NEC-LIST: Single vs Double Nec2

Hi all,
I have haven't been following this list for a while and had a quick
look back through the archive but I can't find anything on this, so
please forgive me if it has been asked lots of times before.
 
Has anybody done any work on the relative pros and cons of the single
vs double precision versions of Nec2, in particular in a PC ( ie.
single IEEE) environment. I assume that single would be faster, and
have less significant figures, but has anyone found that for example in
some models single doesn't converge, or gives wildly different answers
to double? On a similar theme has anyone looked at trade offs between
precision and number of segments, ie. for a given amount of system
resource ( CPU time, memory etc.) I would assume you could fit more
segments in single than you could in an equivalent double, which would
be the better (more efficient?) result?

Thanks
Paul McMahon
VK3DIP

-- 
The NEC-List mailing list <nec-list_at_gweep.ca>
http://www.gweep.ca/mailman/listinfo.cgi/nec-list