Sandia was recently awarded DOE Advanced Scientific Computing Research (ASCR) funding to develop a framework for in situ data management, analysis, and visualization. Sandia’s approach provides a mechanism for specifying the structure and use of data, decouples code specification from optimization, and facilitates the expression of complex scientific workflows—thereby enabling scientists to develop high-performance, portable codes at extreme-scale.
Working closely with application and industry advisors, our technologies will directly benefit DOE Biological and Environmental Research (BER)+ASCR productivity efforts, NNSA’s Advanced Simulation & Computing (ASC), and DOE FastForward and DesignForward activities.
The underlying research ideas to be explored with this funding are
- that a unified, data-driven programming model for applications and in situ analysis and visualization will enable both productivity and performance,
- the ability to express the structure and use of data should be a first-class component of extreme-scale programming models, and
- that policy decisions regarding the mapping of software onto evolving extreme-scale architectures should not affect the correctness of the code.
- enable productivity and performance,
- provide a foundation for building efficient and effective control flow mechanisms for complex in situ workflows, and
- establish enabling capabilities for interactive in situ environments.
This is a multiple-institution effort.
- Sandia Combustion Advisor: Jackie Chen (in Sandia’s Reacting Flow Research Dept.)
- Los Alamos Climate Advisor: Phil Jones
- Argonne Cosmology Advisor: Katrin Heitmann
- NVIDIA Advisor: Steve Keckler
- AMD Advisor: Mike Schulte