Search found 19305 matches

by hgm
Tue May 20, 2008 7:13 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Artificial stupidity - making a program play badly
Replies: 54
Views: 21911

Re: Artificial stupidity - making a program play badly

Underpromotions and moves with negative SEE value are pruned with very high probability, long diagonal moves also have quite high probability, obvious recaptures have very low probability of being pruned, and so on. Finally, the frequency of pruning of course depends on the playing strength. Prunin...
by hgm
Mon May 19, 2008 6:27 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Minimizing Sharing of Data between Physical Processors
Replies: 5
Views: 2571

Re: Minimizing Sharing of Data between Physical Processors

I would say the overhead is non-existent. Assuming the tables are in L2 all the time.
by hgm
Mon May 19, 2008 5:32 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Predateur 0.1.1 : 1498
Replies: 12
Views: 2437

Re: Predateur 0.1.1 : 1498

I never use LGpgnver, as my GUI informs me if claims are correct or not. 8-)
by hgm
Mon May 19, 2008 4:24 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Predateur 0.1.1 : 1498
Replies: 12
Views: 2437

Re: Predateur 0.1.1 : 1498

I suspect that CPP1 (his previous engine) might be stronger. I played few games of cpp1 vs pos, and in mostly all the matches, CPP1 was ahead on material but let POS get a stalemate, three fold rep, or 50 move draw... This was my impression too. NEG (my own abomination... :wink: ), which is also a ...
by hgm
Mon May 19, 2008 4:22 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Predateur 0.1.1 : 1498
Replies: 12
Views: 2437

Re: Predateur 0.1.1 : 1498

I can see that would be a problem when increment = 0... Early Smirf versions used to behave the same. The funny thing is that between engines of approximately equal basic strength, this behavior is not so easy to punish. If the opponent starts thinking only after a book opening, say move 12 or so, h...
by hgm
Mon May 19, 2008 3:50 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Predateur 0.1.1 : 1498
Replies: 12
Views: 2437

Re: Predateur 0.1.1 : 1498

Yes, I know, I missed my vocation: become a politician or a diplomat. :lol: :lol: :lol:
by hgm
Mon May 19, 2008 3:24 pm
Forum: Computer Chess Club: General Topics
Topic: Prédateur 0.1.3b2
Replies: 25
Views: 5640

Re: Prédateur 0.1.3b2

JuLieN wrote:You're right, my time allocation alorithm is very basic. It devides the remaining time by 8, adds the time increment if there's one, and removes 200 microseconds.
Why by 8? You expect your engine only to survive 8 moves? :shock:
by hgm
Mon May 19, 2008 3:07 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Predateur 0.1.1 : 1498
Replies: 12
Views: 2437

Re: Predateur 0.1.1 : 1498

Well, as long as it does not actually lose on time, changing the time management won't work miracles. Even if it would play with half the time, it should lose only about 70 Elo. And the other half is not totally wasted, it did 10 moves in it. And these moves were actually better than what they other...
by hgm
Mon May 19, 2008 1:58 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Predateur 0.1.1 : 1498
Replies: 12
Views: 2437

Re: Predateur 0.1.1 : 1498

Kind of arbitrary, isn't it?

As it did lose each and every single game it played, its rating is just as likely to be 1000, or 500, or 0... You really have no idea about its strength at all, except that it must be very, very weak.

This is as good as a rating based on zero games. :lol:
by hgm
Mon May 19, 2008 7:44 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Minimizing Sharing of Data between Physical Processors
Replies: 5
Views: 2571

Re: Minimizing Sharing of Data between Physical Processors

A lot depends on if the processors share a cache or not, and if the data is read-only or is written. If two CPUs that do not share a cache share a large memory array of read-only data, there is no penalty: They simply load the data in their own caches, and it gets the status 'shared' there. The bus ...