Solver Statistics
From Sokoban Wiki
(→GroupEffort Test Suite) |
|||
| Line 1: | Line 1: | ||
| - | Sokoban | + | A Sokoban solver program is a program that tries to solve Sokoban levels. There are several solver programs available. (See [[Links#Sokoban Solvers |Sokoban solvers]].) |
| - | + | Solver programs may have different objectives: | |
| - | # | + | # Find any solution |
| - | # | + | # Find push optimal solutions |
| - | # | + | # Find push optimal solutions with best moves |
| - | # | + | # Find move optimal solutions |
| - | # | + | # Find move optimal solutions with best pushes |
| - | + | Given the computational complexity of these tasks, a solver program is usually designed for just finding any solution. If a solution is to be optimized, an optimizer program can be used. (See [[Links#Sokoban Solvers |Optimizers]].) | |
| - | + | The statistics presented here provide a picture of what the best freely available Sokoban solvers can do.<br /> | |
For copyrights, credits and redistribution of these statistics, please read the disclaimer. | For copyrights, credits and redistribution of these statistics, please read the disclaimer. | ||
| Line 16: | Line 16: | ||
== General benchmarking information == | == General benchmarking information == | ||
| - | + | For practical reasons, almost all tests have been performed with a 10 minutes time limit. The tests are, however, not a 10 minutes contest but a showcase of what the leading solvers can do, so other time limits are entirely feasible. | |
The hardware used for the tests is a standard PC with a modern operating system (64bit). | The hardware used for the tests is a standard PC with a modern operating system (64bit). | ||
| - | + | Tests are performed with default solver settings, except that RAM usage may be increased manually when this is beneficial. | |
{| class="soko-table feature-list" | {| class="soko-table feature-list" | ||
Revision as of 05:13, 29 November 2021
A Sokoban solver program is a program that tries to solve Sokoban levels. There are several solver programs available. (See Sokoban solvers.)
Solver programs may have different objectives:
- Find any solution
- Find push optimal solutions
- Find push optimal solutions with best moves
- Find move optimal solutions
- Find move optimal solutions with best pushes
Given the computational complexity of these tasks, a solver program is usually designed for just finding any solution. If a solution is to be optimized, an optimizer program can be used. (See Optimizers.)
The statistics presented here provide a picture of what the best freely available Sokoban solvers can do.
For copyrights, credits and redistribution of these statistics, please read the disclaimer.
Contents |
General benchmarking information
For practical reasons, almost all tests have been performed with a 10 minutes time limit. The tests are, however, not a 10 minutes contest but a showcase of what the leading solvers can do, so other time limits are entirely feasible.
The hardware used for the tests is a standard PC with a modern operating system (64bit).
Tests are performed with default solver settings, except that RAM usage may be increased manually when this is beneficial.
| Solver | Version | Author | Architecture | CPU usage | Sourcecode available | More info |
|---|---|---|---|---|---|---|
| Festival | 2.0 | Yaron Shoham | 64bit | 2 threads | no | Website with papers about the solver |
| Sokolution | 5.1 | Florent Diedler | 64bit | 2 threads | no | Sokoban solver "scribbles" by Florent Diedler |
| Takaken | 7.3.1 | Ken'ichiro Takahashi (takaken) | 64bit | 2 threads | no | Description of the used algorithms |
| Curry | 0.3 | Yaron Shoham | 64bit | 1 thread | Curry 0.3 | Website with paper about the solver |
| YASS | 2.146 | Brian Damgaard | 32bit | 1 thread | YASS 2.146 | Sokoban solver "scribbles" by Brian Damgaard |
| JSoko | 2.016 | Matthias Meger | 64bit | 1 thread | JSoko 2.01 | Solver information |
| Rolling Stone | 1.0 | Andreas Junghanns | 32bit | 1 thread | RS | Rolling Stone papers |
Contact: SokobanoWiki@gmail.com
XSokoban Test Suite
The XSokoban level set is the de facto standard test suite in the academic literature on Sokoban solver programming.
| Sokoban Solver Statistics - XSokoban Test Suite - Solved Levels | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Collection | Author | Levels | Rolling Stone | Takaken | YASS | JSoko | Sokolution | Curry | Festival | |
| XSokoban | Thinking Rabbit, ... | 90 | 54 | 90 | 84 | 67 | 90 | 90 | 90 | |
| Sokoban Solver Statistics - XSokoban Test Suite - Push-optimal Solutions | ||||||
|---|---|---|---|---|---|---|
| Collection | Author | Levels | Rolling Stone | Pereira et al. | JSoko | Sokolution |
| XSokoban | Thinking Rabbit, ... | 90 | 6 | 28 | 31 | 59 |
Large Test Suite
The large test suite comprises level packs selected for being a challenge for the solvers, but not too difficult. The statistics shows the results for the leading Sokoban solver programs.
This test suite is not subject to change, as it is meant to be a "Large Standard Test Suite" for Sokoban solver programs.
| Sokoban Solver Statistics - Large Test Suite - Solved levels | |||||
|---|---|---|---|---|---|
| Levels | Festival | Sokolution | Takaken | YASS | |
| Total levels | 3272 | 3121 | 3009 | 2933 | 2687 |
Click here to see the detailed results for the Large Test Suite.
Download the entire test suite here or the individual level sets from www.sokobano.de.
Open Test Suite
The statistics shows the results for the leading Sokoban solvers for various level collections. Please note that the grand totals are simple summations. The number of solved unique levels may differ slightly due to duplicate levels in the level packs.
