Repolode
Submit
ChunelFeng

ChunelFeng/CGraph

A dependency-free C++11 DAG framework for building and running parallel task graphs, with a Python binding (pycgraph).

View on GitHubWebsite
Stars
2,317
Forks
391
Open issues
13
Last commit
5 days ago

What is it?

CGraph is a cross-platform C++ framework for describing and executing directed acyclic graphs of tasks. You subclass a node class, implement its run() method, and declare dependencies between nodes. The scheduler then runs dependent nodes in order and runs independent nodes concurrently, with support for pause, resume and timeouts.

Beyond plain dependencies, the framework provides groups for conditional branching, loops and concurrent execution, as well as parameter passing, events, messages and aspects for cross-cutting logic. It is written against the C++11 standard library, builds on macOS, Linux, Windows and Android, and is also available as the Python package pycgraph.

The project is aimed at developers who need to orchestrate multi-step processing pipelines in native code, such as algorithm workflows or inference pre- and post-processing, without adding external dependencies.

Key features

  • 01Runs dependent nodes in order and independent nodes concurrently
  • 02Supports conditional branches, loops and grouped execution
  • 03Provides pause, resume and per-node timeout control
  • 04Passes parameters, messages and events between nodes
  • 05Has no third-party dependencies and builds on C++11 with Python bindings

Use cases

  • Orchestrating multi-stage algorithm or data-processing pipelines in C++ applications
  • Structuring AI inference pre- and post-processing steps as a dependency graph
  • Embedding a task-graph scheduler in embedded or Android projects without external libraries

Quick start

  1. 1

    Install the Python binding from PyPI

    pip install pycgraph
  2. 2

    Save the README's Python demo as a script and run it with Python

Things to know

  • The C++ version has no package-manager install shown in the README and must be built from source using the separate COMPILE.md guide.
  • The main README is written in Chinese, with a separate English README linked from it.

In collections

Open-source AI agent frameworks & apps

Topics

#dag#task-scheduling#parallel-computing#workflow#pipeline#cpp#python#ai#ai-agents#graph

FAQ

What is CGraph?
A dependency-free C++11 DAG framework for building and running parallel task graphs, with a Python binding (pycgraph). CGraph is a cross-platform C++ framework for describing and executing directed acyclic graphs of tasks. You subclass a node class, implement its run() method, and declare dependencies between nodes. The scheduler then runs dependent nodes in order and runs independent nodes concurrently, with support for pause, resume and timeouts.
Is CGraph free to use? What license does it have?
CGraph is open source under the MIT license. Check the license terms before using it in commercial products.
Is CGraph still maintained?
Actively maintained. The last commit to CGraph was on Nov 3, 2026.
What can I use CGraph for?
Orchestrating multi-stage algorithm or data-processing pipelines in C++ applications Structuring AI inference pre- and post-processing steps as a dependency graph Embedding a task-graph scheduler in embedded or Android projects without external libraries
How do I install CGraph?
Install the Python binding from PyPI: pip install pycgraph Save the README's Python demo as a script and run it with Python
Where can I get CGraph?
The source code and installation instructions are on GitHub: https://github.com/ChunelFeng/CGraph

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects