In-Situ Visualization Solutions Come to the Rescue
DescriptionScientific visualization and analysis are key ingredients in HPC simulation workflows. For decades, the dominant paradigm has been post-hoc visualization; simulation codes iterate and save files to disk, giving the domain scientists the opportunity to read the data back at a later time. However, in recent years where data volumes are growing at an ever-diverging rate as compared to I/O subsystems, this traditional paradigm has been greatly stressed out. In-situ processing helps mitigate these I/O bottlenecks, enabling simulation and visualization calculations to run in-memory, at higher spatial and temporal resolution, while the simulation is running, limiting, or avoiding the transfer of raw data to disks. We will introduce two open-source libraries, “ParaView Catalyst” and “Ascent”, enabling in-situ data analysis and visualization. In-situ libraries can analyze, and transform data, render images, and export results in real-time, but require careful planning because we miss the exploratory, “what-if” analysis process enabled by the interactive visualization applications we are accustomed to. Doing in-situ visualization represents a paradigm shift that will impact the scientist's workflow. We will summarize our early adoption of in-situ visualization scenarios with examples from different simulation fields.
TimeTuesday, June 2717:30 - 18:00 CEST
Computer Science, Machine Learning, and Applied Mathematics