Author: James Swafford
Date: 13:43:28 01/15/00
Go up one level in this thread
On January 15, 2000 at 12:54:15, Bruce Moreland wrote: >On January 15, 2000 at 09:40:51, James Swafford wrote: > >>If you use 20 bits to probe, why not just store the other 44 bits in the >>table as a checksum? You already know the other 20 bits match.... > >What I do in mine is I use the bottom however many bits to probe, and I store >the top 32 bits. So I lose some bits unless my table hits 2^32 elements, which >is not likely for a while at least. I just went back and looked at mine. I store the 48 msb's of the hash signature in the table as a checksum, so if I have less than 2^16 (64k) entries, I will lose bits, too. I thought I had encoded a mandatory minimum, but I didn't. No big deal, though. > >The reason I told the original poster to store all 64 is that I didn't want to >write something more complicated. Storing all 64 would work, it is easy to do, >and it may not be worse than any other solution. > >But the answer to your question is more complicated. There isn't such a thing >as a 44-bit standard data type, so you will end up storing and comparing several >chunks, unless you get tricky and use a union, which would let you use the >redundant bits for something else. Yeah, I realize 44 bits is a bit sticky. I was just trying to make a point. :-) > >Sound like it's extra code and a little extra complication in any case. > >bruce -- James
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.