Scaling Productivity and Innovation on the Path to Exascale with a 'Team of Teams' Approach

摘要

One of the core missions of the Department of Energy (DOE) is to move beyond current high performance computing (HPC) capabilities toward a capable exascale computing ecosystem that accelerates scientific discovery and addresses critical challenges in energy and national security. The very nature of this mission has drawn a wide range of talented and successful scientists to work together in new ways to push beyond the status-quo toward this goal. For many scientists, their past success was achieved through efficient and agile collaboration within small trusted teams that rapidly innovate, prototype, and deliver. Thus, a key challenge for the ECP (Exascale Computing Project) is to scale this efficiency and innovation from small teams to aggregate teams of teams. While scaling agile collaboration from small teams to teams of teams may seem like a trivial transition, the path to exascale introduces significant uncertainty in HPC scientific software development for future modeling and simulation, and can cause unforeseen disruptions or inefficiencies that impede organizational productivity and innovation critical to achieving an integrated exascale vision. This paper identifies key challenges in scaling to a team of teams approach and recommends strategies for addressing them. The scientific community will take away lessons learned and recommended best practices from examples for enhancing productivity and innovation at scale for immediate use in modeling and simulation software engineering projects and programs.