Open the 3D battle viewer — run the full engine here ↗
The engine’s wind.
Running here.
Recoil’s C++ wind simulation, random-number generator and deterministic math now run in WebAssembly. Compare ten minutes of wind updates against the native build.
2,000 frames. Zero checksum differences.
The full Recoil headless engine ran the real BAR synctest on Jade Empress. Every recorded frame checksum matched the unmodified Linux ARM64 build.
Verified in Node.js. This is a saved test result; the browser demo below runs wind only.
Same seed. Same simulation.
18,000 updates · 30 per simulated second · wind range 5–25
Wind direction
ENGINE X / Z PLANETen minutes of changing wind
ACTUAL WASM OUTPUT- WASM state checksum
- Native comparison
- Wind vector (X, Z)
What this result establishes
This compiles Recoil’s Wind.cpp, SpringMath.cpp, float3.cpp and SyncChecker.cpp, plus its streflop library. Wind-generator unit notifications are excluded at the test boundary.
The comparison checks all recorded wind values and RNG state, not just the displayed checksum. The checksum uses Recoil’s checksum implementation over this component’s output; it is not a full-game sync hash. Passing this test does not establish multiplayer compatibility.
Compute time covers this small wind workload in one run. It is not a benchmark of BAR gameplay.