The University of Tennessee's Jack Dongarra is involved with the International Exascale Software Project, a global effort to devise a roadmap that outlines issues, priorities, and the software stack required to support exascale computing. He says the transition to exascale computing is going to be more dramatic than earlier transitions, and this will result in a great deal of strain at the software point.
Dongarra argues that exascale software development should begin as soon as possible, given the steepness of the change from petascale to exascale. "The extreme parallelism, the hybrid design, and because the tightening of the memory bandwidth bottleneck is going to become more extreme as we move to the future, we have to start addressing these issues now," he says. He cites cost and power consumption among the limitations of exascale architecture, while exascale hardware is expected to follow two paths—lightweight processors and commodity processors combined with accelerators.
Dongarra says the development of exascale architectures may require government encouragement, such as financial incentives offered to manufacturers.
From HPC Wire
View Full Article
Abstracts Copyright © 2010 Information Inc., Bethesda, Maryland, USA