Andscacs, developing version

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

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
cdani
Posts: 2104
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Andscacs, developing version

Post by cdani » Sat Oct 22, 2016 3:21 pm

I publish this version of Andscacs. Is not an official one. If nobody finds anything bad on it I will publish a new official version in a few days.

www.andscacs.com/andscacs087215.zip

I decided to do like this as it has quite a few changes:

* Speed improvements.
* Time management improvements.
* Now is able to manage positions with up to 8 of any piece type, like this mate in 4 one, that previously make it crash:
qqqqqqqk/7q/8/8/8/8/Q7/KQQQQQQQ w - - 0 1
* Added Fathom Syzygy.
* Solved bug with Winboard.
* Solved bug in pondering.
* Added contempt parameter. Just a raw substitution of draw value.
* And as always bug solving and a lot of tuning of evaluation and search.

Is maybe 20-25 elo stronger in self play, without counting Syzygy.

So a more serious and complete version.

I would like to thank Ronald for Syzygy and Basil for Fathom. Also a lot of thanks to some people that have run some or a lot of tests for Andscacs: Terence Rattray, Johann Waeselynck, Deeds TacKtyK, and one of Adam Kostas.

User avatar
Guenther
Posts: 3223
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: Andscacs, developing version

Post by Guenther » Sat Oct 22, 2016 3:32 pm

Thanks Daniel. I guess the bug fixed in WB is related to analysis?

User avatar
cdani
Posts: 2104
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs, developing version

Post by cdani » Sat Oct 22, 2016 3:34 pm

Guenther wrote:Thanks Daniel. I guess the bug fixed in WB is related to analysis?
Yes! The one you told :-) I hope it works well now. Anything just tell me.

User avatar
cdani
Posts: 2104
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs, developing version

Post by cdani » Sat Oct 22, 2016 10:52 pm

I found a bug thanks to Alexander Holzhäuser. It prevented to play with some GUIs like Chessbase or Infinitychess.

I solved it and I updated the zip file. Please, download it again.

www.andscacs.com/andscacs087215.zip

Also is stronger, as this patch seems to work really well also for Andscacs, like 10 elo or more:
https://github.com/loco-loco/Stockfish/ ... 664f69cdc8

Also now it shows the tbhits in all the pv's.

As told, I plan to publish an official version soon, after solving any other issue that appears.

cc2150dx
Posts: 176
Joined: Sat Nov 30, 2013 8:51 am
Location: Canada

Re: Andscacs, developing version

Post by cc2150dx » Sat Oct 22, 2016 11:30 pm

[d]8/8/8/5k2/5p1p/7P/4K1P1/8 w - - 0 1

With 5 man syzygy:

