>>106495382
even at -O1 it optimizes to similar assembly. the only difference being the loop counter,
you can get identical assembly by changing it to
for (int* a = arr; a != arr + cnt; ++a)
s += *a;


op using -O0 and then complaining about unoptimized code