The following program: int main() { long double d, e; d = 1.0; while ((e = d / 2.0) != (long double) 0.0) d = e; return 0; } loops forever if it is compiled without optimization: spe149.testdrive.hp.com> /tmp/make/bin/gcc test4.c -mfp-rounding-mode=d -mieee-with-inexact spe149.testdrive.hp.com> time ./a.out ^C0.405u 4.378s 0:04.79 99.5% 10+174k 0+0io 0pf+0w spe149.testdrive.hp.com> /tmp/make/bin/gcc test4.c -mfp-rounding-mode=d -mieee-with-inexact -O2 spe149.testdrive.hp.com> time ./a.out 0.000u 0.004s 0:00.00 0.0% 0+0k 0+0io 0pf+0w spe149.testdrive.hp.com> /tmp/make/bin/gcc -v Reading specs from /tmp/make/lib/gcc/alpha-unknown-freebsd5.2.1/3.4.2/specs Configured with: ./configure --prefix=/tmp/gcc : (reconfigured) ./configure --prefix=/tmp/gcc : (reconfigured) ./configure --prefix=/tmp/make --enable-languages=c Thread model: posix gcc version 3.4.2 The same problem appears with GCC 3.3.3 [FreeBSD] 20031106
Well it should as you used -mfp-rounding-mode=d Dynamic rounding mode. A field in the floating point control register (fpcr, see Alpha architecture reference manual) controls the rounding mode in effect. The C library initializes this register for rounding towards plus infinity. Thus, unless your program modifies the fpcr, d corresponds to round towards plus infinity. Since you are rounding towards postive infinity you will never reach zero.