Andscacs - New version 0.921 with source

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

Moderators: hgm, Harvey Williamson, bob

Post Reply
Damir
Posts: 1785
Joined: Mon Feb 11, 2008 2:53 pm

Re: Andscacs - New version 0.921 with source

Post by Damir » Mon Nov 06, 2017 4:39 pm

cdani wrote:
tttony wrote:Cool!
The source has no license so I guess it's under public domain?
If is the default license, yes. No problem for me.
Damir wrote:I was wondering if you can add parameters to Andscacs like the one Texel has ? I very much like to tweak and tune the engines. The more parameters Andscacs has, the better :) :)
You can do it yourself:
* Decide what eval parameter you want to add. See avaluacio.cpp, in the function carregar_fitxer_avaluacions. For example we take this line
BonusAlfilSuportatPerPeo = FerPun(5, 9);
that means Bonus for bishop supported by pawn.
* in es.cpp add something like this more or less in the line 140:

Code: Select all

afout("option name BonusAlfilSuportatPerPeoMg type spin default 5 min 1 max 100\n");
This is only to show the parameter to the user. The "Mg" is a visual reference to middle game, so we will create a parameter to modify the "5". The "9" is the endgame part of the value.
* in es.cpp add something like this

Code: Select all

if (!_stricmp(name, "BonusAlfilSuportatPerPeoMg")) {
   int tempv= atoi(value);
   if &#40;tempv<= 0&#41;
     tempv= 1;
   if &#40;tempv> 100&#41;
     tempv= 100;
   BonusAlfilSuportatPerPeo = FerPun&#40;tempv, PunF&#40;BonusAlfilSuportatPerPeo&#41;);
&#125;
You can add it for example under this code:

Code: Select all

if (!_stricmp&#40;name, "LoadHashfromFile")) &#123;
   read_hash_binary&#40;);
&#125;	
So this will change the mg part of the value, not changing the current eg part. If you where modifying the eg part, you have to do:

Code: Select all

   BonusAlfilSuportatPerPeo = FerPun&#40;PunI&#40;BonusAlfilSuportatPerPeo&#41;, tempv&#41;;
That's it!
Anything just ask.

A friend of mine have tried and compile Andscacs with the parameter changes you mentioned, but unfortunately he keeps on getting error message when he try and compile Andscacs.
He uses make.exe from SF compilator in src folder.
If you can, can you give some further instructions on how to compile the source, with the changes you just mentioned ?
What compiler are you using etc ? :) :)

pferd
Posts: 92
Joined: Thu Jul 24, 2014 12:49 pm

Re: Andscacs - New version 0.921 with source

Post by pferd » Tue Nov 07, 2017 1:45 pm

Maybe I came across another problem. I tried your binaries for Windows and Linux did a depth 14 search on the start position. I get a different node count for different compiles

Here the result of the Linux version:

