Prerequisites:
- Intro to High Performance Computing (HPC)
- The World is at your Command….line (or knowledge of basic Linux/Unix-like commands)
Overview:
Transitioning scientific workloads from personal local environments (laptops/workstations) to High-Performance Computing (HPC) platforms presents unique architectural challenges and a distinct learning curve. Proper environment configuration, dependency isolation, and toolchain selection are critical for establishing reproducible and efficient research workflows on HPC clusters.
This hands-on workshop equips researchers and developers with the practical skills required to configure, deploy, and manage scientific Python projects on HPC infrastructure. Participants will analyze the practical trade-offs, advantages, and performance considerations of modern Python environment management solutions within HPC environments.
Workshop Structure:
Participants will engage in a practical, hands-on session. Following a short overview, attendees will clone a practical GitHub repository and systematically follow guided steps to migrate a baseline Python project from a local machine to a remote HPC environment.
Topics & Tools Covered:
We will explore, compare, and benchmark leading tools across the Python scientific ecosystem:
- Conda & Mamba: Industry-standard tools for complex binary dependencies and isolated environment management.
- Virtualenv & Virtualenvwrapper: Traditional, lightweight isolation frameworks paired with `pip`.
- uv: High-performance, Rust-backed package and environment management.
- Pixi: Modern ecosystem management powered by `rattler` for multi-language, cross-platform reproducibility.
- Poetry: Dependency management and packaging tailored for structured Python applications.
Contact information:
Please contact researchdata@rice.edu if you have questions about the Data@Rice workshop series.