RCSG TRAINING
DISCOVER YOUR PATH
The Research Computing Support Group (RCSG) at The University of Texas at San Antonio (UTSA) offers specialized training sessions to support researchers with their computational needs. These training sessions cover high-performance computing, data analysis, and software development, ensuring that researchers have the necessary tools and expertise to advance their projects efficiently.
COURSES
AI APPLICATIONS ON THE ARC HPC CLUSTER
This hands-on course guides participants through setting up a Python environment for AI projects on Arc. Topics include launching Jupyter Notebooks, submitting batch jobs, leveraging multiple GPUs, managing long training sessions with checkpoints, and optimizing LLM fine-tuning with Unsloth. Perfect for AI practitioners looking to scale deep learning models efficiently
ARC ONBOARDING TRAINING
This introductory course provides essential guidance on accessing and utilizing the Arc High-Performance Computing (HPC) cluster. Participants will learn how to log in using various methods, run jobs efficiently, and understand job scheduling with Slurm. The training covers best practices for storage, usage policies, and an introduction to parallelization techniques to optimize computing performance.
CUDA PROGRAMMING
Developed by Nvidia, CUDA enables parallel computing on GPUs for general-purpose processing. This hands-on session covers CUDA programming basics in C/C++ and Python, training deep learning models with TensorFlow, and accelerating array calculations using CuPy. Ideal for those looking to harness GPU power for high-performance computing.
MATLAB FOR HIGH-PERFORMANCE COMPUTING
This course explores techniques to optimize applications using UTSA's Arc HPC cluster, with a focus on scaling computations to the cloud. Participants will learn best coding practices, how to identify performance bottlenecks, and how to use parallel processing constructs to enhance performance on multicore systems. The session also covers job monitoring and workload management for efficient execution.
PARALLELIZATION WITH C/C++
This course introduces fundamental parallel computing concepts in C/C++, covering multithreading, multiprocessing, and message passing with MPI. Participants will learn scalability principles, including Amdahl’s Law and the Law of Diminishing Returns, to effectively optimize performance on HPC clusters.
PARALLELIZATION WITH PYTHON
This training explores Python’s parallel computing capabilities, including multithreading, multiprocessing, and MPI for distributed computing. Participants will also learn GPU acceleration with CuPy and key performance optimization principles to enhance large-scale computations on HPC clusters.
PARALLELIZATION WITH R
Designed for statisticians and data analysts, this course explores parallel computing in R to enhance performance on HPC clusters. Attendees will gain hands-on experience with single-node parallelization and leveraging MPI for multi-node computations. Ideal for those seeking to optimize R applications for large-scale data analysis.