Справочник Пользователя для Siemens 500 NIM

 

Floating Point Numbers

The exponent portion of the IEEE format is similar to the IBM format
except that it is 8 bits in length and is biased by 7F

.  It is also different in

that it represents a power of 2 instead of 16.  The exponent for 2

  is

represented in memory by 7F

.  Negative exponents are represented by

numbers less than 7F

.  For example, a 2–  is represented in the

exponent field by by a value of 7E

.  The exponents may be any value from

–7E

  to +7F  (2–   to 2 ).  The eight exponent bits are stored in

bits 1 – 8 of the first memory word. See Figure B-2.

The fraction portion of the IEEE format is similar to the IBM format. It is a
hexadecimal fraction which is interpreted in one of many ways, depending
on the value of the exponent.  Table B-1 describes the different types of
conversions for the IEEE format.





-



 

"

"F

''

''

::HH''

::

::

61
61

61
61

61

61

'

*G

:

!

*G

:

!8 """9= !
8 9

Q,*

8*69J?J86+ 9

8*69J?J8:+ 9

Q,*:89




8*6?C9

*6?1

<=N = N=F

1

Case A is an illegal value.  A Not-a-Number (NaN) will never be returned
from the Series 500 NIM.  If this example is written to the NIM, an
exception will be returned.

Case B, +/– infinity, is used to represent numbers too large in magnitude to
be represented with the IEEE format.

Case C represents a normalized IEEE floating point value.  (There is a one,
not included in f, to the left of the radix point.)

Case D represents a denormalized IEEE floating point (contains a zero to
the left of the radix point).

Case E represents a value of zero.  Numbers too small to be represented in
the IEEE format will be reported as a positive or a negative 0.