7-25
FLOATING-POINT UNIT
When stored in memory, the least significant byte of an FPU data-type value is stored at the
initial address specified for the value. Successive bytes from the value are then stored in succes-
sively higher addresses in memory. The floating-point instructions load and store memory oper-
ands using only the initial address of the operand.
7.4.1.Real Numbers
The FPUs three real data types (single-real, double-real, and extended-real) correspond directly
to the single-precision, double-precision, and double-extended-precision formats in the IEEE
standard. The extended-precision format is the format used by the data registers in the FPU.
Table 7-8 gives the precision and range of these data types and Figure 7-17 gives the formats.
For the single-real and double-real formats, only the fraction part of the significand is encoded.
The integer is assumed to be 1 for all numbers except 0 and denormalized finite numbers. For
the extended-real format, the integer is contained in bit 63, and the most-significant fraction bit
Figure 7-17. Floating-Point Unit Data Type Formats
0
Packed BCD Integers
79
D0
0
Long Integer
63
4 Bits = 1 BCD Digit
0
Short Integer
31
0
Word Integer
15
Sign
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
D13
D14
D15
D16
D17
787271
X
62
14
30
0
Extended Real
79
Sign
78
6463
0
Double Real
63
62
0
Single Real
31
30
2322
Fraction
Exp.
Sign
Implied Integer
Implied Integer
Sign
Exponent
Fraction
5251
Fraction
Exponent
62
Integer
Sign
Sign
Sign
