Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Two interesting positions to test your Programs Materialism

Author: Uri Blass

Date: 14:46:12 11/12/02

Go up one level in this thread


On November 12, 2002 at 16:40:49, ALI MIRAFZALI wrote:

>[D]4r1k1/pp3p2/1q1p2P1/6Q1/2p2Pn1/P5R1/1P2b1P1/5R1K b
>In the above Position 1....Be2xf1 is incorrect.How long does it take your
>program to avoid making this move?

Movei needs less than a minute.

>[D]1qr2rk1/1b2bpp1/pp2pn1p/1N1nN3/2BP4/P5R1/1P2QPPP/2B2RK1 b
>In the above diagram 1....a6xb5 is incorrect. Same Question.

here movei can see negative score for axb5 in less than a minute(0.01 pawn for
white) but it does not find something better than axb5 in a short time.

The score continues to go down to 0.25 pawns for white at depth 12
but movei does not find something better at that depth

Here is the score at the end of depth 12 and at the beginning of depth 13.

depth=12 -0.25 a6b5 c1h6 f6e8 c4b5 d5f6 b5e8 f6e8 e5d7 b8d6 d7f8 g8f8 e2d2 e6e5
Nodes: 131237325 NPS: 117842
Time: 00:18:33.67

depth=13 -0.31 a6b5 c1h6 d5f4 g3g7 g8h8 e2e3 f6h5 g7f7 b5c4 h6f8 e7f8 e5d7 b8d6
d7f8 b7g2
Nodes: 188222338 NPS: 119753
Time: 00:26:11.75

This is not the latest movei but only almost the latest movei and the latest
movei that is in another computer use experimental pruning and it is now working
on the GCP test suite.

I have some evidence that the experimental pruning is probably productive
in very fast bullet games and I believe that it is also productive
in longer time control but I need to do more tests to be sure.

The best version so far in the GCP test suite solved 148 in the GCP test suite
in 5 minutes per move on 1000Mhz and it used experimental pruning.

I hope that I improved the pruning rules in the last version and I hope to
continue to improve them in the future.

Uri



This page took 0.23 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.