AN669

DS00669A-page 4

© 1997 Microchip Technology Inc.

USING 32-BIT FLOATING POINT

MULTIPLY

Using the 32-bit floating point multiply supplied with

AN575 in a C program is straightforward. First, copy the

entire routine from the file fpm32.a16 (from AN575).

Then, create a function with the same name as the

assembly routine.

Lets take a well known formula:

Let,

Find A:

A

πr

2

=

We need to convert the previous decimal numbers to

Microchip 32-bit floating point. Use fpm32 (from

AN575), to solve the equation. We will use MPLAB-C

and use our C function named fpm32(). The main

routine is listed in Example 6.

AN575 comes with a handy utility called fprep.exe.

This Microchip file is a DOS executable. When running

fprep

, you can enter in a decimal number and it dis-

plays the hexadecimal floating point number. Table 1

shows the numbers in our example and their equivalent

floating point formats.

TABLE 1:

PICmicro

™ 32-BIT FLOATING POINT REPRESENTATIONS OF OUR EXAMPLE

EXAMPLE 6:

MAIN ROUTINE TO TEST OUT OUR NEW 32-BIT FLOAT MULTIPLY IN C

#include "16c74a.h"

#include "math16c.c"

#include "fpm32.inc"

// Notice that fpm32 is located in page 0

// Thus, all GOTOs reside in the same page.

void main (void)

{

AEXP

= 0X80;

// PI = 3.141592654

AARGB0 = 0X49;

AARGB1 = 0X0F;

AARGB2 = 0XDB;

BEXP

= 0X82;

// r = 12.34567898

BARGB0 = 0X45;

BARGB1 = 0X87;

BARGB2 = 0XE7;

fpm32();

// AARG =

PI * r

// you must reload r into BARG since

// fpm32() destroys BARG.

BEXP

= 0X82;

// r = 12.34567898

BARGB0 = 0X45;

BARGB1 = 0X87;

BARGB2 = 0XE7;

fpm32();

// AARG = (PI*r)*r

while(1);

}

Microchip Floating Point Equivalent

Decimal Number

EXP

B0 (MSB)

B1

B2 (LSB)

0x80

0x49

0x0F

0xDB

r = 12.34567898 meters

0x82

0x45

0x87

0xE7

0x87

0x6F

0x6A

0x07

using MPLAB 3.12 and PICMASTER 16J probe

0x87

0x6F

0x6A

0x07