FEN: 8/8/8/5k2/5p1p/5K1P/6P1/8 b - - 1 1

Andscacs:
1/1 00:00 29 29k -0.04 Ke2-f3
2/4 00:00 49 49k -0.12 Ke2-f3 Kf5-e5 Kf3-g4
2/3 00:00 85 85k -0.06 Ke2-d3 Kf5-e5
3/5 00:00 122 122k -0.14 Ke2-d3 Kf5-e5 Kd3-e2
3/4 00:00 146 146k -0.12 Ke2-f3 Kf5-e5 Kf3-g4
4/5 00:00 227 227k +1.11 Ke2-f3 Kf5-e5 Kf3-g4 Ke5-e4
5/5 00:00 779 390k +M3 Ke2-f3 Kf5-e5 Kf3-g4 Ke5-e4
6/6 00:00 1k 388k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
7/7 00:00 1k 409k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
8/8 00:00 1k 465k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
9/9 00:00 2k 397k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
10/10 00:00 2k 456k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
11/11 00:00 2k 523k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
12/12 00:00 2k 596k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
13/13 00:00 3k 547k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
14/14 00:00 3k 629k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
15/15 00:00 4k 589k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
16/16 00:00 4k 657k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
17/17 00:00 4k 713k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
18/18 00:00 5k 789k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
19/19 00:00 5k 726k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
20/20 00:00 6k 789k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
21/21 00:00 6k 853k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
22/22 00:00 6k 918k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
23/23 00:00 7k 857k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
24/24 00:00 7k 932k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
25/25 00:00 8k 885k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
26/26 00:00 9k 959k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
27/27 00:00 9k 1,025k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
28/28 00:00 10k 996k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
29/29 00:00 11k 1,058k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
30/30 00:00 11k 1,032k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
31/31 00:00 12k 1,086k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
32/32 00:00 13k 1,173k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
33/33 00:00 14k 1,160k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
34/34 00:00 15k 1,243k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
35/35 00:00 16k 1,218k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
36/36 00:00 17k 1,318k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
37/37 00:00 19k 1,361k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
38/38 00:00 21k 1,413k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
39/39 00:00 23k 1,419k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
40/40 00:00 25k 1,444k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
41/41 00:00 26k 1,548k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
42/42 00:00 29k 1,623k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
43/43 00:00 32k 1,664k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
44/44 00:00 35k 1,769k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
45/45 00:00 39k 1,835k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
46/46 00:00 43k 1,887k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
47/47 00:00 49k 1,872k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
48/48 00:00 55k 1,980k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
49/49 00:00 62k 2,128k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
50/50 00:00 71k 2,227k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
51/51 00:00 80k 2,359k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
52/52 00:00 92k 2,482k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
53/53 00:00 102k 2,549k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
54/54 00:00 116k 2,646k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
55/55 00:00 134k 2,729k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
56/56 00:00 148k 2,791k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
57/57 00:00 160k 2,860k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
58/58 00:00 176k 2,988k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
59/59 00:00 201k 2,993k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
60/60 00:00 222k 3,035k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
61/61 00:00 242k 3,141k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
62/62 00:00 271k 3,222k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
63/63 00:00 303k 3,295k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
64/64 00:00 344k 3,374k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
65/65 00:00 364k 3,404k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
66/66 00:00 391k 3,399k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
67/67 00:00 409k 3,409k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
68/68 00:00 423k 3,385k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
69/69 00:00 452k 3,375k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
70/70 00:00 475k 3,347k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
71/71 00:00 501k 3,318k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
72/72 00:00 531k 3,298k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
73/73 00:00 561k 3,261k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
74/74 00:00 613k 3,224k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
75/75 00:00 683k 3,132k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
76/76 00:00 749k 3,084k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
77/77 00:00 818k 3,062k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
78/78 00:00 907k 3,055k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
79/79 00:00 1,070k 3,031k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
80/80 00:00 1,180k 3,040k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
81/81 00:00 1,288k 3,052k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
82/82 00:00 1,361k 3,058k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
83/83 00:00 1,510k 3,069k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
84/84 00:00 1,740k 3,080k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
85/85 00:00 1,933k 3,082k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
86/86 00:00 2,181k 3,089k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
87/87 00:00 2,370k 3,074k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
88/88 00:00 2,544k 3,076k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
89/89 00:00 2,900k 3,072k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
90/90 00:01 3,150k 3,073k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
91/91 00:01 3,375k 3,077k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
92/92 00:01 3,992k 3,083k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3
93/93 00:01 5,489k 3,056k +M3 Ke2-f3 Kf5-g5 Kf3-e4 f4-f3

Strange behaviour?

But other than that it's working fine :)
Komodo tester

User avatar
cdani
Posts: 2104
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs, developing version

Post by cdani » Sat Oct 22, 2016 11:39 pm

cc2150dx wrote:[d]8/8/8/5k2/5p1p/7P/4K1P1/8 w - - 0 1

With 5 man syzygy:

FEN: 8/8/8/5k2/5p1p/5K1P/6P1/8 b - - 1 1

Andscacs:
1/1 00:00 29 29k -0.04 Ke2-f3
2/4 00:00 49 49k -0.12 Ke2-f3 Kf5-e5 Kf3-g4
2/3 00:00 85 85k -0.06 Ke2-d3 Kf5-e5
3/5 00:00 122 122k -0.14 Ke2-d3 Kf5-e5 Kd3-e2
3/4 00:00 146 146k -0.12 Ke2-f3 Kf5-e5 Kf3-g4
4/5 00:00 227 227k +1.11 Ke2-f3 Kf5-e5 Kf3-g4 Ke5-e4
5/5 00:00 779 390k +M3 Ke2-f3 Kf5-e5 Kf3-g4 Ke5-e4

Strange behaviour?

