home bbs files messages ]

Forums before death by AOL, social media and spammers... "We can't have nice things"

   comp.arch      Apparently more than just beeps & boops      131,241 messages   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]

   Message 130,397 of 131,241   
   EricP to Thomas Koenig   
   Re: Tonights Tradeoff - NaN boxed precis   
   29 Nov 25 15:08:05   
   
   From: ThatWouldBeTelling@thevillage.com   
      
   Thomas Koenig wrote:   
   > Kent Dickey  schrieb:   
   >   
   >> This is my point: I don't see a great way to encode the first NaN, which   
   >> is why I propose not making that a goal.  You're not getting the first   
   >> NaN in any case even if you try to do so in hardware, since the order of   
   >> operations is a fragile thing that's hard to control unless you write   
   >> assembly code, or the most tedious source code imaginable.   
   >   
   > Using Fortran, parentheses have to be honored.  If you write   
   >   
   >  A = (B + C) + (D + E)   
   >   
   > then B + C and D + E  have to be calculated before the total sum.   
   > If you write   
   >   
   >   A = B + (C + (D + E))   
   >   
   > then you prescribe the order completetely.   
   >   
   > I can imagine source code that is much more tedious than this :-)   
      
   That doesn't control which variable is assigned to each source operand.   
   If both operands were Nan's and the two-Nan-rule was "always take src1"   
   then the choice of which to propagate would still be non-deterministic.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]


(c) 1994,  bbs@darkrealms.ca