Performance
Benchmarks vs Paper and Purpur
TPS Under Player Load
20 TPS = perfect
Startup Time
Seconds to first player-ready state, from an empty dir with only the .jar/.exe and EULA. Lower is better.
Memory Usage @ 100 Players
Heap + off-heap in MB. Lower is better.
Chunk Loading Speed
Chunks generated per second. Higher is better.
CPU Usage @ 100 Players
Average across all cores under load. Lower is better.
Max Stable Player Capacity
Maximum concurrent players before TPS drops below 18. Higher is better.
These benchmarks are very unfair, as Paper and Purpur implement the entire Minecraft server logic in Java, while Temper MC is no-where near feature complete and currently doesn't implement as many features of the Minecraft server as Paper and Purpur, so these numbers should be taken with a grain of salt. However, they do demonstrate the potential of our approach and the significant performance improvements that can be achieved by rethinking how Minecraft servers are built.
Some of the numbers don't look right? Come have a chat in our Discord, it's more than likely that we just haven't remembered to update them in a while. Keep in mind that these benchmarks were taken on a high end machine and your mileage may vary based on your hardware and the specific workload, but we hope they give you a good idea of the performance benefits that Temper MC can offer.