Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A test position for chess programs

Author: Gareth McCaughan

Date: 10:12:11 12/10/00

Go up one level in this thread


On December 10, 2000 at 11:17:58, Uri Blass wrote:

> [D]b7/1q2p2k/3p1p1p/1Pr1nP2/6PP/1rN5/2RKBQ2/2R5 b - - 0 1
>
> This position is from Fritz-Gambittiger.
>
> The target is to find Rd5+ that is winning.
>
> How much time does your program need to find Rd5+?
> Is it the simplest way to win the game?

Crafty 17.6 on my machine (K6-2/400) chooses ...Qc7 very early
and sticks with it for at least 12 minutes. It seems to think it's
pretty good. (-0.90 -- i.e., better for Black -- at 12 minutes.)

PV at this point is ...Qc7; g5 hxg5; hxg5 fxg5; Qh2+ Kg7; Qh5 Nf7;
Qg6+ Kf8; Qe6 d5; Bg4. After playing as far as ...fxg5 and letting
it think again for about 2m30, its evaluation and next few moves
mostly agree with that.

After playing ...Rd5+ and letting it think for about 4 minutes,
the evaluation is much the same as after ...Qc7, and the PV begins
with Nxd5 Qxd5+ Ke1. After playing those moves, it takes a little
while to fail high on ...Rh3, but having resolved the fail-high
it still doesn't see the evaluation as being terribly good for
Black. Then it fails high again, resolves it at about 3:45, with
an evaluation of about -2.06: ...Rh3; Rc3 Rh1+; Bf1 Nd3+; Rxd3 Qxd3;
Rd1 Qxb5; h5 Qa5+: Rd2 Qe5+; Qe2 Be4. The first difference between
this PV and the one before the second fail-high is resolved is at
the move Rd1 (before, it had Qe2).



This page took 0 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.