The OSG OSPool provides access to a diverse set of compute resources with varying CPU architectures and specialized hardware like ARM64 and GPUs. This short training introduces researchers to the key differences between ARM64, x86_64 and x86_64 microarchitectures, discusses how to build and run portable software across them, and outlines considerations for targeting GPU-enabled resources. Attendees will learn how to request specific hardware types in their jobs, optimize for performance and compatibility, and ensure their jobs take full advantage of the available heterogeneity in the OSPool.
Training is online; zoom information will be sent the day of the training.
Learn how to scale your research using the OSPool, a high throughput computing (HTC) resource for running large numbers of computational jobs. This training will introduce the fundamentals of HTC, guide you through getting started on the OSPool, and include a hands-on walkthrough where you'll submit and manage your own jobs. Ideal for researchers who want to scale their analyses efficiently and new/prospective OSPool users.
Training is online; zoom information will be sent the day of the training.
Many programs used by researchers are written in Python. In this training, you'll learn how to set up and use Python in the OSPool. We'll start with how to deploy a consistent software environment, primarily through the use of containers. We'll also cover how to use Conda for managing and deploying Python environments. Then we'll cover how to set up your execution script for running Python jobs. Finally, we'll discuss the practical considerations for adapting your Python programs to a high throughput computing system.
Training is online; zoom information will be sent the day of the training.
