 |
|
|
|
|
|
|
|
|
|
|
2005 NEWS AND HIGHLIGHTS
08.10.04 - POP Model Ported to Columbia, Scales to 512 Processors
The Project Columbia applications team has ported and performed initial optimization on the Parallel Ocean Program (POP) model, on 512 processors of the Columbia system. The code has been extensively optimized for the Japanese Earth Simulator and Cray X1 architectures. The applications team's effort involves porting and optimization for the SGI Altix architecture. The code now executes at 5.5 simulation years per compute day on the 0.1-degree resolution North Atlantic problem (about 50 million grid points; see background). This performance allows researchers to compute 0.1-degree decadal simulations in just two days. The team is continuing scaling work, and this performance is expected to double for the same number of processors before the end of the year.
The Parallel Ocean Program (POP) was developed at Los Alamos National Laboratory, under the sponsorship of the Department of Energy. POP is an ocean circulation model derived from earlier models in which depth is used as the vertical coordinate. The model solves the three-dimensional primitive equations for fluid motions on the sphere. Spatial derivatives are computed using finite-difference discretizations that are formulated to handle any generalized orthogonal grid on a sphere, including dipole and tripole grids, which shift the North Pole singularity into landmasses to avoid time step constraints due to grid convergence.
Although originally motivated by the adaptation of POP for massively parallel computers, many changes improved not only its computational performance but the fidelity of the model's physical representation of the real ocean.
In the mid-1990s, POP was used to perform high-resolution (0.28 degrees at the Equator) global ocean simulations. The primary motivation for performing such high-resolution simulations is to resolve mesoscale eddies that play an important role in the dynamics of the ocean. Comparison of sea-surface height variability measured by the TOPEX/Poseidon satellite with that simulated by POP gave convincing evidence that still higher resolution was required. At the time, it was not possible to do a higher resolution calculation on the global scale, so an Atlantic Ocean simulation was done with 0.1 degree resolution at the Equator. Using the 0.1 degree case as a benchmark, lower resolution cases were done at 0.2 degrees and 0.4 degrees. Recently, it has become possible to run a 0.1 degree global simulation and such a simulation [on the Columbia system] is in progress.
|
|
|
|
NEWS ARCHIVE
|
|
|
