Making a Case for Efficient Supercomputing

  title={Making a Case for Efficient Supercomputing},
  author={Wu-chun Feng},
  pages={54 - 64}
A supercomputer evokes images of “big iron“ and speed; it is the Formula 1 racecar of computing. As we venture forth into the new millennium, however, I argue that efficiency, reliability, and availability will become the dominant issues by the end of this decade, not only for supercomputing, but also for computing in general. 

Figures, Tables, and Topics from this paper

The Right Metric for Efficient Supercomputing: A Ten-Year Retrospective
  • Chung-Hsing Hsu, Wu-chun Feng
  • Computer Science
  • 2016 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)
  • 2016
This ten-year retrospective is two-fold: (1) to acknowledge the past work through a historical narrative and (2) to highlight the essence of the remaining issues in this research. Expand
Green Supercomputing in a Desktop Box
This paper presents and evaluates such an architectural solution: a 12-node personal desktop supercomputer that offers an interactive environment for developing parallel codes and achieves 14 Gflops on Linpack but sips only 185 watts of power at load - all this in the approximate form factor of a Sun SPARCstation 1 pizza box. Expand
Assessing the Utility of a Personal Desktop Cluster
The above observations provide the motivation for a personal desktop cluster workstation — a turnkey solution that provides an interactive and parallel computing environment with the approximate form factor of a Sun SPARCstation or 1 “pizza box” workstation. Expand
Green Supercomputing Comes of Age
It is envisioned that holistic power-aware technologies will be available and largely exploited in most, if not all, future supercomputing systems. Expand
The Argus prototype: aggregate use of load modules as a high‐density supercomputer
This paper describes the ARGUS prototype, a high‐density, low‐power supercomputer built from an IXIA network analyzer chassis and load modules. The prototype is configured as a diskless distributedExpand
The Green500 List: Encouraging Sustainable Supercomputing
The Green500 List effort will encourage the HPC community and operators of Internet data centers to design more power-efficient supercomputers and large-scale data centers and to supplement the TOP500 List. Expand
ARGUS: Supercomputing in 1/10 Cubic Meter
This work proposes ARGUS, a high density, low power supercomputer built from an IXIA network analyzer chassis and load modules, and compares and contrast the characteristics of the system against various machines including the 32-node Beowulf and LANL’s Green Destiny. Expand
EPIC: A framework to exploit parallelism in irregular codes
Two refinements to the EPIC framework are presented: one that refines the software design of the EPic framework and another that refine the scheduling algorithm of theEPIC framework, which allows to cope with a special class of sets of tasks: set of tasks where asymmetry is insignificant or can be neglected. Expand
Green Destiny and its Evolving Parts
Although the performance of supercomputers on our n-body cosmology code has improved by a factor of nearly 2000 since 1991, the performance per watt has only improved 300-fold and the performance perExpand
The Green500 List
The performance-at-any-cost design mentality ignores supercomputers' excessive power consumption and need for heat dissipation and will ultimately limit their performance. Without fundamental chang...


What's next in high-performance computing?
We can trace the evolution from Crays, to clusters, to supercomputing centers. But where does it go from here?
BEOWULF: A Parallel Workstation for Scientific Computation
It is shown that the Beowulf architecture provides a new operating point in performance to cost for high performance workstations, especially for file transfers under favorable conditions. Expand
The Bladed Beowulf: a cost-effective alternative to traditional Beowulfs
The results of performance benchmarks on the Bladed Beowulf are presented and two performance metrics that contribute to the total cost of ownership (TCO) of a computing system - performance/power and performance/space are introduced. Expand
Avalon: an Alpha/Linux cluster achieves 10 Gflops for $15k
As an entry for the 1998 Gordon Bell price/performance prize, we present two calculations from the disciplines of condensed matter physics and astrophysics. The simulations were performed on a 70Expand
High-Density Computing: A 240-Processor Beowulf in One Cubic Meter
The performance of the Green Destiny cluster is measured using a gravitational treecode N-body simulation of galaxy formation using 200 million particles, which sustained an average of 38.9 Gflops on 212 nodes of the system. Expand
Speeding up N-body Calculations on Machines without Hardware Square Root
  • A. Karp
  • Computer Science
  • Sci. Program.
  • 1992
This note shows how to cut the time for this part of the calculation of the accelerations of the particles by a factor of 3 or more using standard Fortran. Expand
Cramming More Components Onto Integrated Circuits
  • G. Moore
  • Computer Science, Engineering
  • Proceedings of the IEEE
  • 1998
The future of integrated electronics is the future of electronics itself. The advantages of integration will bring about a proliferation of electronics, pushing this science into many new areas.Expand
Letter to Los Alamos
  • National Labora- tory,
  • 2003
Letter to Los Alamos National Laboratory
  • 2003
The design, implemen- tation, and evaluation of mpiBLAST, Best Paper: Ap- plications Track
  • Proceedings of ClusterWorld Conference & Expo (June
  • 2003