U++ forum: Welcome to the forum

Re: New hash folding function... [message #13686 is a reply to message #13684]

Tue, 22 January 2008 08:55

gprentice
Messages: 260
Registered: November 2005
Location: New Zealand

Experienced Member

Your new hash-fold function seems similar to the last step of that hash function on Wikipedia

hash += (hash << 3);
hash ^= (hash >> 11);
hash += (hash << 15);

Is this a standard type of thing to do after the byte-by-byte bit even when you're not saving the pre-masked result. Is there a mathematical basis for this "extra" bit (like CRC) or was it just found by trial and error?

Why do you call it "hash-fold" if it doesn't actually reduce the range of values
- or if
return (h >> 23) - (h >> 9) - (h >> 15) - h;
does reduce the range of values, how does it do that, since it's unsigned arithmetic?

Graeme

Report message to a moderator