Code: Select all

/linux % ./andscacs 
Andscacs 0.921 by Daniel Jose
uci
id name Andscacs 0.921
id author Daniel Jose
option name Ponder type check default false
option name Hash type spin default 128 min 1 max 65536
option name Clear Hash type button
option name NullMove type check default true
option name MultiPV type spin default 1 min 1 max 100
option name Threads type spin default 1 min 1 max 128
option name AlwaysFullPv type check default false
option name Contempt type spin default 0 min -100 max 100
option name WinGroupAffinity type check default false
option name NeverClearHash type check default false
option name HashFile type string default hash.hsh
option name SaveHashtoFile type button
option name LoadHashfromFile type button
uciok
isready
readyok
ucinewgame
position startpos
go depth 14
info depth 1 seldepth 1 score cp 67 nodes 21 nps 21000 tbhits 0 time 1 pv e2e3 
info depth 2 seldepth 6 score cp 15 nodes 73 nps 73000 tbhits 0 time 1 pv e2e3 d7d5 
info depth 2 seldepth 3 score cp 21 nodes 100 nps 100000 tbhits 0 time 1 pv d2d4 d7d5 
info depth 3 seldepth 7 score cp 44 nodes 216 nps 216000 tbhits 0 time 1 pv d2d4 d7d5 e2e3 
info depth 3 seldepth 7 score cp 46 nodes 321 nps 160500 tbhits 0 time 2 pv e2e3 e7e6 d2d4 
info depth 4 seldepth 8 score cp 15 nodes 512 nps 256000 tbhits 0 time 2 pv e2e3 e7e6 d2d4 c7c6 
info depth 5 seldepth 9 score cp 31 lowerbound nodes 973 nps 243250 tbhits 0 time 4 pv e2e3 
info depth 5 seldepth 9 score cp 46 nodes 1135 nps 227000 tbhits 0 time 5 pv e2e3 c7c6 d2d4 d7d5 b1c3 
info depth 6 seldepth 9 score cp 25 nodes 1723 nps 287166 tbhits 0 time 6 pv e2e3 c7c6 d2d4 d7d5 b1c3 g8f6 
info depth 6 seldepth 9 score cp 31 nodes 2106 nps 300857 tbhits 0 time 7 pv g1f3 g8f6 d2d4 d7d5 e2e3 
info depth 7 seldepth 10 score cp 33 nodes 2597 nps 288555 tbhits 0 time 9 pv g1f3 g8f6 d2d4 d7d5 e2e3 c7c6 b1c3 
info depth 8 seldepth 13 score cp 24 nodes 3944 nps 281714 tbhits 0 time 14 pv g1f3 g8f6 d2d4 d7d5 b1c3 f6e4 e2e3 c8g4 
info depth 8 seldepth 11 score cp 31 nodes 5367 nps 357800 tbhits 0 time 15 pv d2d4 d7d5 g1f3 g8f6 c1f4 f6e4 e2e3 
info depth 9 seldepth 15 score cp 47 lowerbound nodes 11458 nps 603052 tbhits 0 time 19 pv d2d4 
info depth 9 seldepth 14 score cp 40 nodes 13966 nps 665047 tbhits 0 time 21 pv d2d4 d7d5 g1f3 b8c6 b1c3 e7e6 e2e3 f8d6 g2g3 
info depth 10 seldepth 21 score cp 34 nodes 29752 nps 901575 tbhits 0 time 33 pv d2d4 d7d5 e2e3 c7c6 c2c4 e7e6 b1c3 g8f6 c4c5 f6e4 h2h4 
info depth 11 seldepth 16 score cp 41 nodes 38340 nps 983076 tbhits 0 time 39 pv d2d4 g8f6 b1c3 d7d5 c1f4 e7e6 e2e3 b8c6 g1f3 f6e4 f1b5 
info depth 12 seldepth 22 score cp 34 nodes 71245 nps 1113203 tbhits 0 time 64 pv c2c4 e7e6 b1c3 d7d5 d2d4 c7c6 e2e3 g8f6 c4c5 f6e4 h2h4 
info depth 13 seldepth 22 score cp 38 nodes 93801 nps 1158037 tbhits 0 time 81 pv c2c4 e7e6 b1c3 d7d5 d2d4 c7c6 e2e3 g8f6 c4c5 f6e4 h2h4 e4c3 b2c3 e6e5 c1b2 
info depth 14 seldepth 24 score cp 25 nodes 154677 nps 1199046 tbhits 0 time 129 pv c2c4 e7e6 b1c3 d7d5 d2d4 c7c6 g1f3 g8f6 c4c5 b7b6 b2b4 c8b7 e2e3 f6e4 
info depth 14 seldepth 22 score cp 41 nodes 260260 nps 1216168 tbhits 0 time 214 pv e2e4 d7d5 e4d5 g8f6 g1f3 f6d5 f1e2 d5f4 e1g1 f4e2 d1e2 g7g6 f1d1 f8g7 d2d4 e8g8 b1c3 
info nodes 268606 nps 1220936 time 220
bestmove e2e4 ponder d7d5

Here the windows version using wine.

