Notes on Floating Point Representations

There are several forms of floating point representation out there. This page is not endeavouring to catalogue every floating point storage format, but those which are relevant to the architectures spoken of in other sections of this website—mostly relating to Intel 64, VAX, and PDP-11 architectures.

Types of Floating Point

The following table is a summary of various floating point representations on several different platforms:

Floating point representations
Type Representation of 1.0
IEEE 754 128-bit X'00 00 00 00 00 00 00 00 00 00 00 00 00 00 FF 3F'
Intel 80-bit X'00 00 00 00 00 00 00 80 FF 3F'
IEEE 754 64-bit X'00 00 00 00 00 00 F0 3F'
IEEE 754 32-bit X'00 00 80 3F'
VAX G-float X'10 40 00 00 00 00 00 00'
VAX D-float X'80 40 00 00 00 00 00 00'
VAX F-float X'80 40 00 00'

The frep Program

The following is a listing the frep program, which was used to obtain the representations in the table in the previous section.

/* frep.c */
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

int
main(int argc, char *argv[]) {
        char buf[BUFSIZ];
#ifndef __vax__
        __float128 q;
        __float80 e;
#endif
        double d;
        float f;
        int i;

        fprintf(stdout, "FREP is running on a ");
#ifdef __vax__
        fprintf(stdout, "VAX");
#else
        fprintf(stdout, "Intel 64");
#endif
        fprintf(stdout, " system.\n");

#ifndef __vax__
        q = 1.0;
        e = 1.0;
#endif
        d = 1.0;
        f = 1.0;

#ifndef __vax__
        memset(buf, 0, BUFSIZ);
        memcpy(buf, &q, sizeof(q));
        fprintf(stdout, "REAL*16:");
        for (i = 0; i < sizeof(q); ++i)
                fprintf(stdout, " %02X", buf[i] & 255);
        fprintf(stdout, "\n");

        memset(buf, 0, BUFSIZ);
        memcpy(buf, &e, sizeof(e));
        fprintf(stdout, "REAL*10:");
        for (i = 0; i < sizeof(e); ++i)
                fprintf(stdout, " %02X", buf[i] & 255);
        fprintf(stdout, "\n");
#endif

        memset(buf, 0, BUFSIZ);
        memcpy(buf, &d, sizeof(d));
        fprintf(stdout, " REAL*8:");
        for (i = 0; i < sizeof(d); ++i)
                fprintf(stdout, " %02X", buf[i] & 255);
        fprintf(stdout, "\n");

        memset(buf, 0, BUFSIZ);
        memcpy(buf, &f, sizeof(f));
        fprintf(stdout, " REAL*4:");
        for (i = 0; i < sizeof(f); ++i)
                fprintf(stdout, " %02X", buf[i] & 255);
        fprintf(stdout, "\n");

        exit(EXIT_SUCCESS);
}

Note, that on VAX, to use the G-floating point format, you must compile with the -mg flag.



Valid XHTML 1.0 Strict Valid CSS!
Fave icon Database CMYK © 2008 – Barry Mieny
Syntax highlighting using highlight.js by © 2006 – Ivan Sagalaev