roldan@ncar.UCAR.EDU (Roldan Pozo) (07/07/89)
Several days ago, the following message appeared announcing a performance paper on the CM: >We've just submitted a paper, "Performance Charateristics of Scientific >Computations on the Connection Machine 2", which has explicit timings >for floating-point operations, and basic scientific operations. We have >also included benchmarks of a real application ---weather forecasting >using a primitve equation model. >The paper first presents a breif overview of the CM architecture and >hardware, and how to utilize it for best GFLOP performance. I can >send a copy of it if you'd like. This message was originally an email reply to a previous net posting requesting references on CM papers. Apparently, I mailed this reply to the wrong place and it inadvertently wound up on the net; it was not intended as a net posting. Since then, we have received about 70 requests for copies of the paper, mainly from readers of this newsgroup. Although we are quite happy to see such interest, we were not prepared for such a demand. We will gladly send copies to all who wrote in, but please give us a couple of days to service your request. Although e-mail is the best form of distribution, the paper contains several charts and figures in inconsistent formats, thus making distribution through e-mail impractical. Thus, to those people who sent only their e-mail address, we need your US mail address also. Thanks for your interest, Roldan Pozo Center for Applied Parallel Processing Campus Box 430 University of Colorado Boulder, CO 80309-430