http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48906
Summary: Wrong rounding results with -m32
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: libfortran
AssignedTo: [email protected]
ReportedBy: [email protected]
This PR is to carry forward the remaining issue identified in PR48602 comment
#37
Here is a test case showing the problem.
print "(ru,g15.2)", .099d0 ! > 0.10E+00 expected 0.10
print "(rc,g15.1)", .095d0 ! > 0.1E+00 expected 0.1
print "(rc,g15.2)", .0995d0 ! > 0.10E+00 expected 0.10
print "(ru,g15.3)", .0999d0 ! > 0.100E+00 expected 0.100
end
$ gfc bug.f03
$ ./a.out
0.10
0.1
0.10
0.100
$ gfc -O1 bug.f03
$ ./a.out
0.10
0.1
0.10
0.100
$ gfc -m32 -O1 bug.f03
$ ./a.out
0.10E+00
0.1E+00
0.10E+00
0.100E+00
$ gfc -m32 bug.f03
$ ./a.out
0.10E+00
0.1E+00
0.10E+00
0.100E+00
The problem is rooted in excess precision and that we use sensitive floating
point operations to determine the formatting parameters. I am seeking a
solution that will avoid using floating point to do this.
