This example uses two implementations of Euclid’s Greatest Common Divisor (gcd) algorithm. It demonstrates how you can use conditional execution to improve code density and execution speed. The detailed analysis of execution speed only applies to an ARM7™ processor. The code density calculations apply to all ARM processors.

In C the algorithm can be expressed as:

int gcd(int a, int b)
{
    while (a != b)
      {
        if (a > b)
            a = a - b;
        else
            b = b - a;
      }
    return a;
}

You can implement the gcd function with conditional execution of branches only, in the following way:

gcd     CMP      r0, r1
        BEQ      end
        BLT      less
        SUB      r0, r0, r1
        B        gcd
less
        SUB      r1, r1, r0
        B        gcd
end

Because of the number of branches, the code is seven instructions long. Every time a branch is taken, the processor must refill the pipeline and continue from the new location. The other instructions and non-executed branches use a single cycle each.

By using the conditional execution feature of the ARM instruction set, you can implement the gcd function in only four instructions:

gcd
        CMP      r0, r1
        SUBGT    r0, r0, r1
        SUBLT    r1, r1, r0
        BNE      gcd

In addition to improving code size, this code executes faster in most cases. Table 2.3 and Table 2.4 show the number of cycles used by each implementation for the case where r0 equals 1 and r1 equals 2. In this case, replacing branches with conditional execution of all instructions saves three cycles.

The conditional version of the code executes in the same number of cycles for any case where r0 equals r1. In all other cases, the conditional version of the code executes in fewer cycles.