Framework API#
Core infrastructure libraries providing foundational capabilities for high-performance, real-time applications. Built with modern C++20.
Components#
Logging (Log) - Structured, thread-safe logging with minimal runtime overhead
Memory (Memory) - Buffer allocation with device-aware memory pools for zero-copy GPU/CPU transfers
Net (Net) - Low-latency network I/O using DPDK and DOCA
Pipeline (Pipeline) - Library for composing GPU-accelerated processing pipelines from reusable modules
Task (Task) - Lightweight concurrent task execution with priority scheduling and worker thread pool management
Tensor (Tensor) - Multi-dimensional array structures with GPU/CPU memory views
TensorRT (TensorRT) - Integration layer for NVIDIA TensorRT engines
Utils (Utils) - Common utilities and helper functions