The parallel volume rendering algorithm consists of three stages: I/O, rendering, and compositing. The I/O stage reads the dataset in parallel from disk. The rendering step is a software ray caster with lighting. The compositing stage is a many-to-many communication exchange whereby individual images are depth-blended into one final image. Further reading: Peterka, T., Yu, H., Ross, R., Ma, K.-L.: Parallel Volume Rendering on the IBM Blue Gene/P. Proceedings of Eurographics Symposium on Parallel Graphics and Visualization 2008 (EGPGV'08) Crete, Greece, April 2008. pdf bibtex
The overall strong scaling is shown below for three data sizes and three image sizes. Since I/O takes up to 97% of the total run time, studying and optimizing data access patterns is a significant part of our team's research. Further reading: Peterka, T., Yu, H., Ross, R., Ma, K.-L., and Latham, R: End-to-End Study of Parallel Volume Rendering on the IBM Blue Gene/P. Proceedings of ICPP'09 Conference, Vienna, Austria, September 2009. pdf bibtex
Other optimizations include data aggregation, hybrid MPI - threaded programming, and organizing the parallel processes into groups that operate on different time-steps concurrently. Further reading: Peterka, T., Ross, R., Yu, H., Ma, K.-L.: Assessing Improvements to the Parallel Volume Rendering Pipeline at Large Scale. Proceedings of SC08 Ultrascale Visualization Workshop, Austin TX, November 2008. pdf bibtex
