FLOPS requirements

[Joel Sherrill <joel@OARcorp.com>]

Mark Vander Voord wrote:
> I had thought of a doing a timing-based solution like many of you have
> suggested, but I don't think it is really what I am looking for. 
> Since I will be switching micros already, a measurement of how long
> the algorithm takes without hardware floating point support doesn't
> interest me as much as determining what my overall flop requirement
> will be so that I know what part to choose.  I'm not really sure how
> that can be measured, but I think Greg's suggestion of using objdump
> and grep might get me closest.

If you can isolate the code down to some key subroutine, then it should 
be possible to construct a benchmark using some canned data.  That 
benchmark can be run on your current hardware, PCs of varying speeds,
and the RTEMS simulator BSPs which are timing accurate.

If the code can be shared, then some nice sole might be willing to
run it on hardware similar to what you are considering.

> Thanks for the advice, everyone.  If anyone has anything else to add,
> please do so!  =)
> 
> --
> Mark S VanderVoord
> Self-Guided Systems, LLC


-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
    Support Available             (256) 722-9985