Code: Select all

wine andscacs.exe
Andscacs 0.921 by Daniel Jose
uci
id name Andscacs 0.921
id author Daniel Jose
option name Ponder type check default false
option name Hash type spin default 128 min 1 max 65536
option name Clear Hash type button
option name NullMove type check default true
option name MultiPV type spin default 1 min 1 max 100
option name Threads type spin default 1 min 1 max 128
option name AlwaysFullPv type check default false
option name Contempt type spin default 0 min -100 max 100
option name WinGroupAffinity type check default false
option name NeverClearHash type check default false
option name HashFile type string default hash.hsh
option name SaveHashtoFile type button
option name LoadHashfromFile type button
uciok
isready
readyok
position startpos
ucinewgame
go depth 14
info depth 1 seldepth 1 score cp 67 nodes 21 nps 21000 tbhits 0 time 1 pv e2e3 
info depth 2 seldepth 6 score cp 15 nodes 79 nps 13166 tbhits 0 time 6 pv e2e3 d7d5 
info depth 2 seldepth 3 score cp 21 nodes 106 nps 13250 tbhits 0 time 8 pv d2d4 d7d5 
info depth 3 seldepth 7 score cp 44 nodes 209 nps 19000 tbhits 0 time 11 pv d2d4 d7d5 e2e3 
info depth 4 seldepth 8 score cp -15 nodes 389 nps 29923 tbhits 0 time 13 pv d2d4 d7d5 e2e3 d8d6 
info depth 4 seldepth 8 score cp 11 nodes 487 nps 30437 tbhits 0 time 16 pv b1c3 e7e6 e2e3 d7d5 
info depth 4 seldepth 8 score cp 12 nodes 562 nps 31222 tbhits 0 time 18 pv g1f3 d7d5 d2d4 e7e6 
info depth 4 seldepth 8 score cp 28 nodes 822 nps 41100 tbhits 0 time 20 pv c2c3 e7e6 d2d4 b8c6 
info depth 5 seldepth 7 score cp 41 nodes 1052 nps 47818 tbhits 0 time 22 pv c2c3 g8f6 d2d4 d7d5 c1f4 
info depth 6 seldepth 10 score cp 49 nodes 1758 nps 73250 tbhits 0 time 24 pv c2c4 c7c5 d2d4 d7d5 b1c3 
info depth 7 seldepth 16 score cp 34 nodes 4013 nps 143321 tbhits 0 time 28 pv c2c4 c7c6 e2e3 d7d5 c4d5 e7e6 d5e6 c8e6 b1c3 b8d7 
info depth 8 seldepth 12 score cp 20 nodes 6116 nps 191125 tbhits 0 time 32 pv d2d4 d7d5 e2e3 e7e6 c1d2 b8c6 c2c4 
info depth 8 seldepth 13 score cp 50 lowerbound nodes 7133 nps 203800 tbhits 0 time 35 pv e2e4 
info depth 8 seldepth 14 score cp 20 nodes 7772 nps 210054 tbhits 0 time 37 pv d2d4 d7d5 e2e3 e7e6 c1d2 b8c6 c2c4 
info depth 8 seldepth 11 score cp 35 nodes 8376 nps 214769 tbhits 0 time 39 pv g1f3 d7d5 d2d4 e7e6 e2e3 b8c6 b1c3 
info depth 9 seldepth 14 score cp 37 nodes 13091 nps 290911 tbhits 0 time 45 pv g1f3 d7d5 e2e3 e7e6 b1c3 b8d7 d2d4 c7c6 g2g3 
info depth 10 seldepth 14 score cp 28 nodes 16127 nps 329122 tbhits 0 time 49 pv g1f3 d7d5 e2e3 e7e6 b1c3 b8d7 d2d4 c7c6 g2g3 g8f6 
info depth 11 seldepth 18 score cp 32 nodes 25080 nps 432413 tbhits 0 time 58 pv g1f3 d7d5 e2e3 b8c6 d2d4 g8f6 b1c3 a7a6 a2a3 e7e6 f3e5 
info depth 12 seldepth 19 score cp 17 nodes 38754 nps 545830 tbhits 0 time 71 pv g1f3 d7d5 e2e3 b8c6 d2d4 g8f6 f3e5 a7a6 c2c4 c6e5 d4e5 f6d7 c4d5 d7e5 d1d4 d8d6 
info depth 12 seldepth 21 score cp 43 nodes 96071 nps 793975 tbhits 0 time 121 pv c2c4 c7c6 e2e3 d7d5 d2d4 g7g6 b1c3 g8f6 g1f3 d5c4 f1c4 b7b5 c4b3 
info depth 13 seldepth 24 score cp 23 nodes 253287 nps 1017216 tbhits 0 time 249 pv c2c4 e7e6 d2d4 g8f6 g1f3 d7d5 b1c3 f8e7 c4d5 e6d5 e2e3 e8g8 f1d3 f6e4 c3e4 d5e4 d3e4 c7c6 
info depth 14 seldepth 26 score cp 12 nodes 328030 nps 1054758 tbhits 0 time 311 pv c2c4 e7e6 d2d4 g8f6 g1f3 d7d5 b1c3 f8e7 c4d5 e6d5 e2e3 b8c6 f1d3 e8g8 e1g1 c6b4 
info depth 14 seldepth 21 score cp 17 nodes 357157 nps 1056677 tbhits 0 time 338 pv g1f3 e7e6 d2d4 d7d5 e2e3 g8f6 f1e2 f8d6 c2c4 e8g8 e1g1 b8c6 b1c3 b7b6 
info depth 14 seldepth 25 score cp 24 nodes 401461 nps 1062066 tbhits 0 time 378 pv d2d4 d7d5 c1f4 e7e6 e2e3 b8c6 b1c3 f8b4 g1f3 g8f6 a2a3 b4c3 b2c3 f6e4 c3c4 
info nodes 418522 nps 1062238 time 394
bestmove d2d4 ponder d7d5
My Linux compile using gcc from latest sources give yet another result:

