https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106438
--- Comment #6 from Jeevitha <jeevitha at gcc dot gnu.org> --- Interestingly, the issue occurs only with _Float128 and not with __float128. Here's the test case without the issue, [jeevitha@ltcden2-lp1 nb_gcc_106438]$ cat t1.c #include <stdio.h> typedef __float128 GFC_REAL_17; volatile GFC_REAL_17 a; volatile GFC_REAL_17 b; int main() { GFC_REAL_17 sum = a + b; printf("a + b = %Lf\n", sum); return 0; } [jeevitha@ltcden2-lp1 nb_gcc_106438]$ /home/jeevitha/gcc_clean/build_gcc/nb_gcc_106438/./gcc/xgcc -B/home/jeevitha/gcc_clean/build_gcc/nb_gcc_106438/./gcc/ -c t1.c -mabi=ieeelongdouble -O3 -o t1.o -flto [jeevitha@ltcden2-lp1 nb_gcc_106438]$ /home/jeevitha/gcc_clean/build_gcc/nb_gcc_106438/./gcc/xgcc -B/home/jeevitha/gcc_clean/build_gcc/nb_gcc_106438/./gcc/ -flto t1.o -mabi=ibmlongdouble -o t1.out [jeevitha@ltcden2-lp1 nb_gcc_106438]$