Author: Ian Osgood
Date: 08:09:46 09/15/05
Go up one level in this thread
On September 12, 2005 at 13:11:00, Gerd Isenberg wrote: >Hi Peter, > >thanks for the report - really interesting. >There is so much more to research and to try in Go rather than in chess. > >Now and then i have a wistful look to >http://computer-go.org/pipermail/computer-go/ > >May floodfill-algos like kogge-stone with 4*128-bit registers be a usefull >application in computer go? ;-) > >Have fun, >Gerd There certainly is a lot of room for experimentation in the computer go field. GNU Go was already quite strong when they recently made the internal architectural change to go from 2D to 1D board arrays. The author of Many Faces of Go, David Fotland, has designed instruction sets for HP's PA-RISC line of processors, and his deep knowledge of computer architecture shows in his go program algorithms and data structure designs. Most of the strong programs rely on pattern matching for move selection, some using a bit-board approach and others using specialized DFA compilers. The GNU Go effort has also done an amazing job at regression testing. They can evaluate the breakage of each little change, both in terms of failing test cases and in terms of increased number of nodes searched. One nice thing about this, it is inviting for non-programming Go experts (a much larger set than programming Go players) to do nothing but write new test cases. Ian
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.