When viewing the Technical Program schedule, on the far righthand side
is a column labeled "PLANNER." Use this planner to build your own
schedule. Once you select an event and want to add it to your personal
schedule, just click on the calendar icon of your choice (outlook
calendar, ical calendar or google calendar) and that event will be
stored there. As you select events in this manner, you will have your
own schedule to guide you through the week.
You can also create your personal schedule on the SC11 app (Boopsie) on your smartphone. Simply select a session you want to attend and "add" it to your plan. Continue in this manner until you have created your own personal schedule. All your events will appear under "My Event Planner" on your smartphone.
Fast One-Sided Communication on Supercomputers and Application to Three Scientific Codes
SESSION: Research Poster Reception
EVENT TYPE: ACM Student Research Competition Poster, Poster, Electronic Poster
TIME: 5:15PM - 7:00PM
SESSION CHAIR: Bernd Mohr
AUTHOR(S):Jeff R. Hammond, Sreeram Potluri, Zheng (Cynthia) Gu, Alex Dickson, James Dinan, Ivo Kabadshow, Pavan Balaji, Vinod Tipparaju
ROOM:WSCC North Galleria 2nd/3rd Floors
ABSTRACT: We have developed a new library for one-sided communication and applied it to three different scientific codes: the NWChem computational chemistry application, the ScaFaCoS (Scalable Fast Coulomb Solvers) library, and the NEUS (non-equilibrium umbrella sampling) application. All three codes rely upon asynchronous communication, such as one-sided put, get, accumulate and remote atomics (e.g. fetch-and-add). Our library was designed to meet the requirements of current and next-generation interconnects found in Blue Gene/P and Q, PERCS and Cray XE architectures, such as dynamic routing, DMA engines and hardware remote atomics. The synchronization semantics and implementation have been designed for scalability to more than one million ranks. We demonstrate the scaling of all three applications to thousands of cores. As a scaling exemplar, the fast multiple method (FMM) in ScaFaCoS is demonstrated to scale to 300K ranks of Jugene, which is impossible with MPI and ARMCI for both semantic and implementation reasons.
Bernd Mohr (Chair) - Juelich Supercomputing Centre