n GPU computing, performance depends on parallelism. This session begins with an introduction to parallel computing concepts and NVIDIA GPU hardware, followed by an overview of the two main ways of expressing GPU parallelism in code: CUDA and OpenMP.  We show examples of syntax for each and discuss how to compile and run such codes on a cluster, where resources are managed by the Slurm scheduler. We also describe ways of running Python codes on a GPU-enabled cluster. We conclude with an overview of techniques for assessing CPU and GPU utilization, along with a brief introduction to NVIDIA's GPU profiling tools. Prerequisites: Some familiarity with Linux commands and programming concepts.