Massively Parallel Visualization on Leadership Computing Resources - Tom Peterka, ANL
Participants: Tom Peterka, MCS Division, Argonne National Laboratory
Tech Meeting 11:00am
EVL / UIC
842 W. Taylor, RM 2068
EVL alumni Tom Peterka, currently a postdoctoral appointee at Argonne National Laboratory’s, Mathematics and Computer Science Division presents his latest research in massively parallel visualization on the Blue Gene/P (BG/P) at Argonne’s Leadership Computing Facility.
Supercomputer size, computing power, and computational results are rapidly approaching petascale proportions. For example, with the recent deployment of the Blue Gene/P (BG/P) at Argonne’s Leadership Computing Facility, scientists can compute simulations at a rate of hundreds of teraflops (trillions of floating point operations per second) and produce datasets hundreds of terabytes in size. This will soon grow to petaflops and petabytes. Performing post-simulation analysis and visualization operations using a dedicated analysis computer cluster, as we do now, will not scale with the rapidly expanding computational power of the supercomputers performing the simulations. Studying how simulation, analysis, and visualization can be collocated on the same leadership system has the potential benefits of eliminating data transfers and lays the groundwork for performing in situ visualization - viewing results of a simulation as they unfold. In this research, we examine how a popular scientific visualization operation, volume rendering, can be massively parallelized on BG/P and evaluate its performance when scaling from 64 to 32,768 processor cores.
Date: September 10, 2008
Document: Massively Parallel Visualization