The Inter University Computing Centre (IUCC) in Israel and Amazon Web Services have joined forces to train researchers and research software engineers (RSEs) in the use of AWS for high performance computing (HPC) at the PRACE Winter School, 7-9 December 2021. The organizations have issued a call for participation.
PRACE (the Partnership for Advanced Computing in Europe) is an organization dedicated to high-impact scientific discovery and engineering research and development enabled by computational techniques. They help drive European competitiveness for the benefit of society. PRACE delivers on its mission by offering world class computing and data management resources (through a peer-review process) but crucially, by providing advanced HPC training, too.
To see the full program, see the PRACE Winter Workshop event page. To register, fill in the form at the PRACE Winter School Application page. To find out more or get help with registering, contact hank@mail.iucc.ac.il. Registrations close on October 15, 2021.
Graduate students and post-docs across the 26 countries and associated members covered by the PRACE initiative are invited. By participating they’ll gain practical skill using cloud in a wide range of HPC workloads and find out how elastic HPC infrastructure can complement the techniques used in their algorithms.
IUCC recognizes that research computing in the cloud is proving to be transformative. Democratizing access to next generation computational research tools is at the top of their curriculum for the 2021 PRACE Winter School. Customers tell us that researchers can accelerate the time to science when they have access to cost-effective, scalable, and secure computing, available anytime it’s needed. With AWS, they can quickly stand up these complex HPC environments in minutes with scaling capabilities that match the needs of their simulations. This means getting a result sooner to inform the next step in their research. The skills they’ll develop in the Winter Workshop will mean the logistics of creating and using that infrastructure will be a solved problem, leaving them to really focus on the science instead.
In collaboration with IUCC, we’ve put together a day-long program:
- An introduction to the basics of accessing and using AWS services for research computing.
- Building an HPC cluster in the cloud, including enabling an interactive desktop for scientific visualization.
- How to choose from the large number of different Amazon EC2 instance types to optimize for your specific workload.
- Run a highly parallel application like OpenFOAM and GROMACS.
- Go beyond traditional HPC by setting up an event-driven simulation system using EFISPEC3D to simulate ground motion.
- How to deploy containers for HPC and AI/ML workloads.
Supercomputing has come a long way in recent decades, and AWS is working to help customers everywhere access its state-of-the-art techniques. We’ve built P4d-based EC2 Ultraclusters, created our own cloud-scale network fabric for HPC, and bottled the HPC cluster software stack in an open-source project.
Descartes Labs exploited these capabilities with their ‘DIY’ #40 entry into the TOP500, showing how accessible it is for scientists and engineers everywhere to perform highly demanding computational tasks whenever they need to, using their own supercomputer when they need it. Scaling like this means that as the problem-size grows, researchers can estimate early on what volume of resources they might need later, and they can gradually test a new algorithm’s ability to use computational resources, without wasting any of them.