Canonical Wago documentation
Benchmark index
Crawlable Wago benchmark definitions, caveats, raw structured data, and planned many-instance measurements.
Published measurements
- Whole-process startup latency: process spawn to exit, across several runtimes.
- Wago versus wazero: decode, validate, compile, instantiate, Go allocation, and warm execution rows split by architecture.
- ARM64 benchmark interpretation and many-instance plan.
All visible comparison rows are also available without JavaScript in llms-full.txt and as JSON in project.json.
Interpretation rules
- Compare runtimes only within one architecture and workload.
- Do not compare absolute values across machines.
- “Go heap bytes allocated” is allocation traffic measured by Go benchmarks, not total resident memory.
- Guest linear memory, native code mappings, virtual reservations, RSS, and PSS are separate quantities.