Mar 2017 Handout for Scalable Analysis Methods and In Situ Infrastructure for Extreme Scale Knowledge Discovery

The traditional approach for scientific visualization and analysis is to first write data to persistent storage for subsequent, post hoc use. This approach is increasingly impractical for DOE science given the widening gap between computational and I/O rates. An alternative approach, known as in situ processing, is to perform as much visualization and analysis processing as possible without first writing data to persistent storage.

This project is tackling a set of R&D research problems to produce in situ software technology and tools that will help DOE computational science projects perform accurate and timely data analysis without the need to first write data to disk for subsequent analysis. We combine software technology R&D activities with close partnerships with key DOE computational science projects so that our priorities are guided by the most pressing scientific problems and so that the results of our work are put into production use by DOE science projects on DOE high-performance computational (HPC) platforms.