Perspectives on Supercomputing: Three Decades of Change

by Paul R. Woodward

in IEEE Computer, Vol. 29, No. 10, October 1996, pp. 99-111

For 28 years, the author has worked on machines that span three revolutions in supercomputer design: vector supercomputing, parallel supercomputing on multiple CPUs, and supercomputing on hierarchically organized clusters of microprocessors with cache memories. The author describes these revolutions from the perspective of numerical algorithms and the programs that implement them, and he looks toward the future and the coming of distributed shared-memory (DSM) and shared-memory multiprocessor (SMP) architectures.
The new architectures can combine the performance benefits of massively parallel computing with the flexibility of shared-memory multiprocessors. Like all supercomputer systems, however, these new machines will strongly favor certain numerical algorithms and force others to execute at much slower speeds. The usefulness of the favored algorithms and the ease with which they can be implemented, together with the Gflops/sec. or Tflops/sec. these algorithms the scientific output that is the true meaning of a supercomputer.