Cubed-Sphere fvcore (Computational Results)
Computational results suggest
the parallelization with 2-dimensional domain decomposition is performing
exceptionally. At 1-deg resolution the cubed-sphere dynamical core scales
well beyond the pure MPI limitations of the lat-lon core. The model is
scaling linearly at ideal performance out to a full Altix node (512 CPUs)
at c360 (0.25-deg) resolution. With additional tuning, it is expected
that the cubed-sphere dynamical core will achieve a new level of computational
performance, allowing parent models at NASA (GEOS), NCAR (CAM) and GFDL
(AM) to scale well beyond current limitations exceeding 10,000s of processors
as we move toward petascale computing environments.
Graph depicting Finite-Volume Dynamical Core Throughput
for 3D Baroclinic Test Case
Graph depicting Cubed-Sphere Finite-Volume Dynamical
Core Scaling
References:
Jablonowski, C. and D. L. Williamson, 2006: A Baroclinic Instability
Test Case for Atmospheric Model Dynamical Cores, Quarterly Journal of
the Royal Meteorological Society, in review
Lin, S.-J. and R. Rood, 1996: Multidimensional flux form semi-lagrangian
transport schemes. Monthly Weather Review, 124, 2046–2070.
Lin, S.-J. and R. Rood, 1997: An explicit flux-form semi-lagrangian
shallow water model on the sphere. Quarterly Journal of the Royal Meteorological
Society, 123, 2477–2498.
Putman, W., S.-J. Lin, and B.-W. Shen, 2005: Cross-platform performance
of a portable communications module the nasa finite volume general circulation
model. International Journal of High Performance Computing Applications,
19.
Williamson, D., J. Drake, J. Hack, R. Jakob, and P. Swarztrauber, 1992:
A standard test set for numerical approximations to the shallow water
equations in spherical geometry. Journal of Computational Physics, 102,
211–224.
+ Back to Scientific Results
+ Back to Overview
+ Back to Introduction
