machine precision,

Hi,

I found an algorithm for determining machine precision that can be described
as follows. Set x = 0.5. Then if x + 1 > 1, repeat the above but with x =
x * 0.5. So basically, we keep shrinking x until it is so small that adding
it to 1 still results in 1 as far as the computer is concerned. Then x is
the machine precision.

I was wondering, however, if one knows the floating-point format, can one
just use bit operations and pick some floating point number and then add 1
to the least significant digit in the mantissa to see what the next largest
number is. Then the difference between the two is the machine precision?


.

Relevant Pages

  • Re: machine precision,
    ... I found an algorithm for determining machine precision that can be described ... we keep shrinking x until it is so small that adding ... I was wondering, however, if one knows the floating-point format, can one ...
    (sci.math.num-analysis)
  • Re: machine precision,
    ... I found an algorithm for determining machine precision that can be described ... we keep shrinking x until it is so small that adding ... I was wondering, however, if one knows the floating-point format, can one ... where are tricky optimizing compiler which "simplify" ...
    (sci.math.num-analysis)
  • Re: machine precision,
    ... but it is dependend on the floating-point format of the computer you use. ... vsgdp napsal: ... I found an algorithm for determining machine precision that can be described ... I was wondering, however, if one knows the floating-point format, can one just use bit operations and pick some floating point number and then add 1 to the least significant digit in the mantissa to see what the next largest number is. ...
    (sci.math.num-analysis)
  • Re: machine precision,
    ... I found an algorithm for determining machine precision that can be described ... we keep shrinking x until it is so small that adding ... I was wondering, however, if one knows the floating-point format, can one ... to the least significant digit in the mantissa to see what the next largest ...
    (sci.math.num-analysis)