Gaspra - software for astronomical image processing in high performance computing clusters

Author(s)

Andrijauskas, Fabio ; Sampaio Gradvohl, Andre Leon

• Document type: Journal article
• Source: REVISTA BRASILEIRA DE COMPUTACAO APLICADA; v. 6, n. 2, p. 11-pg., 2014-10-01.

Abstract

There is a huge amount of information generated by telescopes as astronomical images. Therefore, software and equipment could process these images to find new phenomena and obtain new knowledge about the space. Considering the need of rapid processing of those images, we present a software for astronomical image processing in high-performance computing clusters, which use shared and distributed memory systems, called Gaspra. We designed Gaspra to batch process large sets of astronomical images, allowing researchers to create scientific workflows to obtain new knowledge from these astronomy imagery data sets. Experiments with Gaspra show a 3.5-fold speedup to process a single image in five processing nodes, each node supporting 64 different threads. (AU)

• FAPESP's process: 10/50646-6 - High performance computing applied in interdisciplinary problems
• Grantee: Vitor Rafael Coluci
• Support Opportunities: Regular Research Grants
• FAPESP's process: 11/00861-0 - Metalanguage for high performance computing on hybrid architectures
• Grantee: André Leon Sampaio Gradvohl
• Support Opportunities: Regular Research Grants