baseConvert() & 32-bit ops

[Dar Scott]

On Sunday, June 22, 2003, at 10:07 PM, Mark Brownell wrote:

>     f = S1[a] + S2[b]
>     f = bitXor( f , S3[c] )
>     f = bitAnd( f + S4[d], -1 )

There are a couple problems, I think the same functionality is the 
cause though.

On the last line is the negative number -1.  It is increased to 0 in a 
bit op.

On the first line and the last line are numbers that could go over the 
32 bit range because of the carry.  They are decreased to the top of 
the range in the bit op if they do.

put (S1[a] + s2[b]) mod 2^32  into f -- only init some var for 2^32
put f bitXor S3[c] into f
put (f + S4[d]) mod 2^32 into f

If we had more than 32 bits to work with in bit ops we could mask 
instead of mod.

I don't know of a faster way than mod.  Anybody?

Maybe...
if f >= 2^32 then put f - 2^32 into f

I doubt that will be faster.

So your F function could be just 3 lines, then: the binaryEncode, the 
binaryDecode and the above 3 combined in the return.

function F halfBlock
    put binaryEncode(...
    put binaryDecode(...
    return ( (  ((S1[a] + S2[b]) mod 2^32) bitXor S3[c]  ) + S4[d] ) mod 
2^32
end F

Dar Scott