February 3, 2026
Benchmarks, brawls, and bragging rights
Show HN: C discrete event SIM w stackful coroutines runs 45x faster than SimPy
Dev drops '45x faster' bomb; HN erupts into C vs Python brawl with a Mojo cameo
TLDR: A new C library claims about 45x faster simulations than a popular Python tool, grabbing attention fast. Hacker News split between “it’s just C vs Python,” technical safety fears, and a surprise Mojo benchmark, turning a speed boast into a lively language showdown.
Hacker News lit up after a developer unveiled Cimba, a C-based engine for running “discrete event” simulations (think: modeling lines at a store, computer networks, factories). The headline claim: up to 45x faster than SimPy, a popular Python library, even beating SimPy’s 64-core run with one C core. Cue the cheers, side-eyes, and popcorn gifs. One excited voice loved the “stackful coroutines” approach (a way to pause/resume tasks cleanly), but the loudest clapback was pure buzzkill: isn’t this just C vs Python speed? A veteran chimed in that they were “surprised it wasn’t 2–3x more,” which poured gas on the fire.
Then the thread split into camps. The tinkerers asked how the magic works (“what’s doing the context switch?”), while the worrywarts warned that DIY coroutine stacks can be brittle. Meanwhile, a Mojo fan rolled in with their own mini-benchmark and a casual “vibe coding” flex, posting numbers and a link, turning it into a three-way faceoff: C vs Python vs Mojo. There were calls for “benchmarks or it didn’t happen,” questions about whether contributions are welcome, and debates over safety vs speed. The project touts heavy testing and lots of built-in checks, but the crowd is clearly split between speed-thrilled and speed-skeptical—and that’s why everyone’s watching.
Key Points
- •Cimba is a multithreaded discrete event simulation library in C/assembly using POSIX pthreads and stackful coroutines.
- •Benchmarks show Cimba runs 40–50× faster than SimPy, including 45× faster on an M/M/1 queue with all CPU cores.
- •On a single core, Cimba processes 20M events/sec, outperforming SimPy on 64 cores (16M events/sec) and reducing run time by 97.8% vs SimPy using all cores.
- •Cimba offers resources, queues, timeouts, condition variables, fast RNGs (ziggurat), logging, and extensive assertions with unit tests.
- •Platform support is stable on x86-64 for Linux and Windows, with Apple Silicon and ARM support planned.