On an Arm9 processor running Debian the following program gives the results
indicated...
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
int main()
{
double in, out;
in = 2.0;
out = sqrt(in);
printf("%f\n", out); // 2.0000
printf("%f\n", sqrt(in)); // 1.414214
printf("%f\n", sqrt(2.0)); // 1.414214
return 0;
}
It works for most values but some obvious ones, all of which seem to be powers
of two, fail though not all powers of two.
This also seems to affect other floating point operations.
--
Summary: Floating point errors on Arm9 processor
Product: gcc
Version: 4.1.2
Status: UNCONFIRMED
Severity: major
Priority: P3
Component: c
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: mp2 at ladybridge dot com
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39085
