BLAZING FAST
Bytecode VM • Native Performance • Smart Cache
Compile pipeline
- Four stages: lexer, parser, IR lowering, bytecode emission
- Portable output: compiled to a
.niaobcmodule
Smart cache
- Compile once: the .niaobc module is reused on every later run
- Skip the front end: no re-lexing, re-parsing, or IR rebuild
- Straight to run: warm starts load bytecode and execute
Bytecode VM
- Register/stack design: tuned for tight dispatch loops
- Native-class speed: runs the same bytecode across platforms
LOW RAM
Lean Runtime • Efficient by Design • Mark-Compact GC
Mark-compact GC
- Mark: live objects are traced from the roots
- Compact: survivors are packed to remove fragmentation
- Automatic: no manual allocation or free calls
Lean runtime
- Small footprint: the runtime leaves RAM for your app
- Runs anywhere: fits modest hardware without losing throughput
- Predictable: steady memory use, no surprise spikes
Benchmarks
Benchmarks publishing as Niao's test suite matures
We won't publish numbers until they're reproducible. Track progress →
