Author: Robert Hyatt
Date: 09:05:35 12/17/99
Go up one level in this thread
On December 17, 1999 at 02:10:19, J. Wesley Cleveland wrote: >On December 15, 1999 at 14:39:18, Greg Lindahl wrote: > >>I'm interested in coordinating an effort to build the next >>world-champion chess program. FPGA technology has advanced since Deep >>Blue's construction such that replicating its architecture is now >>inexpensive. [snip] > >Don't you think you are being too hard on this guy. I thought he had an >interesting idea, that may or may not work, but is certainly worth >investigating. The other responses mostly read like "It won't work. It can't >work. Don't even think about it.", generally without much evidence. If one were >into psycobabble, one could ask "What is so threatening about this that it has >to be rejected without serious consideration ?" I haven't seen much of that happening. When you start something innovative, the very first step is to identify the _hard_ problems that must be solved. It makes absolutely no sense to solve all the easy stuff if there is a hard problem that can't be solved. That is why I mentioned the 'memory' issue first. You _must_ have registers to hold the chess position and other bits of data that the CPU needs to pass over, like side to move, depth to search to, castling status, etc. You must have memory to pass that information down the tree, ply by ply. You must have memory for a repetition stack. You have to have data paths to pass data from one side of the chip to the other so that evaluation can be done in parallel. If those can't be solved, then there is little point in considering the idea of a single-chip FPGA chess engine. If the problems _can_ be solved with just a few external chips, then it has a good chance of success. I don't see how pointing out the _hard_ problems first is considered an attack. The _very first_ thing I did in Crafty when I started was to design and write the incremental bitmap update as that looked like the hardest thing to do, and if it didn't work reasonably well, the idea was doomed. Somehow this is getting overlooked. The problems aren't imaginary. They aren't easy to solve. But maybe they aren't impossible. All of the things I have mentioned can be found in papers by Thompson and Hsu over the years. So it isn't something I made up...
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.