Project deliverable Open Access
Apostolov, Rossen; Abraham, Mark; Laure, Erwin; Lindahl, Erik; Bonvin, Alexandre; de Groot, Bert; Ippoliti, Emiliano
Future hardware development will have a significant impact on all areas of
scientific computing including the Life Sciences. Upcoming extreme-scale compute platforms will offer great opportunities for tackling important, large-scale scientific questions. In this document we analyze from the viewpoint of
biomolecular applications the requirements for different Exascale aspects such as HPC architectures, software management, programming environments, I/O and storage etc. We discuss the near-term hardware developments in processors, network, and memory and I/O in the light of these requirements and we explain the impact that they will have on the core applications. Our findings are already well publicized among the European HPC stakeholders via several working groups which are involved in the development of the ETP4HPC Strategic Research Agenda and the PRACE Scientific Case, as well as the EXDCI (http://exdci.eu) project in which BioExcel is leading the Life Science working group.
Bio-molecular simulation scientists require effective and usable simulation
software that runs well on the hardware resources they can access now. The
development of these codes must also target the likely directions of future
hardware that we have learned from leading vendors and development consortia. This will ensure that the investment in Exascale-era technologies will deliver the expected benefits of improved bio-molecular simulations. These include supporting the design of new drugs, obtaining better understanding of biochemical pathways, and opening new doors for further innovation. This deliverable gives an overview of what we currently see as potential directions and then implementation plans for each of the pilot codes that will suit those directions.
D1.3 - Roadmap of future hardware and long-term development plan for each pilot application.pdf