Electromagnetics at Washington State University

Abstract

A question arises: Should the Given updating be written i OCCAM, to get an even faster code? A simple estimate shows that in he test program, about 40 million floating-point operations were performed in 58 sec on a single transputer. This is approximately 46% of the theoretical speed. In the case of the conjugate gradient method [8], 39% of the theoretical speed has been reached with OCCAM. This, and a comparison with the test of the floating-point operations (Table l), indicates that one will probably not get a faster code with OCCAM in our case.