Good engines for endgame play without tablebases?

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

chesskobra
Posts: 358
Joined: Thu Jul 21, 2022 12:30 am
Full name: Chesskobra

Good engines for endgame play without tablebases?

Post by chesskobra »

What are some engines that are good in endgame play without using tablebases? I have seen some pleasant endgame play by Igel 3.5, but don't remember other good engines.

Today I played a game between Pedone 3.0.7 (White) vs Peacemaker 2.30 (Black) at 40 moves in 5 minutes (intel core i5, 4GB RAM). After 79 moves, it was the following position on the board,

[d]8/1p6/5Kp1/5b1p/7P/2k5/8/6q1 w - - 0 80

Peacemaker (3425 on CCRL) took another 123 moves to win it. At some point I thought Pedone would escape with a draw by 50 move rule.

Should engines be tested without tablebases? The ratings seem very deceptive.

(Incidentally, diagrams are looking beautiful on the new forum. Thank you for the great work.)
Jouni
Posts: 3743
Joined: Wed Mar 08, 2006 8:15 pm
Full name: Jouni Uski

Re: Good engines for endgame play without tablebases?

Post by Jouni »

This position is mate in 6. No help from tablebases.
Jouni
chesskobra
Posts: 358
Joined: Thu Jul 21, 2022 12:30 am
Full name: Chesskobra

Re: Good engines for endgame play without tablebases?

Post by chesskobra »

So how should I test engines for 'basic' endgame play, basic as in the way a strong human, say an FM, would play? I would appreciate suggestions, since I don't have experience running such tests. Are there some test cases I can simply run? Can I run engine matches on an epd collection? I would like to
  • limit engine depth so that we will be testing its overall understanding as to "how to play" endings
  • have test positions that can be won or drawn by a strong human in about 20 moves without too much calculation, especially positions that are not won by a tactical shot, positions for which we can adjudicate a draw if no engine can win in 20 moves.
User avatar
towforce
Posts: 12665
Joined: Thu Mar 09, 2006 12:57 am
Location: Birmingham UK
Full name: Graham Laight

Re: Good engines for endgame play without tablebases?

Post by towforce »

chesskobra wrote: Sat Mar 02, 2024 9:10 pm(Incidentally, diagrams are looking beautiful on the new forum. Thank you for the great work.)

+1

With regards to the subject, I would guess that there's a test suite for endgame positions.
Human chess is partly about tactics and strategy, but mostly about memory
Chessqueen
Posts: 5685
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

Re: Good engines for endgame play without tablebases?

Post by Chessqueen »

Most engines rated above 2000 can find this mate immediately without tablebases, even Snowy rated 1990 :roll

FEN: 8/1p6/5Kp1/5b1p/7P/2k5/8/6q1 w - - 0 80

Snowy_0_2_x64:
1 00:00 4 65 -17.75 Kf6-e5
2 00:00 143 2k -18.02 Kf6-e5 Qg1-d4+
3 00:00 124 2k -19.25 Kf6-f7 Qg1-d4 Kf7-e7
4 00:00 1k 24k -19.25 Kf6-f7 Qg1-d4 Kf7-e7 Qd4xh4+
5 00:00 5k 75k -19.34 Kf6-f7 Qg1-g3 Kf7-f6 Kc3-d4 Kf6-e7
6 00:00 37k 607k -19.41 Kf6-f7 Qg1-d4 Kf7-e7 Kc3-d3 Ke7-f7 Qd4xh4
7 00:00 62k 4,159k -19.49 Kf6-f7 Qg1-d4 Kf7-e7 b7-b5 Ke7-f7 b5-b4 Kf7-e7
8 00:00 822k 5,845k -19.79 Kf6-f7 b7-b5 Kf7-e7 b5-b4 Ke7-d6 b4-b3 Kd6-d5 b3-b2
9 00:00 1,099k 4,690k -27.86 Kf6-f7 b7-b5 Kf7-e7 b5-b4 Ke7-d6 b4-b3 Kd6-d5 b3-b2 Kd5-e5
10 00:01 6,686k 6,293k -27.86 Kf6-f7 b7-b5 Kf7-e7 b5-b4 Ke7-d6 b4-b3 Kd6-d5 b3-b2 Kd5-e5 b2-b1Q
11 00:02 13,428k 4,968k -M6 Kf6-f7 Qg1-b6 Kf7-e7 Qb6-e6+ Ke7-f8 Qe6-d7 Kf8-g8 g6-g5 h4xg5 Bf5-g6 Kg8-h8
Chessqueen
Posts: 5685
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

Re: Good engines for endgame play without tablebases?

Post by Chessqueen »

[pgn][Event "Computer chess game"]
[Site "DESKTOP-4QNC0GS"]
[Date "2024.03.05"]
[Round "?"]
[White "Snowy_0_2_x64"]
[Black "Snowy_0_2_x64"]
[Result "0-1"]
[BlackElo "2000"]
[Time "21:19:08"]
[WhiteElo "2000"]
[TimeControl "300+5"]
[SetUp "1"]
[FEN "8/1p6/5Kp1/5b1p/7P/2k5/8/6q1 w - - 0 80"]
[Termination "normal"]
[PlyCount "12"]
[WhiteType "program"]
[BlackType "program"]

80. Kf7 {(Kf6-f7 Qg1-b6 Kf7-e7 Qb6-e6+ Ke7-f8 Qe6-d7 Kf8-g8 g6-g5 h4xg5
Bf5-g6 Kg8-f8 Qd7-f7+) -M6/13 18} b5 {(b7-b5 Kf7-f6 Kc3-d4 Kf6-f7 Kd4-e5
Kf7-e7 Qg1-a7+ Ke7-e8 Bf5-e6 Ke8-f8 Qa7-f7+) +M6/12 19} 81. Ke7 {(Kf7-e7
Qg1-a7+ Ke7-f6 Kc3-d4 Kf6-g5 Kd4-e5 Kg5-h6 Ke5-e6 Kh6-g5 Qa7-e3+) -M5/11 1}
Qa7+ {(Qg1-a7+ Ke7-f6 Kc3-d4 Kf6-g5 Kd4-e5 Kg5-h6 Ke5-e6 Kh6-g5 Qa7-e3+)
+M5/10 2} 82. Kf6 {(Ke7-f6 Kc3-d4 Kf6-g5 Kd4-e5 Kg5-h6 Ke5-e6 Kh6-g5
Qa7-e3+) -M4/9 0} Kd4 {(Kc3-d4 Kf6-g5 Kd4-e5 Kg5-h6 Ke5-e6 Kh6-g5 Qa7-e3+)
+M4/8 0} 83. Kg5 {(Kf6-g5 Kd4-e5 Kg5-h6 Ke5-e6 Kh6-g5 Qa7-e3+) -M3/7 0} Ke5
{(Kd4-e5 Kg5-h6 Ke5-e6 Kh6-g5 Qa7-e3+) +M3/6 0} 84. Kh6 {(Kg5-h6 Ke5-e6
Kh6-g5 Qa7-e3+) -M2/5 0} b4 {(b5-b4 Kh6-g5 Qa7-e3+) +M2/4 0} 85. Kg5
{(Kh6-g5 Qa7-e3+) -M1/3 0} Qe3# {(Qa7-e3+) +M1/2} 0-1[/pgn]