A minisymposium at SIAM Conference on Computational Science and Engineering, on 15 March 2015, Salt Lake City, UT
There has been a recent renewed focus on software productivity for scientific applications, especially those that require performance at extreme scales (10k+ and GPU-accelerated cores). The goal of this minisymposium is to engage the broader computational science and engineering software community, to discuss productivity issues that arise when developing high-performance applications. The session will include an overview presentation of forums and efforts focusing on productivity (including a recently funded initiative within DOE); two science application presentations, oriented towards software productivity considerations; and an open facilitated session to gather SIAM CSE community concerns and priorities.
Agenda
Overview: Software Productivity Challenges for Extreme Scale Science, Lois Curfman McInnes, Argonne National Laboratory; Michael Heroux, Sandia National Laboratories; Hans Johansen, Lawrence Berkeley National Laboratory
Abstract. The Department of Energy has initiated research in software productivity for application development and software infrastructure for extreme-scale scientific computing. The eventual goal is to enable “grand challenge” science simulations that can survive and even leverage disruptive changes in extreme-scale computer architectures, and thus enable new frontiers in modeling, simulation, and analysis of complex multiscale and multiphysics phenomena. This session will give an overview of DOE activities, including workshops, participating communities, and pilot projects.
Software Productivity Application: Integrated Modeling for Fusion Energy, David E. Bernholdt, Wael R. Elwasif, and Donald B. Batchelor, Oak Ridge National Laboratory
Abstract. Integrated modeling of plasmas is a key scientific capability for the development of fusion as a power source. We will present our experience developing the Integrated Plasma Simulator (IPS), an environment which enables plasma physicists to be highly productive in developing and carrying out a broad range of coupled simulations, while improving supercomputer resource utilization. The design and implementation of the IPS is quite general, and it is also being used in other domains.
Software Productivity Challenges in Environmental Applications, David Moulton and Ethan T. Coon, Los Alamos National Laboratory; Carl Steefel, Lawrence Berkeley National Laboratory
Abstract. Predictive simulations of environmental applications pose significant challenges for multiscale multiphysics frameworks. Problem complexity requires flexibility to compare different models or model features, to add new models, and to explore model coupling. We will present our experience in addressing these issues in two open-source codes, Amanzi and the Arctic Terrestrial Simulator. We will discuss the importance and challenge of leveraging existing frameworks and libraries as well as components of established codes.
Software Productivity Community Input: Concerns and Priorities (discussion), Jeffrey C. Carver, University of Alabama; Michael Heroux, Sandia National Laboratories
Abstract. This session will start with an overview and history of the software productivity activities across the scientific computing community, followed by an open discussion of requirements, forums, and opportunities for further engagement.