Be aware of library usage, Use more efficient data types – Echelon Neuron C User Manual

Page 215

Neuron C Programmer’s Guide

203

The Neuron C utility functions include byte-manipulation functions, such as

high_byte( ), low_byte( ), make_long( ), and swap_bytes( ). There are bit-
manipulation functions such as clr_bit( ), reverse( ), rotate_long_left( ),

rotate_long_right( ), rotate_short_left( ), rotate_short_right( ), set_bit( ), and

tst_bit( ).

For extended precision arithmetic, Neuron C provides the muldiv( ), muldivs( ),

muldiv24( ), and muldiv24s( ) functions. These functions permit a multiply
operation, followed by a divide operation, with the intermediate result and the

operations using either 32-bit or 24-bit precision.
The Neuron C functions also include such utilities as the timers_off( ) function.
This function turns off all application timers with a single function call. This

function call takes less space than the corresponding assignment of zero to a

single timer, although it takes longer to execute. Thus, if your program contains
a single application timer, and you turn it off by assigning zero to it, consider

using this function instead in order to save code space.
Other miscellaneous functions include bcd2bin( ) and bin2bcd( ), delay( ) and
scaled_delay( ), and random( ).
All of the Neuron C functions are described in detail, with examples, in the

Neuron C Reference Guide

Be Aware of Library Usage

Be aware of the system functions that are placed in application memory (see the

table in the

Neuron C Reference Guide

for a complete list of the functions placed

in memory for each chip and each version of firmware). If possible, avoid use of
such things as signed bitfields in structures that cause use of library functions.

Use More Efficient Data Types

The Neuron C compiler generates more compact code when the data items and
operations on them more closely match the underlying machine architecture and

instruction set. If possible, change variables to be local rather than global, to be

short rather than long, and to be unsigned rather than signed.

For example, consider the following function which finds an occurrence of value

in the array a and returns the index where value was located:

find(int a[], int value, count) {

for (i=0; i

if (a[i] == value) break;

}
return

}

When this function is compiled, the following code sizes are obtained

corresponding to the data types shown:

is signed short:

25 bytes

is unsigned short:

24 bytes

is signed long:

34 bytes

is unsigned long:

34 bytes