This User Guide offers a series of Quick Start pages to help PyTorch users on LC systems (also on the left-hand menu):
- PyTorch Quickstart Guide for AMD GPU systems (for users on the Tioga, Tuolumne, RZAdams, RZVernal, El Capitan, and Tenaya systems)
- PyTorch Quickstart Guide for CPU systems
- PyTorch Quickstart Guide for NVIDIA GPU systems (for users on the Matrix and RZVector systems)
- PyTorch via WEAVE Quickstart Guide (for pre-built environments with PyTorch)
- Jupyter Quickstart Guide
- WEAVE Quickstart Guide
Users who wish to use Jupyter notebooks / JupyterHub, must utilize the LC Orbit service.
Using Python in LC
Livermore Computing (LC) maintains Python and a set of site-specific packages (modules) on all production TOSS and CORAL systems. Please see our Python documentation page, which includes information on:
- System installations of Python and accessing them via Lmod
- Setting up a Python environment (with virtualenv or Conda) and working with site packages (and installing your own)
- Running python scripts, at scale, and in an HPC batch job
- Integrated Development Environments (IDEs)
- Debugging python with TotalView and pdb
Users who wish to use Jupyter notebooks / JupyterHub, must utilize the LC Orbit service.
Additionally, users on the El Capitan and Tuolumne platforms should be aware of the Spindle tool as it may impact how python libraries are loaded into large-scale HPC jobs.
LLNL users may also be interested in the WEAVE project, which supports CPU- and GPU-enabled python environments that include many open source software tools, including PyTorch. See also the WEAVE Quickstart Guide.