Echelon Neuron C - Manual (Page 136)

116 Functions
always use muldiv24() without loss of precision, compared to muldiv(), if neither
A nor B ever exceeds 256.
Syntax
#include <stdlib.h>
unsigned long muldiv24 (unsigned long A , unsigned int B ,
unsigned int C );
Example
#include <stdlib.h>
unsigned long a, d;
unsigned int b, c;
...
void f(void)
{
d = muldiv24(a, b, c); // d = (a*b)/c
}
muldiv24s( ) Function
The muldiv24s( ) function permits the computation of ( A * B )/ C where A is a 16-bit
value, and B and C are both 8-bit values, but the intermediate product of ( A * B ) is
a 24-bit value. Thus, the performance of the function is improved while
maintaining the accuracy of the result. There are two versions of this function:
muldiv24s( ) and muldiv24( ). The muldiv24s( ) function uses signed arithmetic,
while the muldiv24( ) function (see above) uses unsigned arithmetic.
See also muldiv( ) and muldivs( ) for functions which use 32-bit intermediate
accuracy for greater accuracy at the expense of slower performance. You can
always use muldiv24s() without loss of precision, compared to muldivs(), if either
A or B always is in the -128..+127 value interval.
Syntax
#include <stdlib.h>
signed long muldiv24s (signed long
A , signed int B , signed int C );
Example
#include <stdlib.h>
signed long a, d;
signed int b, c;
...
void f(void)
{
d = muldiv24s(a, b, c); // d = (a*b)/c
}