On the following program, gcc -O2 creates incorrect code: #include <stdio.h>
#define NUM 10000 #define MAX 10.0 __inline__ float rsqrt(float x) { float xhalf = 0.5f * x; int i = *(int*)&x; i = 0x5f375a86 - (i>>1); x = *(float*)&i; x = x * (1.5f - xhalf * x * x); return x; } int main(void) { int i, j; float x, y; for (i=0; i<NUM; i++) { x = (i+1)*MAX/NUM; y = rsqrt(x); fprintf(stderr,"%10.6f -> %10.6f\n",x,y); } } The same code works as expected with gcc 3.3.6. Interestingly, if "fprintf(stderr," is changed to "printf(" the output changes significantly, but is still incorrect. -- Summary: gcc -O2 creates erroneous code Product: gcc Version: 4.0.2 Status: UNCONFIRMED Severity: normal Priority: P3 Component: tree-optimization AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: steger at mvtec dot com GCC build triplet: i686-pc-linux-gnu GCC host triplet: i686-pc-linux-gnu GCC target triplet: i686-pc-linux-gnu http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26214