Raspberry SCI: An Inexpensive High Performance Computing Cluster for WRF Simulations within the Valparaiso University Meteorology Program

Faculty Sponsor

Kevin Goebbert

College

Arts and Sciences

Discipline(s)

Meteorology

Presentation Type

Poster Presentation

Symposium Date

Spring 5-3-2019

Abstract

An inexpensive high performance computing cluster constructed of Raspberry Pis is compared to a standard multi-core computer server that is capable of running the Weather Research and Forecasting (WRF) Model to simulate various weather phenomenon from severe weather outbreaks to high impact lake effect snow bands that blanket Valpo in snow. The initial idea for this project came from similar work being done at the National Center for Atmospheric Research (NCAR). Numerous Raspberry Pis, small inexpensive computers, are connected to each other, resulting in a total of 40 to 50 cores. The high computing power allows for substantial weather simulations to be completed. The Raspberry Pi cluster is then compared to a 15-core server to assess the effectiveness of running the WRF. The clock-time completion of the WRF simulations serve as a comparison between the two systems. This poster describes the variability of using the Raspberry Pi cluster for use in an educational setting and research within the meteorology program at Valparaiso University. We thank the University Guild for their funding that made this project possible.

Biographical Information about Author(s)

This project was inspired by the National Center for Atmospheric Research. After attending a conference last spring, Isaac and I decided to make a cluster of our own. We applied for a Guild research grant and were granted one. With their support, we worked with both the meteorology and computer science department at Valpo to complete this project.

For Artistic Presentations

wall poster with 8' table in front of it for their Raspberry Pi display.

This document is currently not available here.

Share

COinS