From: krishna.myneni@ccreweb.org   
      
   On 4/18/24 07:09, Stephen Pelc wrote:   
   ...   
   > In retrospect, if I were doing this again I would standardise on an   
   > external double-double library (about 106 bits). In most cases that   
   > we encounter, the desire for 387 FP is to gain the extra precision.   
   > Since very few CPUs support quad precision natively, the most   
   > obvious solution is a double-double library.   
      
   I also thought double-double would be sufficient, until last year, when   
   I ran into having to code an application for which it was unsuitable.   
   The problem with double-double was not in the number of bits in the   
   significand (106) -- the problem was that the exponent range of double   
   precision was not large enough. I believe the double-double type even   
   reduces the available exponent range from that of a double type. In any   
   case I ended up using the MPFR library for this application. IEEE quad   
   precision would be more suitable, since it expands the exponent range.   
      
   --   
   Krishna   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)