The U.S. Exascale Computing Project
Authors: Paul Messina (Argonne National Laboratory)
Abstract: The U.S. Exascale Computing Project (ECP) aims to develop at least two diverse, capable exascale architectures, a broad set of exascale applications that meet mission and science needs, and a software stack to support them. ECP is part of the National Strategic Computing Initiative launched by the Obama Administration in July 2015. This BOF will offer the community the opportunity to hear an update on the current scope, status, and activities of the project, as well as future plans, and to interact with and ask questions of the ECP leadership team.
Long Description: The Exascale Computing Project (ECP) is a new and very important project for the entire high performance computing community. It is a U.S. Department of Energy project, jointly funded and managed by the DOE Office of Science and the National Nuclear Security Administration. ECP consists of four technical focus areas:
1. Application Development (Lead: Doug Kothe, ORNL): Deliver exascale science and mission performance on a suite of applications that are ready for efficient execution on exascale systems.
2. Software Technology (Lead: Rajeev Thakur, ANL): Conduct R&D on tools and methods that enhance productivity and facilitate portability; enhance the software stack that Office of Science and NNSA applications rely on to meet the needs of exascale applications and evolve it to efficiently utilize exascale systems.
3. Hardware Technology (Lead: Jim Ang, SNL): Support supercomputer vendor R&D on hardware-architecture designs needed to build exascale systems.
4. Exascale Systems (Lead: Terri Quinn, LLNL): Support testbeds, advanced system engineering development by the vendors, incremental site preparation, and cost of system expansion needed to acquire capable exascale systems.
ECP's goal is capable exascale computing, not just the development of an ExaFLOP system. The capable exascale goal requires the evolution and development of a new generation of applications, software stack, and hardware technology that is capable of exascale levels of application performance while meeting system-level energy efficiency and reliability goals.
ECP has been making the transition to project status since its inception as the Exascale Computing Initiative in 2015. In the past few months, the ECP technical scope has been defined, reviews have been conducted, and calls for proposals have been issued in applications, codesign, and software and hardware technologies.
This BoF offers an excellent opportunity for the community to engage with the full technical leadership of the ECP, hear a brief synopsis of the technical scope and plans of ECP, and ask questions about the project. The ECP Project Director as well as the leaders of each of the four focus areas will give a brief update on current activities and future plans. There will be ample time for the audience to interact with the ECP leadership team, ask questions, and offer suggestions.
Plans for the ECP call for launching many projects nationwide, touching many stakeholders throughout the HPC community. Anyone interested in the U.S. national initiative for achieving capable exascale computing should attend this session.
Birds of a Feather Index