The floating-point preferred exponent of a decimal operation in the IEEE 754-2008 standard is the operation-dependent target value of the exponent used to select a floating-point representation of an exact result. If that value cannot represent the result in the destination format, the nearest possible exponent is used.
Writing 
for the exponent
of a finite operand 
, the preferred exponent
is 
for addition and subtraction,
for multiplication, and 
for division (IEEE Computer
Society 2008, pp. 18 and 21). Thus preserving the operands' quantum
means respecting these operation-dependent rules, rather than keeping each operand's
quantum unchanged. For example, 
has preferred exponent 
,
although each operand has exponent 
.
