The International Conference for High Performance Computing, Networking, Storage, and Analysis
The International Conference for High Performance Computing, Networking, Storage, and Analysis
Authors: Mary Thomas (University of California, San Diego (UCSD)), Thomas Papatheodore (Oak Ridge National Laboratory (ORNL)), Kevin Gott (Lawrence Berkeley National Laboratory (LBNL), National Energy Research Scientific Computing Center (NERSC)), Meifeng Lin (Brookhaven National Laboratory), Mozhgan Kabiri chimeh (NVIDIA Corporation), Sunita Chandrasekaran (University of Delaware)
Abstract: Given the anticipated growth of the HPC market, HPC is challenged with expanding the size, diversity, and skill of its workforce. As we move toward exascale computing, how best do we prepare future computational scientists, and enable established domain researchers to stay current and master tools needed for exascale architectures?
This BoF invites scientists, researchers, trainers, educators, and the RSEs that support them to discuss current learning and development programs, explore adding in-person and virtual hackathons to existing training modalities, and brainstorm implementation strategies to bridge between traditional programming curricula and hands-on skills needed by diverse communities within different environments.
Long Description: The potential for high-performance computing (HPC) to accelerate science is limitless, but not without challenges. Given the anticipated growth of HPC into exascale regions for both scientific computing and the broader enterprise, HPC is feeling the pressure of recruiting and retaining people. It faces the quandary of expanding the size, diversity, and skill of its workforce. At the same time, the COVID-19 pandemic necessitated isolation and remote/distributed work and education approaches that are continuing even as certain restrictions ease. Newer architectures, accelerators, and software stacks are turned out faster than ever before. All creating a perfect storm that HPC must navigate. However, new options have begun to have a significant, positive impact on HPC training programs.
As we move forward in exascale computing we must ask: How can we improve recruitment and better prepare future computational scientists for the upcoming challenges in exascale computing? How do we enable established domain researchers to stay current with the latest software and hardware trends and master the tools needed for the newer compute node architectures? How do we make exascale and HPC more accessible? How do we facilitate the creation of sustainable software?
Traditionally, HPC has had a high barrier to use. Numerous training and development modalities exist, but often are independent of each other. Adding in-person and virtual hackathons to the training mix can bridge traditional programming curricula and hands-on skills needed among the diverse communities across national laboratories, supercomputing centers, and academic environments.
Hackathons and coding bootcamps have evolved from early coding and “bug discovery” sessions to become modern innovation events that combine agile programming and intense mentoring- We have found that using an approach for hackathons that pairs teams of researchers with programming experts who are often experienced in the scientific domain is successful in helping to accelerate and optimize scientific applications on a variety of HPC architectures. The collaborative approach of these events provides the critical accelerated computing skills needed by the scientific community and the professionals that support them and aids in preparing researchers to use current and upcoming supercomputing resources.
To date, over 100 hackathons using this approach have been run worldwide and ~550 scientific applications across multiple scientific domains have been accelerated wholly or in part at Hackathons. Examples include BerkeleyGW, Quantum ESPRESSO, CASTRO, Gkeyll, QUICK, CASTEP, and NWChem/NWChemEx. For additional information, please refer to the published paper: Best Practices in Running Collaborative GPU Hackathons: Advancing Scientific Applications with a Sustained Impact (https://www.hzdr.de/publications/PublDoc-12325.pdf).
This interactive BOF will bring together computational scientists, researchers, trainers, educators, and research software engineers (RSEs) and the staff that support them to: discuss current learning and development programs available; explain some benefits and challenges in implementing hackathons for training; share specific use cases of hackathons, including training “readiness,” outcomes and continued progress; discuss how to engage diverse communities—from students to early career researchers to veteran scientists—within the organization; and brainstorm best practices for initiating and executing these events into their training mix.
URL:
Back to Birds of a Feather Archive Listing