Hi!
So finally I published the new version, 0.88.
http://andscacs.com/andscacs088.zip
Changes:
* Added ability to use Syzygy endgame tablebases.
* Improved various evaluation parameters.
* Add ability to treat heterodox positions.
* Add "Contempt" parameter to try to avoid draw.
* Solved little bugs.
* Speed improvements.
* Improved time management.
I hope it works well for everybody!! Anything just tell me.
Andscacs - New version 0.88
Moderators: hgm, Rebel, chrisw
-
- Posts: 2204
- Joined: Sat Jan 18, 2014 10:24 am
- Location: Andorra
Andscacs - New version 0.88
Daniel José - http://www.andscacs.com
-
- Posts: 325
- Joined: Sat Nov 30, 2013 9:51 am
- Location: Canada
- Full name: Jason Coombs
Re: Andscacs - New version 0.88
Thanks!
-
- Posts: 170
- Joined: Sun Oct 28, 2012 9:46 pm
Re: Andscacs - New version 0.88
Ran a quick 500 game match against 0.88 and 0.872. Looks like some good work!
Code: Select all
Rank Name Elo + - games score oppo. draws
1 andscacs 0.88 21 25 25 500 54% -3 34%
2 NirvanaDev -3 18 18 1000 49% 2 31%
3 andscacs 0.872 -18 26 26 500 48% -3 28%
-
- Posts: 2204
- Joined: Sat Jan 18, 2014 10:24 am
- Location: Andorra
Re: Andscacs - New version 0.88
Thanks for the match!! Terrible strong your nirvana dev also .
I also ran some matches:
I also ran some matches:
Code: Select all
50 seconds + 0.06 xeon E5-2630 v4 @ 2.2 GHz
# PLAYER : RATING ERROR POINTS PLAYED (%)
1 Andscacs 0.88 : 2872.4 3.2 2823.0 5168 54.6%
2 Andscacs 0.872 : 2839.6 3.2 2345.0 5168 45.4%
67 seconds + 0.05 amd fx-8350 @ 4 GHz
# PLAYER : RATING ERROR POINTS PLAYED (%)
1 Andscacs 0.88 : 2868.9 3.5 2547.0 4748 53.6%
2 Andscacs 0.872 : 2843.1 3.5 2201.0 4748 46.4%
120 seconds + 0.1 I7-5820K @ 3.3 GHz
# PLAYER : RATING ERROR POINTS PLAYED (%)
1 Andscacs 0.88 : 2872.0 9.0 330.5 606 54.5%
2 Andscacs 0.872 : 2840.0 9.0 275.5 606 45.5%
Daniel José - http://www.andscacs.com
-
- Posts: 10948
- Joined: Wed Jul 26, 2006 10:21 pm
- Full name: Kai Laskos
Re: Andscacs - New version 0.88
Andscacs seems to be not only scaling well at long TC, but also knowledgeable. It comes third and improved compared to 0.872 in my test suite on quite openings positional test:
10 seconds/position 4 cores:
Significantly above Houdini 4!
10 seconds/position 4 cores:
Code: Select all
Komodo 10.1 128/200
Stockfish dev 124/200
Andscacs 0.88 106/200
Andscacs 0.872 102/200
Houdini 4 99/200
Houdini 4 Tactical 95/200
Hannibal 1.7 93/200
Texel 1.06 91/200
Giraffe 73/200
Fruit 2.1 73/200
Sjaak 1.3.1 52/200
Significantly above Houdini 4!
-
- Posts: 1346
- Joined: Sat Apr 19, 2014 1:47 pm
Re: Andscacs - New version 0.88
You re really good at improving your engine quickly. That's really nice. At this rate you might catch up with Houdini 4 soon and maybe Fire 5 / Shredder dev too.
-
- Posts: 550
- Joined: Thu Apr 24, 2008 9:31 am
- Location: Belgium
Re: Andscacs - New version 0.88
Thank you very much Daniel..will be tested
JP.
JP.
-
- Posts: 1539
- Joined: Thu Mar 09, 2006 2:02 pm
Re: Andscacs - New version 0.88
Thx Daniel
More than 3 month since my last test.
You got 30+Elo in self test. I am curious to see how this one will perform in "the wild"
Ingo
More than 3 month since my last test.
You got 30+Elo in self test. I am curious to see how this one will perform in "the wild"
Ingo
-
- Posts: 2204
- Joined: Sat Jan 18, 2014 10:24 am
- Location: Andorra
Re: Andscacs - New version 0.88
Nice! Thanks for running it on your new test. I will investigate a bit some positions that does not solve fast, to try to get some ideas of improvement. An undetermined part of the improvement comes from the more optimized search, more than the not much added knowledge to this version.Laskos wrote:Andscacs seems to be not only scaling well at long TC, but also knowledgeable. It comes third and improved compared to 0.872 in my test suite on quite openings positional test:
10 seconds/position 4 cores:Code: Select all
Komodo 10.1 128/200 Stockfish dev 124/200 Andscacs 0.88 106/200 Andscacs 0.872 102/200 Houdini 4 99/200 Houdini 4 Tactical 95/200 Hannibal 1.7 93/200 Texel 1.06 91/200 Giraffe 73/200 Fruit 2.1 73/200 Sjaak 1.3.1 52/200
Significantly above Houdini 4!
For this helps a lot having 30 cores running 24/7, having rented a 20 cores server for three months, and of course the help of some testers as I mentioned in previous thread, which I repeat here: Terence Rattray, Johann Waeselynck, Deeds TacKtyK, and one of Adam KostasJJJ wrote:You re really good at improving your engine quickly. That's really nice. At this rate you might catch up with Houdini 4 soon and maybe Fire 5 / Shredder dev too.
Thank you! It will certainly be very interesting to see the results.jpqy wrote:Thank you very much Daniel..will be tested
JP.
Thanks! I will follow your test. On it the improved pondering should add something more, and also of course the 4 pieces Syzygy bases.IWB wrote: More than 3 month since my last test.
You got 30+Elo in self test. I am curious to see how this one will perform in "the wild"
Ingo
Daniel José - http://www.andscacs.com
-
- Posts: 1539
- Joined: Thu Mar 09, 2006 2:02 pm
Re: Andscacs - New version 0.88
I added them, but I personaly doubt that 4 pc tbs add a single elo in gameplay.cdani wrote:
Thanks! I will follow your test. On it the improved pondering should add something more, and also of course the 4 pieces Syzygy bases.
They are fine in Analysis but gameplay ...?
Ingo