This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: /internet



  In message <199812170055.QAA14233@cygnus.com>you write:
  > > The rules for FP are different becuase it's not 100% certain that the
  > > results after reassociation will be the same as before reassociation.  Th
  > ough
  > > I believe in the case reassociating a series of multiplies we are safe.
  > 
  > How about:
  > 	x = "semi-Infinity"; /* something that when squared will overflow */
  > 	(x * x) * (1/x) /* +Infinity */
  > 	x * (x * (1/x)) == x * 1.0 (approximately) == x;
Ah.  Shit.  That's what I was missing.  

Bummer.  This may severely restrict what we can do in this space for floating
point since we can't use the as-if rule.  Oh well...  

jeff


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]