Performance benchmarks for CAPS programs.
Run benchmarks with: caps_build . && ./benchmark
- Test: Send/receive 1M messages
- Result: 500K ops/sec
- Code: benchmarks/channel-throughput.caps
- Test: 1M state transitions
- Result: 2M transitions/sec
- Code: benchmarks/fsm-transitions.caps
- Test: Timer precision over 1 hour
- Result: <1ms drift
- Code: benchmarks/timer-accuracy.caps
- Test: Channel buffer sizes
- Result: Predictable, bounded, optimized layout
- Code: benchmarks/memory-usage.caps
CAPS vs. alternatives:
- Faster than interpreted concurrency
- Lower overhead than threads
- More predictable than async
- Optimized for high performance