In article <87a4wnayix.fsf@nightsong.com>,   
   Paul Rubin   wrote:   
   >Stephen Pelc  writes:   
   >> The application is construction planning. Sometimes over 10 currencies   
   >> are used. The app must allow for a total cost of well over 10 billion   
   >> dollars US.  They tried 80 bit FP, but the difference in dollars given   
   >> by an 128 bit integer calculation (effectively a huge spreadsheet) and   
   >> an 80 bit floating point version was 10 million dollars for capping   
   >> the piles in the new Hong Kong airport.. There are many examples in   
   >> construction of dealing with two quantities, one of which is large and   
   >> the other small. Capping piles is one of these.   
   >   
   >I wonder if this was due to some kind of numerical instability in which   
   >case 128 bits might not be enough either.  80 bit FP has a 53 bit   
   >mantissa and 2**53 pennies is 9e15 dollars.  Did you try the same   
   >calculation with say 256 bit FP, bignums, or whatever?   
      
   Financial calculations donot like floating point. If you started with 32 bits   
   integers, stepping up to quad precision is a smaller step, then introducing   
   fp.   
      
   Groetjes.   
   --   
   The Chinese government is satisfied with its military superiority over USA.   
   The next 5 year plan has as primary goal to advance life expectancy   
   over 80 years, like Western Europe.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)