Zilog EZ80F916 - Manual (Page 166)

UM014423-0607 Using the ANSI C-Compiler
ZiLOG Developer Studio II
eZ80Acclaim! ® User Manual
146
Basic Fractional Arithmetic
The compiler supports signed/unsigned fractional arithmetic for the +, -, and * operators.
The binary operators +, -, and * on the
fract
data type give correct results only if both
the operands are of same size and of same sign-type. Otherwise, the values are distorted
for the reasons cited in “Assigning Values to fract Variables” on page 145.
All other arithmetic operators on
fract
treat the data as if they were of integral types
(ignoring the
fract
keyword), and hence the resulting value might not make sense within
the context of
fract
.
Bitwise Logical and Shift Operators
The compiler supports all bitwise logical operators, including &, |, ^, <<, and >> for frac-
tional types, although ANSI does not allow these operations for floating-point types. You
can use these operators to vary the position of the binary point in a fractional number. This
feature can be very useful when the result of a fractional expression is not within the range
of –1 to +1 or 0 to +1 but a wider range such as –3 to +3.
Scaled Fractional Arithmetic
You can alter the range of a fractional fixed-point value by modifying where its implied
binary point resides. You can perform this operation using the shift operators as follows:
unsigned short fract f1,f2;
int int_part;
unsigned short fract fract_part;
f2 = .55;
f2 >>= 3; /* move the binary point to right */
f1 = f2 + f2; /* .55 + .55 = 1.1 */
int_part = (f1 & 0xe000) >> 13; /* 1 */
fract_part = f1 << 3; /* 0.1 */
Although
fract
arithmetic assumes that the binary point is just after the most significant
bit (or after the sign bit in the case of a signed type), you can perform the same operations
on
fract
s that have been shifted to move the binary point.
NOTE: However, be careful when dealing with negative
signed fract
s and make sure that all
values involved in an operation have their binary points aligned.
Char and Short Enumerations
The enumeration data type is defined as int as per ANSI C. The C-Compiler provides
language extensions to specify the enumeration data type to be other than int to save
space. The following syntax is provided by the C-Compiler to declare them as char or
short :
enum
{
RED = 0,
YELLOW,