C2ME built for 26.3 on 15 September, the day Wilderness Bound shipped. That is fast for a mod that rewrites how the game threads chunk work, and it is worth looking at what the mod claims, because the numbers are larger than most performance mods are willing to print.
What C2ME does
Concurrent Chunk Management Engine is a Fabric mod aimed at three things: chunk generation, chunk I/O and chunk loading. It does not touch rendering, entity ticking or lighting. It takes work the game does on one thread and spreads it across the cores your machine actually has.
That focus is why it pairs rather than competes with the rest of the stack. Lithium handles general server tick optimisation, ScalableLux handles lighting, and C2ME handles chunks. The project explicitly recommends running all three, which is unusual honesty for a mod page.
It is also why a client-side player may not notice it at all. Server side is required, client side is optional. If your world is already generated and you are not exploring, C2ME has nothing to do.
The benchmark numbers
The project publishes benchmarks run in August 2026 on 26.2 with Terratonic and Terralith installed, measured in chunks per second. Two curves, one for high core counts and one for a more realistic range.
- 48-core system: C2ME scales to roughly 1,200 chunks per second at about 45 threads, reaching around 1,400 at 96. Paper stops scaling around 20 threads and tops out near 170. Vanilla reaches about 42.
- 1 to 16 threads: C2ME scales linearly to 458 chunks per second at 16 threads. Paper’s scaling breaks down from around 7 threads and reaches 142. Vanilla manages 41.
- Single thread: 29.1 versus vanilla’s 13.2, so roughly 120 percent faster before any parallelism is involved.
The headline figures work out to about 34 times vanilla and 8.2 times Paper on the 48-core machine. Those are generation throughput numbers on a heavily modded worldgen setup, not framerate numbers, and they should not be read as a general speedup.
Reading those numbers honestly
Three caveats matter.
First, the benchmark uses Terratonic and Terralith. Custom worldgen is far more expensive than vanilla generation, which means there is more work available to parallelise. On a vanilla world the gap narrows.
Second, the single-thread result is the one most home servers should look at. A four-core VPS is not going to see 34x. It might see the 1-to-16-thread curve, which is still a large improvement, but it is a different number.
Third, chunks per second only matters while chunks are being made. Once your players have explored the area they play in, chunk generation stops being the bottleneck and something else becomes it.
The alpha question
The 26.3 build is version 0.4.2-alpha.0.85, marked alpha, and it has picked up over 26,000 downloads since release. C2ME has been publishing alpha-tagged builds as its normal release channel for a long time, so the label means less here than it would elsewhere.
It is still a mod that changes threading behaviour on a server, which is the category where bugs are hardest to reproduce. The sensible position for a live server is a backup before updating and a day of watching logs after, the same discipline that applies to any version migration.
The changelog for the 26.3 build is a single automated line: build 26.3. No behaviour changes, just a port. That is the good kind of changelog for a mod you are about to trust with your world.
Where it fits on a server
C2ME is at its most valuable in two situations. A freshly created world where players are exploring outward constantly, and a pre-generation run where you are deliberately asking the server to make as many chunks as fast as it can.
For the second case it stacks with Chunky, which also reached 26.3. Pre-generating with C2ME installed is the single biggest time saving most admins can get out of it, and it is a one-off cost that permanently removes generation lag from normal play – the argument for pre-generation and a world border in the first place.
If your server’s problem is tick lag with a stable, fully explored map, C2ME is not the mod you need. That is a tuning problem elsewhere in the stack.
Loader availability
The main project is Fabric only. NeoForge users need the separate C2ME NeoForge project, which is maintained independently and does not always update at the same time. There is also C2ME-OCL, an experimental build with hardware-accelerated worldgen, which is not something to put on a server anyone else is playing on.
Licence is MIT, so redistribution in a modpack is unrestricted.
Bottom line
If you run a Fabric server where players are still exploring, C2ME on 26.3 is worth the update, ideally alongside Lithium and a pre-generation pass. Treat the 34x figure as the ceiling on a workstation-class machine with heavy worldgen mods, and expect something between two and five times on ordinary hardware.
Our server performance stack write-up covers the order to install the rest in.
Images: Moderner Beta project gallery.