleneck for modded servers. Most discussions focus on CPU and RAM, but I/O performance determines:

• World save time — How long it takes to write chunks to disk
• Chunk load speed — How fast players can move through the world
• Startup time — Reading thousands of mod JARs and world data

HDD vs SSD vs NVMe

The random read speed is what matters for Minecraft. The server constantly reads and writes small chunks of data scattered across the drive. HDDs are catastrophically bad at random I/O because the read head has to physically move.

A modded world save can be 5–10 GB of region files. On an HDD, autosaves cause visible lag spikes. On NVMe, saves complete before players notice.

When HDD Is Acceptable

If you are budget-constrained, a 7200 RPM HDD can work for:

• Light modpacks with few players
• Servers with long view distances disabled
• Worlds that are played infrequently

But if you are buying hardware specifically for a modded server, put an SSD — ideally NVMe — in the budget. A 500 GB NVMe drive costs about the same as a mediocre dinner out and will save you hours of troubleshooting "lag spikes every 30 seconds."

Forge vs Fabric: Resource Patterns

The two major mod loaders have different performance characteristics. Understanding them helps you pick hardware and troubleshoot issues.

Forge

Forge is the older, more established mod loader with the largest mod library. Resource-wise:

• Startup: Forge servers take longer to start. The loader does extensive class transformation at boot. Kitchen-sink packs can take 5–10 minutes on slow hardware.
• Memory: Forge tends to use more RAM at steady state. The core loader itself has overhead.
• CPU: Similar single-threaded behavior to Fabric once running.
• Memory leaks: Forge packs are more prone to gradual memory creep over long uptimes. Scheduled restarts (every 12–24 hours) are standard practice.

Fabric

Fabric is newer and designed to be lightweight:

• Startup: Significantly faster startup times. The loader does less at boot.
• Memory: Lower baseline RAM usage. The loader itself is minimal.
• CPU: Same single-threaded behavior, but less overhead per tick.
• Optimization mods: The Fabric ecosystem has more performance mods (Lithium, Sodium server-side equivalents, Starlight).

Which to Choose

The choice is usually driven by which mods you want, not performance:

• Forge: More content mods, especially tech/magic mods. Create, Mekanism, Botania, most major modpacks.
• Fabric: More optimization mods, popular for performance-focused packs. Strong Vanilla+ scene.

If your modpack supports both loaders, Fabric will generally use less RAM and start faster. But if the mods you want are Forge-only, there is no real decision to make.

For a detailed comparison of server software options, see our Paper, Purpur, Fabric, or Forge guide.

Real-World Spec Recommendations

Putting it together: here are hardware recommendations by use case. These assume a dedicated machine running Linux (lower overhead than Windows).

Small Friends Server (5–10 Players, Light/Medium Modpack)

Cost: Used hardware in the 100–150 EUR range works here.

Community Server (15–25 Players, Medium/Heavy Modpack)

Cost: 200–400 EUR for used/refurbished hardware, or hosted.

Large Server (30+ Players, Kitchen-Sink Modpack)

At this scale, self-hosting at home becomes impractical. Network stability, power, and maintenance overhead push most server owners toward managed hosting — for reference, our nodes are built around the Ryzen 9 7950X3D today (with a 9950X3D rollout in progress) on top of NVMe storage, specifically because of the cache-hungry workload modded Minecraft creates.

When Hardware Isn't the Problem

Before you upgrade hardware, verify that hardware is actually the bottleneck:

1. Check TPS: Use /forge tps or the Spark profiler. If TPS is below 20, something is causing slowdown.
2. Check Spark profiles: Spark shows exactly where server time is going. Often it is a specific mod or player's base, not CPU limits.
3. Check server configs: Many modpacks ship with view-distance 12 or higher. Reducing to 8 or 10 dramatically cuts workload.
4. Check for memory leaks: If RAM usage climbs over time until the server lags, restart more frequently and report the leak to mod authors.

See our server performance optimization guide for specific tuning steps.

Quick Reference

Related Guides

• Minecraft Server RAM Guide — Baseline RAM recommendations for vanilla through light modpacks.
• Paper, Purpur, Fabric, or Forge? — A practical comparison of server software to help you pick the right loader.
• How to Fix Minecraft Server Lag — Complete troubleshooting for TPS drops, memory issues, and network lag.
• Optimize Server Performance — Configuration tweaks and tuning steps for better performance.
• One-Click Modpack Install — How to install CurseForge and Modrinth modpacks without manual file management.
• Mob Spawning and Entity Lag — Why entities cause lag and how to control them.