MIN < EXPONENT < MAX I SIGNIFICAND = ANY BIT PATfERN
SIGN OF SIGNIFICANO, 0 OR 1
EXPONENT = O [
SIGN OF SIGNtFICAND. 0 OR 1
SIGNIFICAND = ANY NON-ZERO BIT PATTERN
EXPONENT = 0 ] SIGNIFICANO = 0
SIGN OF SIGNIFICAND, 0 OR 1
J
l EXPONENT = MAXIMUM J SIGNIFICANO = O"
I
SIGN OF SIGNIEICANO, O OR 1
I[ EXPONENT = MAXIMUM ( SIGNIFICANO = ANY NON-ZERO BIT PA'I'rERN
l-- SIGN OF SIGNIFICAND. 0 OR 1
*For the extended precision format, the moat significant bit of the significand (the integer bit) is a don't care.
Figure 3-8. Binary Real Data Type Summary
91 80 67
~XPONENT
[ [ 1 1 [ 1
,~PUCIT OEaMAl ~01~
2 BITS. USED ONLY FGR _+INFINITY OR NAN(S), ZERQ OTHERWISE
SIGN OF EXPONENT
-- SIGN OF MANTISSA
0
I'7-oIG,'T ', I I ; = " I
PACKEBOECIMAL,E
M,ANT~SS, A I I I I I I
*Unless a binary-to-decimal conversion overflow occurs
Figure 3-9. Packed Decimal Real Data Format
3.4 INTERNAL DATA FORMAT
AI] FPCP internal operations are performed in extended precision. All external operands,
regardless of data format, are converted to extended precision values before the specified
operation is performed,
The format used in the eight floating-point data registers is identical to the extended
precision data format described previously and in 3.6 DATA FORMAT DETAILS (with the
MC68881/MC68882 USER'S MANUAL FREESCALE
3-7
To Next Page
Loading...