[TOC]
CoreFlow is a modern, high-performance graph execution runtime designed for AI, ML, and computer vision applications. Our platform combines:
- π No-code Graph Builder: Visual, intuitive UI to build pipelines with automatic optimization
- π€ Cross-platform Support: Seamless deployment across ARM or x86 with unified APIs.
- β‘ High Performance: Optimized determinstic execution for real-time processing across CPU, GPU, and specialized hardware
- π οΈ Developer Friendly: Simple C++ API with comprehensive error handling
- π§ Extensible: Easy to add custom kernels and processing behaviors
CoreFlow uses a graph-based execution model where:
- Graphs represent execution modeled as DAG (Directed Acyclic Graph)
- Nodes represent processing operations
- Edges define data flow between operations
- Kernels define a runnable operation or algorithm
- Targets define a collection of suppported operations
- ONNX Runtime: Deploy models using Open Neural Network Exchange
- TensorFlow Lite: Leverage Google's on-device AI framework
- Executorch: Leverage Meta's on-device AI framework
- Model Management: Easy model loading and switching
- Memory Management: Automatic memory optimization
- Hardware Acceleration: CPU, GPU, and specialized hardware support
- Streaming and Pipelining Support: Automatic graph runtime optimization
- Error Handling: Comprehensive error reporting and recovery
This project is of confidental and contains proprietary material. See the LICENSE file for more details.
- Email: support@runedge.ai
Ready to build the future with CoreFlow?
Start with our samples and discover the power of CoreFlow's execution model.
| Previous | Next |
|---|---|
| Getting Started |