Hello, A change in allocation has been requested: Requester: norris (Boyana Norris) Project: pflotran Title: PFLOTRAN Performance Analysis and Tuning Description: PFLOTRAN is a massively parallel code developed at LANL for the accurate prediction of multiscale subsurface processes. PFLOTRAN has many applications and guides safer industrial techniques; for example, it can help geoscientists determine how carbon dioxide flows through aquifers and deep geologic formations during carbon-capture and storage. PFLOTRAN's performance has been studied extensively on Intrepid and Jaguar (Cray XT/5), but no performance studies on Intel-based clusters have been performed. The efficiency of the code is a small fraction of the peak on both BG/P and Cray XT/5. In the first year of this project, we focused mainly on analyzing and improving the performance of the sequential portions of key computational kernels (matrix-vector product, triangular linear solution, and some of the chemistry computations), which required many single-core runs of these kernels. During the next year, we plan to study and possibly improve the scalability of the Jacobian computation and the linear system solution. We'll also investigate approaches for managing the load imbalance between the flow and transport stages. This will require a number of parallel scaling runs. We will use input decks that can run on different ranges of cores, from one node to the whole machine. We'll target runtimes of less than an hour in most cases (by changing the number of time evolution steps). Current: undetermined amount Justification: Requested: 200000 A specific reason has been given: We request additional time to continue our research on automatically tuning key computations in applications such as PFLOTRAN. Our goal is to advance the state-of-the-art in automatic empirical tuning and adaptation of scientific applications on high performance architectures and thereby significantly increasing scientific productivity. Our current focus is on computational kernels that occur in PFLOTRAN. We have systematically evaluated the strengths and limitations of autotuning techniques using our growing collection of kernels on Fusion. This study has improved our understanding of large-scale auto tuning and also helped us to identify open issues that need to be addressed, which requires developing novel, flexible algorithmic approaches. During the current allocation period, this research has produced two conference/workshop articles. – P. Balaprakash, S. M. Wild, and P. Hovland. Can search algorithms save large-scale automatic performance tuning? in Proceedings of the International Conference on Computational Science, ICCS 2011, Procedia Computer Science, Vol. 4, pp. 2136-2145, 2011. – P. Balaprakash, S. M. Wild, and B. Norris. SPAPT: Search Problems in Automatic Performance Tuning, Preprint ANL/MCS-P1872-0411, April 2011. (will be submitted to ISPASS 2012) This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System