Control Data Corporation 3300 Reference Manual page 75

60229400
The binary point is assumed
to
be immediately
to
the right of the lowest
order bit (00). The upper bit of any fixed number designates the sign of the
coefficient (23 low order bits).
Sign Bit
1
o
Coefficient
Negative
Positive
COMPUTATIONAL-2 describes an item in floating point format. Such an
item occupies two computer words (48 bits) and is stored as follows:
4746 35
II
Exponent
Coefficient
Sign bit
The coefficient consists of a 36-bit fraction in the low order positions of the
floating point word. The coefficient is a normalized fraction; it is equal to
or greater than 1/2 but is less than 1. The highest order bit position (47)
is occupied by the sign bit of the coefficient.
Sign Bit
1
o
Coefficient
Negative
Positive
The floating point exponent is expressed as an II-bit quantity with a value
ranging from 0000 to 37778.
It
is formed by adding a true positive exponent
and a bias of 2000 8 or a true negative exponent and a bias of 1777 8 . This
results in an effe~tive exponent modulus of ±1023
10
.
For further information concerning fixed point and floating point arithmetic,
refer to the 3300 Computer Systems Reference Manual, Pub. No. 60157000.
The only meaningful clauses to be used with COMPUTATIONAL-n are the
VALUE and REDEFINES clauses. COMPUTATIONAL-n items are
synchronized left.
USAGE IS INDEX describes an elementary item as an index data item. An
index data item is always one computer word in length in binary format;
USAGE IS INDEX has the same effect as USAGE IS COMPUTATIONAL-I.
An index data item is referred to by a SEARCH or a SET statement. It
contains the character address bias of a table element. (The character
address bias is the value added to the base character address of a table to
00
give the character address of a particular element in the table.) An index
data item can also appear in a relation condition, but it cannot be a conditional
variable.
3-41