LANL CSSE L2: Case Study of In Situ Data Analysis in ASC Integrated Codes [Slides]
John M. Patchett, USDOE National Nuclear Security Administration (NNSA), James Paul Ahrens, Boonthanome Nouanesengsy, Patricia Fasel, Patrick O’Leary, Christopher Sewell, Jonathan Woodring, Christopher Mitchell, Li‐Ta Lo, Kary Myers, Joanne R. Wendelberger, Curtis Canada, Marcus Daniels, Hilary Abhold, Gabriel Rockefeller · 2013
The massive, complex datasets generated by scientific simulations on next-generation computer architectures at extreme scale pose new challenges for data management and scientific understanding. A critical element in meeting these new challenges is in situ data analysis where information is reduced and analyzed as close to the point of generation as possible, for example, at run time or as the data stream to longer term storage. This milestone will demonstrate in situ data analysis in xRAGE, an ASC Eulerian Radiation Hydrodynamics code. We will work with the Eulerian Applications Project to identify target problems that will exercise physics of interest to the program, and we will work closely with users to maximize the achieved usability and productivity improvements. We will investigate enhancements to the in situ process that enable and improve techniques of data reduction and data triage in addition to improving in situ performance by using portable acceleration technology. We will report on their viability and performance in xRAGE or another ASC code. These products will provide ASC IC developers with reference codes and tools that will benefit DSW by advancing the technology for predictive science at extreme scales.