Code: Select all

info depth 14 seldepth 22 score cp 41 nodes 260266 nps 1204935 tbhits 0 time 216 pv e2e4 d7d5 e4d5 g8f6 g1f3 f6d5 f1e2 d5f4 e1g1 f4e2 d1e2 g7g6 f1d1 f8g7 d2d4 e8g8 b1c3 
info nodes 268582 nps 1209828 time 222
bestmove e2e4 ponder d7d5
Kind regards

User avatar
cdani
Posts: 2040
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs - New version 0.921 with source

Post by cdani » Tue Nov 07, 2017 8:56 pm

Damir wrote: A friend of mine have tried and compile Andscacs with the parameter changes you mentioned, but unfortunately he keeps on getting error message when he try and compile Andscacs.
He uses make.exe from SF compilator in src folder.
If you can, can you give some further instructions on how to compile the source, with the changes you just mentioned ?
What compiler are you using etc ? :) :)
Visual Studio 2013 for Windows, and gcc for linux. You need to give more details to be able to help.

User avatar
cdani
Posts: 2040
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs - New version 0.921 with source

Post by cdani » Tue Nov 07, 2017 9:06 pm

pferd wrote:Maybe I came across another problem. I tried your binaries for Windows and Linux did a depth 14 search on the start position. I get a different node count for different compiles
Aha. Thanks. I really never worried much on being compatible with other compilers than Visual Studio or other operating systems than Windows, so is very probable that I have added stuff that is not portable or also undefined behavior, like shifting of negative numbers, extensively used by Andscacs. At some point I will try to see about it. I'm sorry that I don't have an immediate solution.

Sure someone has tried a linux version of Andscacs and maybe can say if is comparatively strong than windows one. The Mac one tried by Michael seems that maintains the expected level.

User avatar
cdani
Posts: 2040
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs - New version 0.921 with source

Post by cdani » Tue Nov 07, 2017 9:09 pm

I just saw this website with Andscacs compiled for Android:

http://chesstroid.blogspot.com/

Someone has tried it? It works well?
Thanks

pferd
Posts: 92
Joined: Thu Jul 24, 2014 12:49 pm

Re: Andscacs - New version 0.921 with source

Post by pferd » Wed Nov 08, 2017 2:56 pm

cdani wrote: Sure someone has tried a linux version of Andscacs and maybe can say if is comparatively strong than windows one. The Mac one tried by Michael seems that maintains the expected level.
I played 1000 quick games between the windows (wine) and the linux version today. Whatever causes the difference does not seem to affect the playing strength.

