Search found 17077 matches

by bob
Fri Sep 07, 2007 9:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

log.001: time=6.47 mat=0 n=12083521 fh=94% nps=1.9M log.002: time=6.49 mat=0 n=12083521 fh=94% nps=1.9M log.003: time=6.47 mat=0 n=12083521 fh=94% nps=1.9M log.004: time=6.40 mat=0 n=12083521 fh=94% nps=1.9M Or, for longer searches: log.001: time=18.13 mat=0 n=36425309 fh=93% nps=2.0M log.002: time...
by bob
Fri Sep 07, 2007 8:08 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Threads and cores questions
Replies: 1
Views: 1483

Re: Threads and cores questions

Some newbie questions. Suppose we have a processor of n cores: 1) Can we auto detect number of cores (specific for C++ and Debian Linux)? 2) What is the maximum number of threads should we create? 3) Is there any method to start and run a thread in a given core (C++ and Debian Linux)? 4) (Similar t...
by bob
Fri Sep 07, 2007 8:03 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: botnet vs BlueGene
Replies: 1
Views: 1417

Re: botnet vs BlueGene

This article: http://www.itnews.com.au/News/60752,storm-worm-botnet-more-powerful-than-top-supercomputers.aspx makes the claim that the "Storm Worm" botnet would defeat BlueGene in a game of chess. (Ok, suppose such programs existed.) I believe the claim is wrong, simply because: 1. There is no kno...
by bob
Fri Sep 07, 2007 7:59 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

You are having far better luck than I am. I am using 40 positions, 2 games each, with 4 standard opponents. A single run has a wild variance when compared with a second run or a third run... I guess the way that I am testing now can be looked at like a type of genetic evolution. A small change and ...
by bob
Fri Sep 07, 2007 7:55 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

Also I test each change individually from some baseline version, not incrementally building up the changes on top of each other. Hi Martin, can you tell me more about this? I'm currently always testing from the current best version, but I do that only because I like to see the program do some progr...
by bob
Fri Sep 07, 2007 7:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

You said that you don't finish an iteration, you cut the search off when time runs out. That alone should make it impossible to choose the same move every time, unless your eval is _very_ simple. Because there should be plenty of cases where you change your mind if you finish an iteration, but if t...
by bob
Fri Sep 07, 2007 7:50 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

Now, let me explain what I do and why. Yes, that sounds like a very sensible way to do things. It is in fact very close to what I do in Joker: I also have an absolute time limit in case of emergensies, to interrupt cases where a single move would already cause a loss on time. (This sometimes happen...
by bob
Fri Sep 07, 2007 7:16 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

[ Now, let me explain what I do and why. [snip] I've been using this for at least 10+ years in Crafty, and for at least 15 years prior to that in Cray Blitz. I have tweaked things from time to time, but I have never looked at a game and thought "dang, If I had just searched another 2 seconds, I wou...
by bob
Thu Sep 06, 2007 9:08 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

And yes, if your program is basically a "fixed depth" version because it always finishes an iteration, then that will make repeatability very high. But it is not the best way to manage time overall, and not many do that... I have often wondered about this. The strength of a chess program is only a ...
by bob
Thu Sep 06, 2007 7:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: nondeterministic testing
Replies: 86
Views: 34772

Re: nondeterministic testing

I have a theory about it. There may be knowledge that is not important at long time control because search can replace it. Here is an example. Assume that no tablebases are used. A program may not know how to win KR vs K but know that KR vs K is winning for the rook. Adding special knowledge for th...