Communications of the ACM
Programming Model for Supercomputers of the Future
Carsten Lojewski (left), Christian Simmendinger (center), and Rui Machado (right) developed a programming model that uses high-performance computers as efficiently as possible.
Credit: Dirk Mahler / Fraunhofer-Gesellschaft
Next-generation supercomputers could benefit from the Global Address Space Programming Interface (GPI), an asynchronous programming model developed at the Fraunhofer Institute for Industrial Mathematics.
Even as demand for supercomputers rises, the message passing interface, which supercomputers rely upon to ensure that microprocessors in distributed systems can communicate, has been in place for 20 years and is nearing the limits of its capabilities. "I was trying to solve a calculation and simulation problem related to seismic data," says the Fraunhofer Institute's Carsten Lojewski. "But existing methods weren't working. The problems were a lack of scalability, the restriction to bulk-synchronous, two-sided communication, and the lack of fault tolerance. So out of my own curiosity I began to develop a new programming model."
Lojewski created GPI based on the parallel architecture of high-performance computers with maximum efficiency. GPI is an asynchronous communication model based on remote completion that enables each processor to access all data directly without impacting other parallel processes, regardless of memory type.
Lojewski notes GPI was created as a parallel programming interface rather than a language, which means it can be used universally.
From Fraunhofer Institute
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found