Author: Keith Evans
Date: 15:37:34 11/20/02
Go up one level in this thread
On November 20, 2002 at 16:22:29, Dann Corbit wrote: > >It seems that they are popular for hardware systems, as Marc Boule's thesis "An >FPGA Move Generator for theGame of Chess" also mentions them briefly: "In >concert with pre–calculated attack tables, bit–wise operations are used to >speed–up move generation." > Marc is talking about software implementations there. A Belle-style hardware move generator like Marc built, can tell which squares are attacked without the need for separate attack tables. You build a 2-D silicon chessboard and pieces basically radiate attack rays. Squares know that they are under attack when they receive one of these rays. The rays all propagate in parallel - so you can get results in a single cycle. Those CMU projects like the "Scooby Doo" project that you reference were based on TSCP, and the students didn't have a lot of experience with computer chess. So sometimes they implemented software-like methods in hardware, rather than coming up with techniques which could have taken better advantage of the hardware. (Of course if they had tried, then they probably wouldn't have been able to finish on time given the short time frame.) Their hardware was also rather limited in capacity. Keith
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.