High-Performance Engine

Build and run deterministic simulations with a graph-scheduled, fixed-step solver designed for interactive iteration and real-time use.

Outcomes

• Deterministic execution with a reproducible step schedule
• Real-time responsiveness at small fixed steps (e.g., 1–10 ms)
• Clear execution order across blocks and subsystems
• Diagnostics for algebraic loops and scheduling issues
• Profiling signals to identify hotspots and tune performance

Key capabilities

• Graph scheduler: topologically orders blocks and resolves execution dependencies
• Fixed-step time base: stable integration at a user-defined step size and duration
• Event handling: edge events and discrete updates within the fixed-step loop
• Deterministic I/O: well-defined read/write phases per step to avoid race conditions
• Loop detection: algebraic loop warnings with guidance to insert delays/state
• Profiling hooks: per-block timing and per-step aggregates to guide optimization

How it works

1. Choose a step size and duration

   Typical: 1e-3–1e-2 s for interactive work; increase for faster previews

2. Configure scheduling

   Inspect the generated execution order; break cycles by inserting state where needed

3. Validate model timing

   Ensure sources/sinks have compatible sample times within the fixed step

4. Run and profile

   Enable per-block timing to spot long-running operations

5. Tune and iterate

   Adjust step size, simplify hot paths, and split subsystems where appropriate