This GitHub organization hosts the Auto-Tuning Framework (ATF) and related tools. ATF is a research-driven framework for automatically tuning performance-critical program parameters.
The repositories in this organization provide language bindings, libraries, and infrastructure to integrate auto-tuning into real-world systems.
ATF focuses on:
- automatic exploration of performance parameter spaces,
- support for constrained tuning parameters,
- integration into existing applications and toolchains,
- portability across platforms and execution environments.
The projects here are primarily research-oriented, but aim to be usable and extensible in practice.
Python interface and tooling for ATF (see also this paper). Provides bindings, examples, and utilities for driving the auto-tuning process from Python-based workflows.
Repository: pyATF
C++ interface for integrating ATF into native applications (see also this paper). Designed as a lightweight, header-only interface suitable for performance-critical code.
Repository: cppATF
The organization also hosts documentation and web resources related to ATF, including project overviews and usage instructions.
See: atf-tuner.github.io
- Select the repository relevant to your use case (Python or C++).
- Follow the README and documentation in the respective repository.
- For questions, issues, or extensions, please use GitHub Issues in the appropriate repository.
Contributions are welcome. Please open an issue to discuss larger changes before submitting a pull request. Repository-specific contribution guidelines apply where provided.
ATF-Tuner is developed in an academic research context and is intended to support experimentation and reproducible performance studies in auto-tuning and performance optimization.
