Search found 469 matches

by rvida
Thu Jul 30, 2009 10:02 am
Forum: Computer Chess Club: General Topics
Topic: Diminishing returns of increasing search depth
Replies: 43
Views: 7372

Re: Diminishing returns of increasing search depth

I would furthermore advise to do this not with a single engine, but select a group of about 6 engines that is about equally strong when limited to 12 ply. And then just play a number of round-robins with them inclucing 4 'generations' (N, N+1, N+2, N+3) for every N that is feasible. Perhaps a small...
by rvida
Sun Jul 26, 2009 11:31 pm
Forum: Computer Chess Club: General Topics
Topic: Critter 0.40 released
Replies: 4
Views: 1182

Re: Critter 0.40 released

From the changes you've made to the evaluation, it looks as though testers should consider this a completely new version rather than combining results with 0.39a? Cheers, Graham. Yes, it is somewhat stronger with totally different playing style, thus merging results with previous version is meaning...
by rvida
Sun Jul 26, 2009 10:31 pm
Forum: Computer Chess Club: General Topics
Topic: Critter 0.40 released
Replies: 4
Views: 1182

Critter 0.40 released

Hi everyone, Critter 0.40 is available from SourceForge . It is still work in progress, main reason of this relase is a fix of a rather serious bug which caused occasional crashes (especially with fast TC and/or pondering). From the changelog: Version 0.40 (2009.07.26) ======================...
by rvida
Fri Jul 24, 2009 1:29 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: DIRECT BITBOARD MOVEGENERATION
Replies: 56
Views: 9383

Re: DIRECT BITBOARD MOVEGENERATION

Im a PASCAL guy, I dont use C, but let me guess...

If you return ONLY "rnk", compiler will optimize away whole "fle" calculation.
If you return ONLY "fle", compiler will optimize away whole "rnk" calculation.

If you return logical OR of both, it must include both calculations.
by rvida
Thu Jul 16, 2009 7:34 am
Forum: Computer Chess Club: General Topics
Topic: New chess applet
Replies: 15
Views: 2275

Re: New chess applet

M ANSARI wrote:... it mysteriously managed to add a rook on the board. This happened quite a few times. Still buggy I am afraid.
Just a guess - maybe a castling move making/unmaking issue?
by rvida
Tue Jun 30, 2009 9:58 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: hashing in quiescent
Replies: 17
Views: 2154

Re: hashing in quiescent

That said, I have for many years kept a small eval cache that holds recent full evaluations (it doesn't hold evals that are cut off early by futility checking in the eval code). The call to evaluate() checks the cache and returns the cached value if it is there. Depth doesn't matter for this: eithe...
by rvida
Tue Jun 16, 2009 9:41 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: alpha beta hashing and move ordering
Replies: 11
Views: 1367

Re: alpha beta hashing and move ordering

Pulsar uses root ordering prior to every iteration of search. If on ply 1 the best move is move 20 then move 20 moves to spot 1. On ply 2 if the best move is now move 13 then move 13 moves to spot 1 and move 20 moves to spot 2. This means on interation 3, a 3 ply search it will search move 13 then ...
by rvida
Sat May 30, 2009 12:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Problems with this position
Replies: 36
Views: 6323

Re: Problems with this position

Critter has some problems here too, although I'm not sure how bad it is. At ply 23 it found something but after 2x fail-high it returns with original score (??). At ply 24 it clearly sees the promotion though. setboard 4k3/8/8/8/8/8/4P3/4K3 w go 2 00:00 7 0 +1.21 e2e4 Ke8d7 3 00:00 39 0 +1.54 e2e4 K...
by rvida
Tue May 26, 2009 5:17 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: transposition table details
Replies: 22
Views: 5130

Re: transposition table details

You will notice that the first "go depth 15" command takes 29.3 seconds in this case. The second time I run this it takes 0.002 seconds! This makes me think that Vas has a rule not to ever overwrite PV nodes, it's probably as simple as that or something similar. Or simply feed back the PV to the TT...