In Situ Post-Processing with VisIt, Libsim, and FieldView
Speaker: Brad Whitlock, Intelligent Light
Presentation
Abstract
The widening gap between compute performance and the ability to store computation results dictates that data analysis and visualization be performed in situ during simulation runs. Saving analysis results and data extracts offers tremendous opportunities for reduction in storage costs while also permitting greater access to data in the temporal domain. Libsim is a library which enables a fully-featured visualization tool, VisIt, to request data as needed from the simulation and apply visualization algorithms in situ with minimal modification to the application code. In this talk, the motivation for in situ processing is presented and the steps to instrument a simulation using Libsim are covered. Instrumenting simulation codes with Libsim opens up new possibilities for higher productivity workflows such as automatic creation of FieldView extract databases (XDBs) which can be visualized efficiently using the post-processing tool FieldView.
Biography
Brad Whitlock is a visualization and post-processing engineer in Intelligent Light’s Applied Research Group. He is also a founding developer of VisIt, which started at Lawrence Livermore National Laboratory. VisIt is a powerful, massively parallel visualization tool designed for high performance computing environments. Brad joined Intelligent Light in 2013 to develop a hardened version of VisIt that will serve as an enabling technology for Intelligent Light’s commercial data analysis and visualization products.