But other than that it's working fine :)
I see. It does not change the outcome of the game or of the analysis, but it shows a bad evaluation, mate instead of known win.
I will solve it.
Thanks!!!

cc2150dx
Posts: 176
Joined: Sat Nov 30, 2013 8:51 am
Location: Canada

Re: Andscacs, developing version

Post by cc2150dx » Sun Oct 23, 2016 12:02 am

You're welcome. I like your engine. Keep up the good work :)
Komodo tester

User avatar
Ozymandias
Posts: 1154
Joined: Sun Oct 25, 2009 12:30 am

Re: Andscacs, developing version

Post by Ozymandias » Sun Oct 23, 2016 7:45 am

cdani wrote:* Now is able to manage positions with up to 8 of any piece type, like this mate in 4 one, that previously make it crash:
qqqqqqqk/7q/8/8/8/8/Q7/KQQQQQQQ w - - 0 1
Has this one also been solved?
4rrk1/P4p1p/1R3Pp1/3B2P1/4P1K1/6P1/8/8 w - - 1 53

(I know, more than 8 pieces) It really messed up Andy's performance and prevented it from entering the top 8, in the rapids.

Your engine should be ahead of the discontinued Rybka, in that regard, if no CB representative has contacted you already, about continuing the Fritz tradition, they're wasting time. The private engines still ahead, don't seem to be going anywhere, unlike yours.

User avatar
cdani
Posts: 2104
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs, developing version

Post by cdani » Sun Oct 23, 2016 8:32 am

Ozymandias wrote: Has this one also been solved?
4rrk1/P4p1p/1R3Pp1/3B2P1/4P1K1/6P1/8/8 w - - 1 53

(I know, more than 8 pieces) It really messed up Andy's performance and prevented it from entering the top 8, in the rapids.

Your engine should be ahead of the discontinued Rybka, in that regard, if no CB representative has contacted you already, about continuing the Fritz tradition, they're wasting time. The private engines still ahead, don't seem to be going anywhere, unlike yours.
So I did not explain well. I meant that now should be able to manage strange positions with upto 8 queens for one side + upto 8 knights... and the same for the other side.
I'm not 100% sure about the position you tell, but I was not able to make Andscacs crash with it, so I hope is also solved.

About commercialize Andscacs, I'm not really interested, because then making it will be like working, and now I do, and I want to do it for fun.

User avatar
Ozymandias
Posts: 1154
Joined: Sun Oct 25, 2009 12:30 am

Re: Andscacs, developing version

Post by Ozymandias » Sun Oct 23, 2016 10:21 am

cdani wrote:
Ozymandias wrote: Has this one also been solved?
4rrk1/P4p1p/1R3Pp1/3B2P1/4P1K1/6P1/8/8 w - - 1 53

(I know, more than 8 pieces) It really messed up Andy's performance and prevented it from entering the top 8, in the rapids.

Your engine should be ahead of the discontinued Rybka, in that regard, if no CB representative has contacted you already, about continuing the Fritz tradition, they're wasting time. The private engines still ahead, don't seem to be going anywhere, unlike yours.
So I did not explain well. I meant that now should be able to manage strange positions with upto 8 queens for one side + upto 8 knights... and the same for the other side.
I'm not 100% sure about the position you tell, but I was not able to make Andscacs crash with it, so I hope is also solved.

About commercialise Andscacs, I'm not really interested, because then making it will be like working, and now I do, and I want to do it for fun.
It was well explained, I just took the opportunity to parade that position, to see if you had any idea what it was that happened there, which seems like you don't.

Well, let's hope it keeps being fun for a long time, a lot of us will be grateful.

Post Reply