Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Test Positions / Alternative Chess Test 1-B
- - By dorszcz (**) Date 2019-10-22 14:00 Upvotes 1
Alternative Chess Test 1-B (ACT1-B) - is a collection of 183 hard chess positions for testing engines. This is the revised version of ACT which had previously 287 positions. Now the easiest positions are removed, as well as some faulty ones and these that could be solved by luck.

Conditions:
Time limit: 15 seconds per position
CPU: i5-3570 (MP engines: 4x3600 MHz, SP engines: 1x3800MHz)
Tablebases: 6-pieces Syzygy, 5-pieces Gaviota, 5-pieces (and ~80 GB of most popular 6-pieces) Nalimov, all on SSD
Strongest version of engine is used: 64 bit, POPCNT, SSE4.2 etc.
256 MB hash, no books, no learning, 50 move rule is enabled

Results:
141 out of 183 = 77.0% - Crystal 260819
132 out of 183 = 72.1% - MateFinder 260819
132 out of 183 = 72.1% - CorChess 3.1 260819
128 out of 183 = 69.9% - Eman 4.00
125 out of 183 = 68.3% - Sting 14
118 out of 183 = 64.4% - Houdini 1.5a
115 out of 183 = 62.8% - Cfish 240719
115 out of 183 = 62.8% - SugaR NN 130819
114 out of 183 = 62.2% - Critter 1.6a
114 out of 183 = 62.2% - Stockfish 10
111 out of 183 = 60.6% - Honey X5i
108 out of 183 = 59.0% - AsmFishWCP_2019-07-23
104 out of 183 = 56.8% - Crystal-Honey X5i
92 out of 183 = 50.2%  - IvanHoe 9.46b
90 out of 183 = 49.1%  - Naum 4.6
89 out of 183 = 48.6%  - Spark 1.0
86 out of 183 = 46.9%  - Komodo 10
84 out of 183 = 45.9%  - Deep Rybka 4
84 out of 183 = 45.9%  - Rybka 3 Dynamic
82 out of 183 = 44.8%  - Booot 6.3.1
79 out of 183 = 43.1%  - Gull 3
79 out of 183 = 43.1%  - Sting SF 3 VE
73 out of 183 = 39.8%  - Wasp 3.75
70 out of 183 = 38.2%  - Rybka WinFinder 2.2
67 out of 183 = 36.6%  - Nemorino_5.00
61 out of 183 = 33.3%  - Fire 2.2+ xTreme GH
61 out of 183 = 33.3%  - Spike 1.4
60 out of 183 = 32.7%  - Andscacs 0.95
56 out of 183 = 30.6%  - Wasp 3.60
54 out of 183 = 29.5%  - Stockfish 5
52 out of 183 = 28.4%  - Equinox 3.30
50 out of 183 = 27.3%  - Xiphos 0.5.3
48 out of 183 = 26.2%  - Fire 7.1
43 out of 183 = 23.4%  - Alfil 13.1
42 out of 183 = 22.9%  - Laser 1.7
41 out of 183 = 22.4%  - SmarThink 1.98 (1CPU)
41 out of 183 = 22.4%  - Xiphos 0.5.6
37 out of 183 = 20.2%  - Topple 0.7.3
35 out of 183 = 19.1%  - RofChade 2.2
35 out of 183 = 19.1%  - Protector 1.9.0
34 out of 183 = 18.5%  - RubiChess 1.5
34 out of 183 = 18.5%  - Texel 1.08a11
32 out of 183 = 17.4%  - Fritz 11 SE (1CPU)
32 out of 183 = 17.4%  - Rodent III 0.273
31 out of 183 = 16.9%  - Senpai 2.0
30 out of 183 = 16.3%  - RofChade 2.1
29 out of 183 = 15.8%  - Nirvanachess 2.4
28 out of 183 = 15.3%  - Pedone 1.9
22 out of 183 = 12.0%  - Arasan 21.3
20 out of 183 = 10.9%  - iCE 3.0 (1CPU)
16 out of 183 = 8.7% - Marvin 3.4.0
13 out of 183 = 7.1% - Bobcat 3.25
9 out of 183 = 4.9% - Amoeba 3.0

Download here - http://dorszcz.blogspot.com
Parent - By Akbarfan (***) Date 2019-10-22 15:29
Many Thanks !
Parent - - By Vinvin (***) Date 2019-10-23 10:12
Hi,
Thanks for all this results !

Where can I find a working version of "Rybka WinFinder 2.2" on Windows 10 ?

