help !!

[Paul Pham <pham@xmission.com>]

sorry, need helps on another topic: representation of
floating point in Digital computer ..

assume real-number is represented as 32-bits as follows:

   -------------------------------------------------
      fraction                       ! exponent
   -------------------------------------------------
         24 bits for fraction         8 bits for exponent

the 1st bits of fraction and exponent is sign bit

the floating point must be normalized ;ie that the first bit and second
bit in the fraction must be of opposite sign. for example

positive number   0 1 .......
negative number   1 0 .......

so a real number is computed as  +- a * (2 ** b)
   where  0.5  <=  abs(a) <= 1.0  (abs = absolute value)
   and    -128 <= b <= 127

question: given a real number, need an alogorithm to convert it
          into the above machine reprensentation.
PDP


hi all,

assume: negative integer is represented in digital computer
        as 2 complement and assume 4 bits computer

 so  -3 looks like:  1101 
             and     1101 = 13 as a positive number (forget the sign bit)
                     13 = 16 - 3

so in general if we have n-bits computer, a negative integer number
b in 2 complement is 2**n + b (in postive representation).

the above statement is intuitively true.
Comment anyone ???.

PDP