4/28/2009
Written by
The Parallel Programming Laboratory, which is part of the University of Illinois' computer science department, hosted the 7th Annual Workshop on Charm++ and its Applications April 15-17, 2009. The workshop was supported in part by the university's Institute for Advanced Computing Applications and Technologies (IACAT).
The event included a panel discussion, called "A Single Programming model for clusters and multiprocessor nodes: Dream, Nightmare, Reality, or Vision," featuring leaders in high-performance computing and software from Illinois. The conversation explored the costs and benefits of developing a single programming model that will work for individual multiprocessor nodes, clusters of uniprocessor nodes, and clusters of shared-memory nodes.
Video of the panel and other presentations from the event are available at: http://charm.cs.uiuc.edu/charmWorkshop/program.html.
The workshop also showcased the latest leading research in parallel processing using Charm++, including processor virtualization techniques, the use of GPUs (graphics processing units) and other accelerators, multicore programming, parallel discrete event simulations used for performance prediction, performance analysis and visualization, and Charm++'s role in the Blue Waters sustained-petascale computing project.
Charm++ and the frameworks built upon it are powerful parallel programming systems. By allowing the programmer to divide the computation into a large number of entities, which are mapped to the available processors by an intelligent runtime system, Charm++ improves both productivity and performance.
Charm++ is a crucial part of the NAMD molecular dynamics code developed by Illinois' Theoretical and Computational Biophysics Group. NAMD is used by hundreds of researchers worldwide. It is also one of the first three science and engineering codes being supported by Illinois' Blue Waters project, to scale the code to run at sustained-petascale rates of more than one quadrillion calculations per second.