PERFORMANCE MODELING OF IN SITU RENDERING by MATTHEW CRIVELLI LARSEN A DISSERTATION

Original approval signatures are on file with the University of Oregon Graduate School. With the push to exascale, in situ visualization and analysis will play an increasingly important role in high performance computing. Tightly coupling in situ visualization with simulations constrains resources for both, and these constraints force a complex balance of trade-offs. A performance model that provides an a priori answer for the cost of using an in situ approach for a given task would assist in managing the trade-offs between simulation and visualization resources. In this work, we present new statistical performance models, based on algorithmic complexity, that accurately predict the run-time cost of a set of representative rendering algorithms, an essential in situ visualization task. To train and validate the models, we create data-parallel rendering algorithms within a lightweight in situ infrastructure, and we conduct a performance study of an MPI+X rendering infrastructure used in situ with three HPC simulation applications. We then explore feasibility issues using the model for selected in situ rendering questions. vi ACKNOWLEDGEMENTS I thank Hank Childs for being an exceptional advisor and for his guidance through the Ph.D. processs. Advanced Simulation & Computing (ASC) program. vii For my parents and my brother, who have supported and encouraged me throughout my life. I would not been able to achieve this without them. To my advisor, Hank Childs, who unwittingly encouraged me to go down this path during a conversation between fellow basketball players, years before he joined the faculty at the University of Oregon. His advising during my time at Oregon was outstanding, and I would not have been able to walk this path without it. I thank my co-authors who all have made valueable contributions to our publications, and those who contibuted to material within this dissertation. They include Navrátil, and Dave Pugmire. I thank all the members of the CDUX research group who have all been supported of one another. Specifically I would like to thank Ryan Bleile,