Issue No. 05 - September/October (2002 vol. 4)
Geoffrey Fox , Indiana University
<p>Over the past decades,the computational science community has debated the best architecture for parallel computing.sometimes it's distributed memory,sometimes single instruction, multiple data (synchronous as in Thinking Machines' CM-1 and CM-2). Sometimes it's multiple instruction, multiple data, as in networked computers; or shared memory; or vector nodes;or multi-threaded. And at times, it's more or less all of these. The debate recently perked up when the Japanese Earth Simulator supercomputer achieved 40 Tflops using a slightly heretical architecture. The arguments are accompanied by a related discussion as to the appropriate parallel computing model. </p>
G. Fox, "Message Passing: From Parallel Computing to the Grid," in Computing in Science & Engineering, vol. 4, no. , pp. 70-73, 2002.