Performance evaluation of open source seismic data processing packages
Contribuinte(s) |
Xiang, Yang Cuzzocrea, Alfredo Hobbs, Michael Zhou, Wanlei |
---|---|
Data(s) |
01/01/2011
|
Resumo |
In many businesses, including hydrocarbon industries, reducing cost is of high priority. Although hydrocarbon industries appear able to afford the expensive computing infrastructure and software packages used to process seismic data in the search for hydrocarbon traps, it is always imperative to find ways to minimize cost. Seismic processing costs can be significantly reduced by using inexpensive, open source seismic data processing packages. However, hydrocarbon industries question the processing performance capability of open source packages, claiming that their seismic functions are less integrated and provide almost no technical guarantees for one to use. The objective of this paper is to demonstrate, through a comparative analysis, that open source seismic data processing packages are capable of executing the required seismic functions on an actual industrial workload. To achieve this objective we investigate whether or not open source seismic data processing packages can be executed using the same set of seismic data through data format conversions, and whether or not they can achieve reasonable performance and speedup when executing parallel seismic functions on a HPC cluster. Among the few open source packages available on the Internet, the subjects of our study are two popular packages: Seismic UNIX (SU) and Madagascar.<br /> |
Identificador | |
Idioma(s) |
eng |
Publicador |
Springer-Verlag |
Relação |
http://dro.deakin.edu.au/eserv/DU:30043196/aziz-performance-evid-2011.pdf http://dro.deakin.edu.au/eserv/DU:30043196/aziz-performanceevaluationof-2011.pdf http://dx.doi.org/10.1007/978-3-642-24650-0_37 |
Direitos |
2011, Springer-Verlag Berlin Heidelberg |
Palavras-Chave | #High performance computing #Madagascar #open source seismic data processing packages #seismic data processing #Seismic UNIX |
Tipo |
Conference Paper |