RS6000 REAL*16

Bob Alverson uunet!tera.com!bob
Fri Dec 13 08:46:38 PST 1991


Do I have old software, or does the RS/6000 really mangle its
real*16 infinities?  For example, inf*1 gives Nan(!) using IBM's
"doubled" precision.  I guess Tera's real*16 (doubled precision as
well) won't match IBM's.

Bob




More information about the Numeric-interest mailing list