On Fri, 26 Jul 2002, Sternbach, William [IT] wrote:
> Also, I haven't seen any benchmark programs which calculate MIPS
> (Millions of Instructions per Second) for a computer.
The notion of "MIPS" in benchmarking is a misnomer. However, the
standard definition of "MIPS" is defined by the performance of the
Dhrystone benchmark on a VAX 11/780 computer, which produces a result
of 1757. "MIPS" certainly has nothing to do with floating point
You can learn more at
> My "C" program calculates the MIPS of a computer, which is of interest to
> people who write scientific or number crunching programs.
Interesting. I am not sure how you can calculate a "MIPS" value based
on your own benchmark. The "IPS" stands for Instructions Per Second,
which is a meaningless unit (unless you take it literally and count
machine instructions) since a processor may execute many more
instructions per second than another processor, yet accomplish less
Normally floating point performance is measured in "FLOPS" rather than
> Also, this program can be compiled and run on different platforms to compare
> speed of Unix Boxes, Apple Macintoshes, Intel based computers, Mainframe
> Cray Computers, ETC. So I thought the post would be of value to the group.
There is no shortage of benchmark programs. I think I trust standard
benchmarks such as from Spec (http://www.spec.org/) more than any