Author: Joshua Lee
Date: 07:19:46 06/29/01
Go up one level in this thread
>>Thankyou > >There is a fair bit of suspicion that the famous Chaos Nxe6 was due to a bug. I >heard third hand that someone who was there asked one of the Chaos programmers >why it made this brilliant move, the answer was "I don't know, but when I find >out, that's one bug I'm going to fix." > >It's actually quite easy to write a program that makes brilliant sacrifices like >this with very little tree searching. The only trouble is, it also makes a lot >of brilliant looking sacrifices that don't actually work. > >Go through some of the games Chaos played, and get them commented by Fritz. >You're going to find that Chaos made heaps of tactical mistakes and Fritz will >find most of them in about a millisecond. > >All of these older programs would get smashed by any decent modern program. >There are two reasons for this. > >1. We know a lot more now about writing strong programs. > >2. Playing a program designed for a 1 MIPS computer on a 1000 MIPS computer >doesn't get you as much advantage as you think, because it just isn't written to >take advantage of that much power. In fact it can even get worse due to bugs >showing up that don't matter at slower speeds. > >For the other old programs you list: > >Pioneer never existed as an actual program. It has been suggested that it never >even existed as a complete paper design, but that is hard to verify either way. > >Hitech and the better versions of Belle only worked on special hardware. They >won't run on any normal computer, although the hardware they ran on may still >exist in a closet somewhere and is only about ten times slower than a modern >cheap PC. > >Chess 4.X was in CDC-6600 assembly language. It won't run on anything modern. > >Nuchess was Fortran, so it would run without much trouble. It would also be one >of the better old programs, since it was very solid and had a well-tuned eval. >It would not be competitive with modern programs because it didn't have >null-move or anything equivalent so it's going to do at least 2 plies less than >anything modern when running on modern hardware. > >Elsewhere in the thread someone said Kaissa is in Fortran, and probably still >exists. This may be the most interesting of the old programs in a lot of ways. > >Bob still has some runnable versions of Cray Blitz i think, but probably not >really old ones. I wish more posts could be like this one... Well this clears up one thing just because they had small node amounts doesn't necessarily mean that they programed so well of a heuristic eval that on todays 1ghz they will beat all the new programs. I don't know that this move is a bug as several programs find it but hey when i write a program i hope mine has lots and lots of bugs if it plays moves like that! I guess the most appropriate question now is is there any ideas used in the older programs that aren't in use today? If pioneeer doesn't exist how come nobody has thought to try to re-create it with the added benifit of 1 C++ 2 Newer Algorithms 3 Faster Hardware 4 Tablebases, Opening books Also my idea is one of Putting all possible techniques into a program .... Alot of work considering i know very little about programming in general.
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.