Question

I am trying to optimize some code, that seems simple but is giving me a hard time. So basically, I am trying to make a number the value 1, if it is greater than 0. The problem is that I don’t want to use any comparisons, as they are very expensive and getting a solution without a comparison will save me 40 seconds as it gets called alot. So all I want is bit wise operators, adding, subtracting, dividing and multiplying.

Extra: The number will ever only be 1 or 2.

The type is a unsigned int.

Full Algorithm:

        DWORD num = (blockNum / 0xAA) * blockStep[0];

        switch (blockNum / 0xAA)
        {
            case 0:
                return num + hashOffset;

            default:

                num += ((blockNum / 0x70E4) + 1) << (BYTE)packageSex;
                switch (blockNum / 0x70E4)
                {
                    case 0:
                        return num + hashOffset;

                    default:
                        return num + (1 << (BYTE)packageSex) + hashOffset;
                }
        }

Answer 1

This code snippet produces the effect you want. I checked the assembly output (MSVC++ 2010) and it’s branchless.

n = !!n;