Re: NEC Timing

In the process of setting out benchmarks for calculation performance, two
concerns occur to me (and either or both may not be real considerations).

1. Different implementations of NEC-2 have been set up to use memory
differently. Will this affect performance figures? How will available
RAM affect calculation speeds, relative to different implementations?

2. In comparing different calculation machines (NEC-2, MININEC Pro,
etc.), will complexities in the card deck alter the relative speed of
calculation. The 299-segment dipole is simply a large file, but of
simple construction. Are there other files which can help evaluate the
performance of calculation machines on a broader basis?

Although initial results are suggestive, a more thorough study of what
makes a good benchmark test for various purposes seems in order. Most
test results, for example, have reported only the CPU chip and not other
factors of computer configuration and program implementation that may
affect results.

L. B. Cebik
Received on Fri May 31 1996 - 16:03:00 EDT