Vista is targeted towards scientific computing projects that require highly capable resources for AI and other HPC applications. Researchers may need to adjust their user environment and application program to make full use of Vista's capabilities. While we recommend a full read-through of this roadmap, here are some common workflows you may want to consider.

Training a Machine Learning Model with PyTorch

• User Environment and Storing and Moving Data topics for basic understanding of the Vista system
• ML and AI page to set up a PyTorch environment (an Apptainer container could be used for a simple model running on a single node)
• Running Jobs and Managing Jobs topics for model training through Slurm

Compiling and Running a Program with GPU Acceleration

• User Environment and Storing and Moving Data topics for basic understanding of the Vista system
• Compiling for GPUs, though this may not be necessary if a prebuilt, GPU-enabled program already exists in one of TACC's modules
• Running Jobs and Managing Jobs topics for running the program through Slurm

Accessing the TACC Analysis Portal (TAP) from Cornell University Center for Advanced Computing

Running Jupyter Notebooks on the TACC Analysis Portal (TAP) from Cornell University Center for Advanced Computing