Updating Inshimtu with Catalyst2 and Integrating an HPC MiniApp: Lessons Learned

Overview

KVL has updated Inshimtu, a pedagogical tool designed to facilitate experimentation with in situ visualization techniques. The growing disparity between system processing power and storage capabilities has underscored the importance of in situ analysis. However, adoption of these technologies remains limited due to hesitancy around new coding techniques and perceived workflow complexity. Inshimtu addresses this by serving as a “shim” library, allowing users to experience in situ workflows using their own data without committing to full-scale integration.

In this work, we updated Inshimtu from Legacy ParaView Catalyst to the modern Catalyst2 standard to leverage its improved API and versatile Python scripting capabilities. Furthermore, we developed and integrated a version of the Gray-Scott reaction-diffusion MiniApp—a two-variable application using a memory-bound 7-point stencil kernel—to demonstrate in situ workflows at scale.

The system was successfully tested on the Shaheen III supercomputer, demonstrating the combined functionality of Inshimtu and Gray-Scott in a scientific workflow. This project details the complexities and lessons learned during the transition to Catalyst2 and highlights Inshimtu's utility as a teaching aid for the simulation community. The paper entitled “Updating Inshimtu with Catalyst2 and Integrating an HPC MiniApp: Lessons Learned” can be accessed here.