Component architectures in the next generation of ultrascale scientific computing: challenges and opportunities

  • Authors:
  • David E. Bernholdt

  • Affiliations:
  • Oak Ridge National Laboratory, Oak Ridge, TN

  • Venue:
  • Proceedings of the 2007 symposium on Component and framework technology in high-performance and scientific computing
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Component architectures for high-end scientific computing are still a relatively new idea, and their most effective use and benefits are active areas of research for both developers and users of such approaches. At the same time, however, the scale of both scientific simulations and the computer hardware on which they're run has been growing rapidly, generating an increasing desire for more complex software systems (such as coupled simulations), new programming models and languages, and more complex hardware environments. This paper presents the author's analysis of how these trends might interact with component-based software engineering (CBSE) for scientific computing, in some cases exacerbating existing challenges or posing new ones, or in other cases offering opportunities in which the component environment might be leveraged to give software developers new capabilities or simplify challenges they face.