ACM's Gordon Bell Prize for special achievement in high performance computing has been won by a team of researchers from the U.S. Department of Energy's (DOE) Lawrence Berkeley National Laboratory for their work concerning nanostructures' energy harnessing potential. The scientists utilized the DOE's National Energy Research Scientific Computing Center (NERSC) at Berkeley Lab, the Argonne Leadership Computing Facilities at Argonne National Laboratory, and the National Center of Computational Sciences (NCCS) at Oak Ridge National Laboratory to test the Linearly Scaling Three Dimensional Fragment (LS3DF) technique. LS3DF presents a more efficient method for calculating energy potential because it is based on the observation that the total energy of a large nanostructure system can be divided into small fragments, and each fragment can be calculated independently. The LS3DF application attained a speed of 135 teraflops per second on NERSC's Cray XT4 system, 224 teraflops per second on Argonne's IBM BlueGene/P supercomputer, and 442 teraflops per second on a Cray XT5 system at the NCCS. The electronic composition of a 3,500-atom ZnTeO alloy was computed by the algorithm using the NERSC supercomputer, and the run confirmed that the code could be employed to compute properties of the alloy that previously had been experimentally observed. The model cleared the way for a projection of the alloy's efficiency as a new solar cell material, illustrating LS3DF's effectiveness as the first electronic structure code that runs efficiently on systems with tens to hundreds of thousands of cores. "Using a linear scaling algorithm, we can now study systems that would otherwise take over 1,000 times longer on even the biggest machines today," says Berkeley Lab researcher Juan Meza. "Instead of hours, we would be talking about months of computer time for a single study."
From Berkeley Lab
View Full Article