Code: Select all

   # PLAYER     &#58;  RATING  POINTS  PLAYED   (%)
   1 linux      &#58;  2301.6   504.5    1000    50
   2 windows    &#58;  2298.4   495.5    1000    50

User avatar
cdani
Posts: 2040
Joined: Sat Jan 18, 2014 9:24 am
Location: Andorra
Contact:

Re: Andscacs - New version 0.921 with source

Post by cdani » Wed Nov 08, 2017 6:27 pm

pferd wrote:
cdani wrote: Sure someone has tried a linux version of Andscacs and maybe can say if is comparatively strong than windows one. The Mac one tried by Michael seems that maintains the expected level.
I played 1000 quick games between the windows (wine) and the linux version today. Whatever causes the difference does not seem to affect the playing strength.

Code: Select all

   # PLAYER     &#58;  RATING  POINTS  PLAYED   (%)
   1 linux      &#58;  2301.6   504.5    1000    50
   2 windows    &#58;  2298.4   495.5    1000    50
Thanks!! :-)

Carlos777
Posts: 615
Joined: Sun Dec 13, 2009 5:09 pm

Re: Andscacs - New version 0.921 with source

Post by Carlos777 » Thu Nov 09, 2017 11:30 am

Hi Daniel,

It seems the new update is weaker than the 0.92 version, at least under TCEC conditions. Only 0.5 out of 5 games, keeping in mind that in the previous stage Andy remained unbeaten. Unfortunately the last defeat was by illegal move on this position:

[d]6k1/R4Rp1/p2P3p/6nP/p5P1/6BK/4r3/8 w - -

Andscacs 0.921 played Rxf7?

Leo
Posts: 447
Joined: Fri Sep 16, 2016 4:55 pm
Location: Saint Paul, Minnesota USA

Re: Andscacs - New version 0.921 with source

Post by Leo » Thu Nov 09, 2017 11:37 am

Carlos777 wrote:Hi Daniel,

It seems the new update is weaker than the 0.92 version, at least under TCEC conditions. Only 0.5 out of 5 games, keeping in mind that in the previous stage Andy remained unbeaten. Unfortunately the last defeat was by illegal move on this position:

[d]6k1/R4Rp1/p2P3p/6nP/p5P1/6BK/4r3/8 w - -

Andscacs 0.921 played Rxf7?
Also, this stage is with 44 cores. The other engines might be scaling better in this stage.
Brainfish fan. AMD fan.

User avatar
Guenther
Posts: 2246
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Contact:

Re: Andscacs - New version 0.921 with source

Post by Guenther » Thu Nov 09, 2017 11:57 am

Leo wrote:
Carlos777 wrote:Hi Daniel,

It seems the new update is weaker than the 0.92 version, at least under TCEC conditions. Only 0.5 out of 5 games, keeping in mind that in the previous stage Andy remained unbeaten. Unfortunately the last defeat was by illegal move on this position:

[d]6k1/R4Rp1/p2P3p/6nP/p5P1/6BK/4r3/8 w - -

Andscacs 0.921 played Rxf7?
Also, this stage is with 44 cores. The other engines might be scaling better in this stage.
Oops, I didn't realize a different version is playing in stage 2.
Might be the first time an illegal move happened in TCEC?

AFAIK the version can be replaced after such a rare event:
Critical Engine Bugs
In the case of a serious, play-limiting bug (like crashing or interface communication problems, not including losses on time) not discovered during the pre-Season testing, the engine can be updated once per Stage to fix this/these bug/bugs only.
If this update still doesn't fix the problem(s) or if there is no update available, the engine might have the number of cores reduced, have the hash size reduced or have the tablebase access disabled - these changes will remain for the rest of the Stage.
Some debugging work for Daniel.

Guenther
Guenther Simon
http://rwbc-chess.de/

Post Reply