Speed ​​up a program with multiple processors

I found that sometimes it’s faster to split one cycle into two or more

for (i=0; i<AMT; i++) {
    a[i] += c[i];
    b[i] += d[i];
}
     ||
     \/
for (i=0; i<AMT; i++) {
    //a[i] += c[i];
    b[i] += d[i];
}
for (i=0; i<AMT; i++) {
    a[i] += c[i];
    //b[i] += d[i];
}

On my desktop, win7, AMD Phenom (tm) x6 1055T, the two-loop version is faster by about 1/3 of the time less.

But if I am dealing with an appointment,

for (i=0; i<AMT; i++) {
    b[i] = rand()%100;
    c[i] = rand()%100;
}

dividing the assignment of b and c into two loops is not faster than in one cycle.

I think there are some rules for using the OS to determine if certain codes can be executed by multiple processors.

I want to ask if I guess correctly, and if I am right, what are the rules or cases when several processors will be automatically (without thread programming) used to speed up my programs?