Skip to main content

(emsl2691)Development of a Scalable High-Performance Cloud Model


EMSL Project ID
2591

Abstract

Objective. This is a companion proposal to the LDRD project under the same name currently funded under the Computational Science and Engineering Initiatives. The LDRD project is designed to bring state-of-the-art parallelization tools to Pacific Northwest National Laboratory (PNNL) and apply these tools to a Large-Eddy Simulation (LES) cloud model. The direct outcome of the project is the development of a parallel and scalable version of the cloud model and testing it under distributed- and shared-memory multiprocessor environments. The specific objective of this project is to test the newly developed parallel code on a suit of Massively-Parallel Processing (MPP) system at EMSL. The proposed test will address the model portability and performance. Approach. Under the current LDRD project, the parallel version of the cloud model has been developed using the Scalable Modeling System (SMS) provided by the Advanced Computing Branch at NOAA's Forecast Systems Laboratory (FSL). At the user's end, the system consists of high-level directives inserted into the code in the form of Fortran comments. Using a preprocessor, these comments are converted into calls to the run-time library built on top of the MPI. The SMS software package has been installed on a dual-processor Sun workstation and 16-node Linux cluster where the parallel code is being tested. In this project, we propose to implement the SMS and transfer the model to the 240-node Colony cluster at the Molecular Science Computing Facility (MSCF). First of all, this will provide a test of model portability, which is very important for future production runs on a variety of computational platforms. Second, access to the 240-node cluster will allow us to study the model scalability to a much greater degree than is possible on a 16-node cluster. We estimate that for these test we will require 2000 node hours of computer time on Colony. Mikhail Ovtchinnikov will port and set the model up in new environment and Joel Malard will conduct performance tests. From this project we expect to obtain an adequate assessment of the model performance, which is crucial for writing intelligent scientific proposals in the future and obtaining further programmatic funding.

Project Details

Project type
Capability Research
Start Date
2002-07-19
End Date
2002-09-30
Status
Closed

Team

Principal Investigator

Konstantin Ovchinnikov
Institution
Pacific Northwest National Laboratory

Team Members

Joel Malard
Institution
SIMUCAD Design Automation

Related Publications

Govind N, KA Lopata, RJ Rousseau, A Andersen, and K Kowalski. 2011. "Visible Light Absorption of N-Doped TiO2 Rutile Using (LR/RT)-TDDFT and Active Space EOMCCSD Calculations." The Journal of Physical Chemistry Letters 2(21):2696-2701. doi:10.1021/jz201118r
Lopata KA, R Reslan, MI Kowalska, D Neuhauser, N Govind, and K Kowalski. 2011. "Excited-state studies of polyacenes: A comparative picture using EOMCCSD, CR-EOMCCSD(T), range-separated (LR/RT)-TDDFT, TD-PM3 and TD-ZINDO." Journal of Chemical Theory and Computation 7(11):3686-3693.
Ovtchinnikov M, and SJ Ghan. 2005. "Parallel Simulations of Aerosol Influence on Clouds Using Cloud-Resolving and Single-Column Models." Journal of Geophysical Research. D. (Atmospheres) 110(D15):D15S10.
Reslan R, KA Lopata, CD Arntsen, N Govind, and D Neuhauser. 2012. "Electron transfer beyond the static picture: A TDDFT/TD-ZINDO study of a pentacene dimer." Journal of Chemical Physics 137(22):22A502-1 - 22A502-6.
Zhang Z. 2014. "Spin–orbit DFT with Analytic Gradients and Applications to Heavy Element Compounds." Theoretical Chemistry Accounts 133:1588. doi:10.1007/s00214-014-1588-0