Path: utzoo!attcan!uunet!lll-winken!sun-barr!cs.utexas.edu!radar!banach!mishra
From: mishra@banach.ACA.MCC.COM (Aditya Mishra)
Newsgroups: comp.lang.c
Subject: floating point multiplication BUG in C (cc compiler)
Keywords: bug in C float multiplication
Message-ID: <1348@banach.ACA.MCC.COM>
Date: 19 Oct 90 19:16:22 GMT
Organization: MCC Human-Interface
Lines: 66

BUG !!!   BUG !!!   BUG !!!

THIS PROGRAM DEMONSTRATES WHAT I BELIEVE IS A BUG IN FLOATING POINT
MULTIPLICATION IN 'C' !! 

I ACCIDENTALLY STUMBLED ON A PATTERN OF
FLOATING POINT OPERAND VALUES FOR WHICH RESULTS ARE QUITE INACCURATE !!!

I WAS GETTING UNDESIRED BEHAVIOR IN A PROGRAM BECAUSE OF THIS (APPARENT) BUG.

I FELT IT MIGHT BE SERIOUS AND IMPORTANT TO ALL C USERS.

 
bug.c
-----

Aditya Mishra, MCC Austin  (512) 338 3481
-----------------------------------------

(1)   In the program, if float f = ... 0.64, 0.32, 0.16, 0.08, 0.04, 0.02, 
	0.01, 0.005, 0.0025, ..., etc (see the pattern?), the value 
	of i1 was one less than what it should be. 
      This error was absent if the value of f was not 
        from the above pattern (if f = 0.65, e. g.).

(2)   Also, if float g = 1.0025, 1.32 or 9.0025 and some more values, 
        the result in float r2
	(r2 = factor * g) was wrong after 2 or 3 decimal places.

(3)   These bugs persist on cc (C compiler) on about six machines 
	that I could get my hands on.

(4)   I am not sure if this is really a bug. I am just expressing an
        experience that teaches me to be careful with floating point
        results of multiplication.
 
*/

/*--	--	--	--	--	--	--*/

#include <stdio.h>

main()
{
	int i1, i2;
	float r1, r2;

	float g = 9.0025;
	float f = 0.32;

	float factor = 10000.0;
 	
	i1 = factor * f; /* *even if*  i = (int) (factor * f); */
	r1 = factor * f;

	i2 = factor * g; /* *even if*  i = (int) (factor * g); */
	r2 = factor * g;
	
	printf("%d   ", i1);
	printf("%f \n", r1);

	printf("%d   ", i2);
	printf("%f \n", r2);
}

/*--	--	--	--	--	--	--*/