Geocomputation's future at the extremes: high performance computing and nanoclients

  • Authors:
  • K. C. Clarke

  • Affiliations:
  • Department of Geography/NCGIA, University of California, Santa Barbara, CA

  • Venue:
  • Parallel Computing - Special issue: High performance computing with geographical data
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Recent developments in computing are reviewed, with consideration given to the nature of the human-computer interface and its development. The emergence of the desktop metaphor and the Windows-Icons-Menus Pointers interface, while suitable for most computing tasks, has now played out its course as far as innovation in hardware is concerned. The next phase of computing is only now emerging, and is being fed by changes at the "extremes" of computing, the parallel supercomputer at one end, and the extremely thin or "nano" client at the other. The paradox, of fat servers and ultrathin clients implies a further separation toward the extremes as systems develop. Examples are given of nanoclients, including wearable computers, smart dust, microelectromechanical systems and wireless integrated network sensors. Clearly, the user interface for nanoclients must explicitly or implicitly include geographical space, because the primitive of location, coordinates, will often be a critical part of the data stream. At the parallel and HPC end, the advantages and types of parallel computing are discussed, and examined in the light of what they can do for geographical systems. Some examples of the computing extremes are given in terms of application, including wearable computers, and geo-smart dust. Finally, the relationship between massive combined computing power at both the client and server side are considered in the light of tractability in geoscience. It seems quite likely that a new suite of methods, based in computing, will emerge to replace the standard set of mathematical tools, such as optimization methods, in use today.