Thanks,
Vincent
Parent - - By dorszcz (**) Date 2019-10-23 11:46
Hi!
I don't know, I don't use engines on Windows 10. I run the tests on Windows 7.
Parent - - By Akbarfan (***) Date 2019-11-02 11:49 Edited 2019-11-02 12:22
You're planning a new act3-test?

What kind of test is the act3-test?

I find mate in x not so pleasant. :grin:

I like positions from tournament chess or Corr chess more, i know they are rare and not so easy to find.
Parent - - By dorszcz (**) Date 2019-11-02 13:46
ACT3 is almost ready. I am planning to release ACT3 this month. It will be similar to ACT1-B and ACT2.

For other kind of tests I will use different test names. After releasing ACT3 my next test will contain mostly middle part of game. It is already partly done.
Parent - - By Akbarfan (***) Date 2019-11-06 15:52 Edited 2019-11-06 15:54
In your ACT1-B  Pos.92

[Event "ACT1-B"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "6k1/2p5/6K1/3P2p1/3R4/2p3p1/3P2b1/8 w - - 0 1"]
[PlyCount "43"]

1. d6 cxd6 2. Rc4 Bh3 3. Rxc3 g2 4. Rc1 Bd7 5. Rg1 Bc6 6. Kxg5 Kf7 7. Kf4 Ke6
8. Re1+ Kd5 9. Ke3 Ke5 10. Kf2+ Kd4 11. Re3 Bd5 12. Rg3 Be4 13. Rc3 Bb7 14. Ra3
Be4 15. Kg1 Bc6 16. Rh3 Bd5 17. Rg3 Be4 18. Re3 Bd5 19. Rh3 Be4 20. Rg3 Bd5 21.
Rg4+ Be4 22. Rh4 *

Continuation 4.Rc1 leads to draw.

Only 4.Rg3 mates, e.g. 4...g4  5. Re3 Kf8  6. Re1 g3  (6... g1=Q  7. Rxg1 {57# 7st} Ke7 8. Kf5)
7. Rg1 Ke7  8. Kg5 Ke6  9. Kf4 Kd5  10. Kxg3 {74# 7st} Bf5 (10... Bd7 11. Rxg2 {73#} Ke4 12. Kf2 {72#})
11. Kxg2 {73#} Kd4!  12. Kf2 {72#}
Parent - By dorszcz (**) Date 2019-11-06 18:16
Sometimes some mistakes happen somewhere in the notation, but it does not change the results obtained by the engines. Thanks for pointing it out.
Parent - - By user923005 (****) Date 2019-11-06 20:33
The pv in the analysis record is a correct solution:
6k1/2p5/6K1/3P2p1/3R4/2p3p1/3P2b1/8 w - - acd 277; acs 88; bm d6; c1 "
3.hm ARVES AT (tt); 2008; Akobia=I; (+0130.24g6g8); 1-0; 2008"; c3 "d6"; c6 Arves.06575; ce 32492; dm 138; id "ACT1-B.092"; pm d6; pv d6 cxd6 Rc4 Kf8 Rxc3 g4 Kf6 Kg8 Rxg3 Bf3 Rg1 Kh8 Re1 g3 Kg6 Bc6 Re3 g2 Rg3 Bd5 Kf6 Bf3 Ke6 d5 Kf7 Bh5+ Kf8 Kh7 Rxg2 Bf3 Rh2+ Kg6 d3 Kg5 Kf7 Kf4 Ke6 Kg3 Ra2 Bd1 d4 Kf4;
Parent - - By Akbarfan (***) Date 2019-11-07 13:34 Edited 2019-11-07 13:36

> The pv in the analysis record is a correct solution:
>dm 138; id "ACT1-B.092"; pm d6; pv d6 cxd6 Rc4 Kf8 Rxc3 g4 Kf6 Kg8 Rxg3


.

???? .....
6k1/8/3p1K2/8/6p1/6R1/3P2b1/8 b - - 0 5

I have here:
5.Rxg3 is a mate in 36 (7 men),  NOT dm 138 :eek:

5...Bf3  6.Rg1 #35  :razz::razz:
Parent - By user923005 (****) Date 2019-11-07 16:50
Sure.
I said a correct solution, not an optimal solution.  I expect that many of the long mates can be shorter, especially if the Lomonosov files are used.
Parent - By dorszcz (**) Date 2019-11-07 17:40
Providing the optimal solution for all positions is impossible, it would take unbelievable much time that I do not have. The main purpose of my tests is to test engines. I am more focused to allow only the positions that are solved intentionally by engines, rather than verifying all possible continuations. The files with analysis are just to help you find a solution when your engine can't do it.
Up Topic The Rybka Lounge / Test Positions / Alternative Chess Test 1-B

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill