From: jgd@cix.co.uk   
      
   In article <10mq853$106co$1@paganini.bofh.team>, antispam@fricas.org   
   (Waldek Hebisch) wrote:   
      
   > Note that IBM floating point format effectively lost about 3 bits of   
   > accuracy compared to modern 32-bit format.  I am not sure how much   
   > they lost compared to IBM 7090 but it looks that it was at least 5   
   bits.   
      
   It's somewhat worse than that. Because the mantissa is in whole hex   
   digits, accuracy is lost in 4-bit lumps during a calculation. And because   
   normalisation is of whole hex digits, and Barnard's Law applies, accuracy   
   and its loss are quite data-dependent.   
      
   > But it does not mean that 36-bits are somewhat magical.   
      
   Definitely not.   
      
   Quadi, have your computer architectures included IBM 360 floating point   
   support? There is probably more demand for that than for 36-bit these   
   days.   
      
   John   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)