The peril of the petascale: challenges to scientific computing

John Clyne
NCAR/CISL

Abstract
Thanks to over 40 years of steady advancement in microprocessor chip design and fabrication we are now poised to enter an era of petascale computing. Yet unsolved problems exist for computational scientists in the current period of terascale computing that will only become further exasperated as we move forward. High resolution numerical simulations can lead to vast volumes of data that may be impenetrable with the analysis resources available to many scientists today. The result is often a significantly reduced scientific return on our largest computational efforts. This talk will explore the challenges associated with large scale data analysis and will discuss fundamental changes in computational science that we feel are increasingly necessary to ensure full return on our scientific investments as we approach the era of the petascale.

Back to agenda