Search found 16846 matches

by bob
Tue Jul 03, 2007 8:38 pm
Forum: Computer Chess Club: General Topics
Topic: Bob Hyatt Crafty 21.5 and 21.6
Replies: 10
Views: 4790

Re: Bob Hyatt Crafty 21.5 and 21.6

I hope not...

:)
by bob
Tue Jul 03, 2007 8:37 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Iterative DTS
Replies: 17
Views: 10318

Re: Iterative DTS

anything is possible, but it will be anything but easy. That's why I chose to stick with the recursive formulation of the search, it is more readable.

Cray Blitz was a mess in that regard.
by bob
Tue Jul 03, 2007 3:26 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Iterative DTS
Replies: 17
Views: 10318

Re: Iterative DTS

Sorry. I was simply explaining the usual issue with an iterated search conversion. Your code was too long to look at on the older computer I was using at the time (a 1024 x 768 laptop with a small screen). It sounds like you did things the right way. Next step would be to compare old vs new code to ...
by bob
Tue Jul 03, 2007 3:59 am
Forum: Computer Chess Club: General Topics
Topic: Bob Hyatt Crafty 21.5 and 21.6
Replies: 10
Views: 4790

Re: Bob Hyatt Crafty 21.5 and 21.6

glorfindel wrote:
bob wrote: Hopefully we will have a "final" 21.6 version before long and that will be released to everyone.
Together with the source code?
yep...
by bob
Mon Jul 02, 2007 11:59 pm
Forum: Computer Chess Club: General Topics
Topic: Nalimov Tablebases
Replies: 6
Views: 3018

Re: Nalimov Tablebases

tbs files are not needed. They are "summary" files that give information such as deepest mate, etc. They are not used by the chess engines...
by bob
Mon Jul 02, 2007 11:06 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 'undercut' hash-replacement scheme
Replies: 9
Views: 5816

Re: 'undercut' hash-replacement scheme

my only quibble here is the term "distribution" as if that can be sub-optimal. The distribution of nodes by draft is not a uniform distribution, so I'm not sure how we can conclude this distribution is better than that distribution, unless you first search the tree, then compare the tree to the stuf...
by bob
Mon Jul 02, 2007 9:02 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Iterative DTS
Replies: 17
Views: 10318

Re: Iterative DTS

I'm talking about the specific example code in the root post. It's a not exactly 'giant' switch but still a switch statement implementing a state machine. MvH Dan Andersson OK. I didn't look at the code, it was too long to scroll through. I commented only on how I have seen "iterative searches" imp...
by bob
Mon Jul 02, 2007 7:46 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Iterative DTS
Replies: 17
Views: 10318

Re: Iterative DTS

That I don't follow. What "giant switch table"??? I didn't have one in Cray Blitz at all so I am not sure what you are talking about here...
by bob
Mon Jul 02, 2007 7:44 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 'undercut' hash-replacement scheme
Replies: 9
Views: 5816

Re: 'undercut' hash-replacement scheme

My only comment is this. Given the choice of replacing one of N entries, I can't imagine a circumstance where I would choose to replace with a deeper draft as opposed to replacing one with a shallower draft. The two-tiered table (Belle approach) addresses two main issues. (1) keep the deep draft ent...
by bob
Mon Jul 02, 2007 6:21 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 'undercut' hash-replacement scheme
Replies: 9
Views: 5816

Re: 'undercut' hash-replacement scheme

The question has to be "why" however. Because of this: http://home.hccnet.nl/h.g.muller/hashnew.jpg The graph shows that shallowest-of-4 replacement, which is effectively an implementation of the Bell scheme with a depth-preferred table 3 times the size of the always-replace table, performs signifi...