Page 1 of 1

Pawns ending

Posted: Fri Dec 10, 2010 5:50 pm
by Andres Valverde
From a engine vs engine blitz game :

[D] 8/8/3kPp1p/5P1p/p1p5/p1K3P1/8/8 b - - 0 60

Both engines played many moves giving +3/+5 scores for black .... being a clear draw for an human.

Re: Pawns ending

Posted: Sat Jan 01, 2011 1:58 pm
by Kurt Utzinger
Andres Valverde wrote:From a engine vs engine blitz game :

[D] 8/8/3kPp1p/5P1p/p1p5/p1K3P1/8/8 b - - 0 60

Both engines played many moves giving +3/+5 scores for black .... being a clear draw for an human.
What a shame for most engines not realizing that this is a draw. Even a human player of 1700 Elo would do better.
Kurt

Re: Pawns ending

Posted: Sat Jan 01, 2011 2:28 pm
by Vinvin
The strength a connected passed pawn vs the weakness of 6 is isolated pawns worth the 3 pawns difference ... 8-)

..converting it into a Testposition.White to play and draw!!

Posted: Sat Jan 01, 2011 3:27 pm
by George Tsavdaris
Vinvin wrote: [d]8/8/3kPp1p/5P1p/p1p5/p1K3P1/8/8 b - - 0 60
The strength a connected passed pawn vs the weakness of 6 is isolated pawns worth the 3 pawns difference ... 8-)
Pretty position and it becomes a good testposition(because the current one is not a good one for white/black to play and find the best move-since computers in this position actually are good at this in playing conditions as they would always find the best move so will not lose as white or black, but they are very bad at analysis conditions since they will deceive their observer believing the position is won for black) if you add some other pieces. :D

White to play and draw with 1.Rxd6!!
1.Rc6? or 1.Ra6? lose as i saw by analyzing a bit.

[d]8/4k3/1R1nPp1p/2p2P1p/p1p5/p5P1/8/2K5 w - - 0 1

Re: ..converting it into a Testposition.White to play and dr

Posted: Sat Jan 01, 2011 11:17 pm
by Dirt
George Tsavdaris wrote:White to play and draw with 1.Rxd6!!
1.Rc6? or 1.Ra6? lose as i saw by analyzing a bit.

[d]8/4k3/1R1nPp1p/2p2P1p/p1p5/p5P1/8/2K5 w - - 0 1
Houdini gets it at ply 30 with two threads.

Stockfish 1.91 with one thread took until ply 32, but only used a little over 1/4 the time.