Cimba: High-Performance Discrete Event Simulation Library in C with Multithreading and Coroutines

Summary

Cimba is a high-performance discrete event simulation library written in C and assembly that uses POSIX pthreads for parallelized replications and stackful asymmetric coroutines (fibers) for concurrent simulated processes within each thread. The library demonstrates significant performance advantages over Python's SimPy, running 40-50 times faster in multithreaded scenarios and 25% faster on a single core than SimPy using all 64 cores. The implementation focuses on efficient event processing with benchmarks showing 20 million events per second on a single core.