This test suite may be changed by adding new level collections from time to time.
| Sokoban Solver Statistics - Open Test Suite - Solved levels | |||||
|---|---|---|---|---|---|
| Levels | Festival | Sokolution | Takaken | YASS | |
| Total levels | 4762 | 4487 | 4283 | 4171 | 3697 |
Click here to see the detailed results for the Open Test Suite.
GroupEffort Test Suite
The only reason for benchmarking the GroupEffort test suite is to demonstrate the limited veracity of the statement made in section "6 Conclusion" of the thesis, where the author writes that the GroupEffort solver "outperforms existing solvers".
The benchmarks show that this was only true for the cherry-picked small test set at the time the thesis was published, never for the presumably more representative large test set.
| Sokoban Solver Statistics - GroupEffort Test Suite - Solved Levels | |||||||
|---|---|---|---|---|---|---|---|
| Collection | Levels | GroupEffort | Festival | Sokolution | Takaken | JSoko | YASS |
| Small Set | 200 | 196 (a)(c) | 200 (b) | 200 (a) | 192 (b) | 180 (b) | 191 (b) |
| Large Set | 2851 | 1566 (a)(c) | 2642 (b) | 2471 (a) | 2462 (b) | 2165 (b) | 2260 (b) |
| Total | 3051 | 1762 | 2842 | 2671 | 2654 | 2345 | 2471 |
(a): Time limit: 5 minutes (b): Time limit: 10 minutes (c): Best GroupEffort solver configuration
Disclaimer
Sokoban =======
Sokoban (R) Registered Trademark of Falcon Co., Ltd., Japan Sokoban Copyright (C) 1982-2004 by Hiroyuki Imabayashi, Japan Sokoban Copyright (C) 1989, 1990, 2001-2004 by Falcon Co., Ltd., Japan
License =======
Redistribution and use of the Sokoban Solver statistics are permitted provided that the following conditions are met:
* Redistributions may not be sold, nor may they be used in a commercial product or activity. * Redistributions must reproduce the entire content of this disclaimer in the documentation and/or other materials provided with the distribution.
Credits (in no particular order) ================================
CosmoNotes, CosmoPoly, LOMA, MicroCosmos, MiniCosmos, NaboCosmos and PicoCosmos level sets Copyright (C) by Aymeric du Peloux. E-mail: aymeric.du-peloux@gadz.org Web site: unknown
dh1, dh2, bagatelle, cantrip, and maelstrom level sets Copyright (C) by David Holland E-mail: unknown Web site: unknown
Microban, Mas Microban, Sasquatch, Mas Sasquatch, Sasquatch III, Sasquatch IV, Sasquatch V, Sasquatch VI and Sasquatch VII level sets. Copyright (C) by David W. Skinner E-mail: sasquatch@bentonrea.com Web site: unknown
Grigr2001, Grigr2002 and GrigrSpecial level sets Copyright (C) by Evgeny Grigoriev (aka GRIGoRusha) E-mail: grigr@yandex.ru Web site: http://grigr.narod.ru/
SokEvo and SokHard level sets Copyright (C) 2002-2006 by Lee J Haywood E-mail: fruise@gmail.com Web Site: http://leehaywood.org/games/sokoban/
Sven level set Copyright (C) 2007 by Sven Egevad E-mail: sven.egevad@telia.com Web Site: unknown
Sokoban Perfect and Sokoban Revenge level sets Copyright (C) 1989, 1991 by Thinking Rabbit Web Site: https://sokoban.jp/
Y.M. Auto and Y.M. Handmade level sets Copyright (C) Yoshio Murase E-mail: yoshio@asahi.email.ne.jp Web Site: http://www.ne.jp/asahi/ai/yoshio/sokoban/index.html
JSoko Copyright (C) by Matthias Meger E-mail: JSoko@mail.de Web site: https://www.sokoban-online.de
Sokoban Automatic Solver (aka Takaken solver) Copyright (C) 2002 - 200X Ken'ichiro Takahashi (aka takaken) E-mail: takaken@ic-net.or.jp Web site: http://www.ic-net.or.jp/home/takaken/e/soko/index.html
YASS - Yet Another Sokoban Solver and Optimizer - For Small Levels Copyright (C) 2007 by Brian Damgaard Web site: https://sourceforge.net/projects/sokobanyasc/
Sokoban YASC Copyright (C) 2001-2020 by Brian Damgaard Web Site: https://sourceforge.net/projects/sokobanyasc/
YSokoban Copyright (C) 2007 by George Petrov E-mail: YGP@WriteMe.com Web Site: http://ygp.orgfree.com/sokoban.html
XSokoban Copyright (C) by Andrew Myers E-mail: andru@cs.cornell.edu Web Site: http://www.cs.cornell.edu/andru/xsokoban.html
Sokofan Copyright (C) 2005-2007 by Eric Leung E-mail: eleung2@hotpop.com Web Site: unknown
Sokolution Copyright (C) 2016-2019 by Florent DIEDLER E-mail: fdiedler2000@gmail.com Web site: http://codeanalysis.fr/sokoban/
Festival Copyright (C) 2020 by Yaron Shoham Web site: https://festival-solver.site/
Trademarks ==========
Company names, brand names and product names are trademarks or registered trademarks of their respective holders.
Contact =======
E-mail: sokobanowiki@gmail.com
