pawn endgame

[lech]

A standard evoluation sometimes doesn't match to many pawn endgames.

Rybka 4.1 4CPU - Komodo 5 1CPU (The Champions 2012 - 29 minutes / 40 moves)
[d]8/ppk5/6p1/4pbP1/8/P3N3/1PP5/2K5 b - - 0 43
In draw position, Black plays 43...Kd6 and after 44.Nxf5 gxf5 45.c4 loses immediately.
Two chain passed pawns are weaker than isolated.

[Kyodai]

That reminds me of a recent game 40/2 I played vs Chessmaster personality
GM Zukertort. After the last inviting move 24.Rd1-d4! I was hoping for
some exchanges...

[d]6k1/1pp4p/p1p5/3r4/3R2b1/P1P2N2/1P3KPP/8 b - - 0 24

And black happily went into a lost pawn ending by taking on f3 and d4
(24.-Bxf3 25.Kxf3 Rxd4 26.cxd4) and we have this position (won for white)

[d]6k1/1pp4p/p1p5/8/3P4/P4K2/1P4PP/8 b - - 0 26

[zullil]

A standard evoluation sometimes doesn't match to many pawn endgames.

Rybka 4.1 4CPU - Komodo 5 1CPU (The Champions 2012 - 29 minutes / 40 moves)
[d]8/ppk5/6p1/4pbP1/8/P3N3/1PP5/2K5 b - - 0 43
In draw position, Black plays 43...Kd6 and after 44.Nxf5 gxf5 45.c4 loses immediately.
Two chain passed pawns are weaker than isolated.

Critter avoids the error:

Code: Select all

Critter v1.6a 64-bit, by Richard Vida
hardware POPCNT enabled
setboard 8/ppk5/6p1/4pbP1/8/P3N3/1PP5/2K5 b - - 0 43
info string Opening book: "book.cbk" - 57853 entries [native]
info string GTB Init OK (5 piece set)
info string Using 8 thread(s), 2048 MB hash LP=off
go infinite
 2/ 2  00:00      247       0  +0.10 Kd6 Kd2
 3/ 3  00:00      382       0  +0.07 Kd6 Kd2 Bd7
 4/ 5  00:00      612  612000  +0.10 Kd6 Kd2 Bd7 Kd3
 5/ 7  00:00     1087   1087K  +0.01 Kd6 Kd2 Bd7 Kd3 b6
 6/ 8  00:00     1946  973000  +0.03 Kd6 Kd2 Bd7 Kd3 b5 c4
 7/ 9  00:00     4072   1018K  +0.01 Kd6 Kd2 Be4 Nc4+ Ke6 b3 Bc6
 8/10  00:00     8620   1077K  +0.01 Kd6 Kd2 Bd7 Kd3 Bc6 c4 Bf3 b3
 9/14  00:00    26747   1485K  -0.03 Kd6 Kd2 Bd7 Kd3 b5 b3 a6 c4 Bc6 cxb5 axb5
10/15  00:00    95206   2720K  -0.01 Kd6 Kd2 Ke6 c4 Be4 b4 b6 Ng4 Kf5 Nf6 Bb7
11/15  00:00   156920   2905K  +0.00 Kd6 Kd2 Ke6 c4 Be4 b3 b6 a4 Bc6 Kd3 Bf3 Nc2
12/15  00:00   197469   3085K  +0.00 Kd6 Kd2 Ke6 c4 Be4 b3 b6 a4 Bc6 Kd3 Bf3 Nc2
13/17  00:00   330288   3711K  +0.00 Kd6 Kd2 Ke6 c4 Be4 Ke2 b6 b3 Bc6 a4 Be4 Kd2 Ba8 Kd3 Bf3 Kd2 Be4 Ng4 Kf5
14/19  00:00   730632   4970K  +0.00 Kd6 Kd2 Be4 b3 b6 a4 Ke6 Ke2 a6 Kd2 Bb7 Nc4 b5 axb5 axb5
15/20  00:00    1323K   5656K  +0.01 Kd6 Kd2 Be4 Ke2 b6 Nc4+ Ke6 Ne3 Kd6
16/21  00:00    2140K   5996K  +0.01 Kd6 Kd2 Be4 Nc4+ Ke6 b3 Bc6 Ke3 Kf5
17/24  00:00    3805K   6652K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bb5+
18/26  00:00    6069K   7132K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bb5+
19/30  00:01   11840K   7713K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bb5+
20/30  00:02   17449K   7986K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bb5+
21/32  00:03   30093K   8422K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bb5+
22/35  00:06   53611K   8745K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bb5+
23/37  00:09   87991K   8923K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bd5 Nxd5 Kxd5 b4 e4+ Ke3 a6 Ke2 Ke5 Ke3 Kd5
24/39  00:16  146977K   9084K  +0.00 Kd6 Kd2 Be4 c4 Ke6 c5 Bc6 Kd3 Be8 b3 Bb5+ Kc3 Bc6 Kd3 Bd5 Nxd5 Kxd5 b4 e4+ Ke3 a6 Kd2 Kc4 Ke3 Kd5
25/41  00:27  257552K   9323K  +0.00 Kd6 Nxf5+ gxf5 c4 f4 c5+ Kd5 b4 a6 a4 e4 g6 Ke6 b5 axb5 axb5 Kf6 c6 bxc6 bxc6
25/45  00:54  528525K   9667K  +0.00 Be6 Kd2 Kd6 Kd3 b6 Ke4 b5 Kd3 Bd5 b3 Be6 c4 bxc4+ Nxc4+ Kd5 Ne3+ Kd6 Nc4+
26/45  01:07  656105K   9681K  +0.00 Be6 Kd2 Kd6 Kd3 b6 Ke4 b5 Kd3 Bd5 b3 Be6 c4 bxc4+ Nxc4+ Kd5 Ne3+ Kd6 Nc4+

[lech]

Code: Select all

25/41  00:27  257552K   9323K  +0.00 Kd6 Nxf5+ gxf5 c4 f4 c5+ Kd5 b4 a6 a4 e4 g6 Ke6 b5 axb5 axb5 Kf6 c6 bxc6 bxc6
25/45  00:54  528525K   9667K  +0.00 Be6 Kd2 Kd6 Kd3 b6 Ke4 b5 Kd3 Bd5 b3 Be6 c4 bxc4+ Nxc4+ Kd5 Ne3+ Kd6 Nc4+
26/45  01:07  656105K   9681K  +0.00 Be6 Kd2 Kd6 Kd3 b6 Ke4 b5 Kd3 Bd5 b3 Be6 c4 bxc4+ Nxc4+ Kd5 Ne3+ Kd6 Nc4+

54 seconds, it could be too little time to avoid in this tournament.

[Uri Blass]

Movei is relatively fast here.
with one thread(movei is not SMP) less than 1 minute(when houdini3 needs more than it even with 4 threads).
The reason is that I simply extend transition to pawn endgames when the extension is not constant and I extend more plies when the remaining depth is bigger.

Movei needs depth 19 to see a positive score after Nxf5+ gxf5 but needs only depth 16 to find Nxf5+ with a positive score

avoiding Kd6 is depth 17 for movei

Note that I did not work on movei lately and I am very surprised that my simple extension trick make it faster even today relative to top programs like houdini or critter(critter used 8 threads in the analysis).