Crate feagi_npu_burst_engine 

FEAGI Burst Engine

High-performance neural burst processing engine.

Performance Targets

• 30Hz burst frequency with 1.2M neuron firings
• 50-100x faster than Python implementation
• SIMD-optimized for modern CPUs
• Cache-friendly data structures

Architecture

• Pure Rust, no Python overhead
• Rayon for multi-threading
• Zero-copy data access where possible
• Minimal allocations in hot paths

Re-exports

pub use burst_loop_runner::SensoryIntake;

pub use dynamic_npu::DynamicNPU;

pub use tracing_mutex::TracingMutex;

pub use dynamic_npu::DynamicNPUGeneric;

pub use parameter_update_queue::ParameterUpdate;

pub use parameter_update_queue::ParameterUpdateQueue;

pub use backend::*;

pub use burst_loop_runner::*;

pub use fire_ledger::*;

pub use fire_structures::*;

pub use fq_sampler::*;

pub use neural_dynamics::*;

pub use npu::*;

pub use sensory::*;

pub use synaptic_propagation::*;

pub use viz_shm_writer::*;

Modules

async_burst_loop

Async burst loop implementation using FeagiAsyncRuntime.

backend

Compute Backend Abstraction

burst_loop_runner

Pure Rust Burst Loop Runner

dynamic_npu

Runtime type dispatch for NPU

fire_ledger

FireLedger - Dense, burst-aligned firing history for memory + STDP.

fire_structures

Fire Queue data structures for NPU processing

fq_sampler

Fire Queue Sampler for Motor and Visualization Output

motor_shm_writer

Motor SHM Writer

neural_dynamics

Neural Dynamics (Phase 2)

npu

Complete Rust NPU

parameter_update_queue

Asynchronous parameter update queue for cortical areas.

sensory

Rust Sensory Injection System

synaptic_propagation

Synaptic Propagation Engine

tracing_mutex

Tracing Mutex wrapper that automatically logs all lock acquisitions and releases This provides visibility into lock contention across all code paths

viz_shm_writer

Visualization SHM Writer

Structs

BurstEngineStats

Burst engine performance statistics

Constants

VERSION

Crate version from Cargo.toml