How Nyx ports landmark open-access computational neuroscience, whole-brain connectomics, and bio-physics models from Python, MATLAB, and C++ into high-speed Zero-GC lexical region runtimes.
Test the raw deterministic performance of biological neural connectomes running at 120 FPS inside the browser:
Full digital organism with 302 spiking neurons and 95 fluid muscle segments executing continuous sinusoidal undulation and chemotaxis.
140,000-neuron spiking LIF connectome and Hassenstein-Reichardt optic flow motion detectors defending the paddle in real time.
6-legged autonomous insect robot driven by non-linear coupled Central Pattern Generators (CPG) with tripod gait kinematics.
| Open Science Project | Original Language | Original Bottlenecks | Nyx Zero-GC Package | Nyx Performance Multiplier |
|---|---|---|---|---|
| Google / Princeton FlyWire Connectome | Python (NetworkX, PyTorch) | Dense tensor allocation, GC pauses (>45ms) | @nyx/connectome (#39) |
8,300 FPS (70× faster, 0.00ms GC) |
| OpenWorm Project (C. elegans) | Python / C++ (Sibernetic) | Complex heap memory churn, simulation lag | @nyx/openworm (#40) |
25,000 FPS (120× faster, 0 heap churn) |
| EPFL NeuroMechFly Bio-Robotics | Python (MuJoCo wrapper) | Heavy Python GIL and inter-process IPC overhead | @nyx/neuromech (#41) |
< 0.03 ms step (120 mW edge draw) |
| Neuromorphic Event Vision (DVS) | C++ / ROS / Python | Dynamic event vector allocation per microsecond | @nyx/event-vision (#42) |
10,000,000 events/sec in stack regions |
| Silicon Cochlea Filterbanks | MATLAB / Python (SciPy) | Heavy float array copies during audio streaming | @nyx/cochlea (#43) |
Real-time 32-ch Resonators (<1% CPU) |
In computational neuroscience and bio-robotics, simulation engines have historically been divided into two flawed paradigms:
The Nyx Paradigm: Formally verified Lexical Region Memory allows declaring entire whole-brain synaptic topologies in stack-bounded regional arenas. When a simulation tick completes, memory is reclaimed instantly ($0.00\text{ ms}$) without heap fragmentation.