https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92768
--- Comment #1 from Martin Liška <marxin at gcc dot gnu.org> --- clang seems to be happy with -ffast-math: $ clang -O2 matrix.c -ffast-math && ./a.out matrix.c:17:63: warning: incompatible pointer types passing 'int const[4]' to parameter of type 'const float *' [-Wincompatible-pointer-types] return (graphene_simd4f_t) _mm_xor_ps((r1_sum), _mm_load_ps(__pnpn.f)); ^~~~~~~~ /usr/lib64/clang/9.0.0/include/xmmintrin.h:1723:26: note: passing argument to parameter '__p' here _mm_load_ps(const float *__p) ^ 1 warning generated.