Discussion of anything and everything relating to chess playing software and machines.
Moderators: hgm, Harvey Williamson, bob
Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
-
vladstamate
- Posts: 161
- Joined: Thu Jan 08, 2009 8:06 pm
- Location: San Francisco, USA
-
Contact:
Post
by vladstamate » Sat May 15, 2010 1:57 am
Hi all,
A new version of my engine, Plisk 023h is now available on the website (
http://sites.google.com/site/pliskchess/).
The results of STS tests and others are on the website. 023h version has improved in the STS tests and now got 583/1000 correct (previous version was 573/1000).
A quick 500 game tournament showed about 54elo improvement.
Code: Select all
Program Elo + - Games Score Av.Op. Draws
1 Plisk023_h : 2027 26 26 500 57.6 % 1973 28.0 %
2 Plisk022_i : 1973 26 26 500 42.4 % 2027 28.0 %
There was a 023 earlier uploaded there that I noticed some people picked up, that had a bug with hashing pawn attacks. The 023h version fixes that.
Kind regards,
Vlad.
-
vladstamate
- Posts: 161
- Joined: Thu Jan 08, 2009 8:06 pm
- Location: San Francisco, USA
-
Contact:
Post
by vladstamate » Sat May 15, 2010 2:07 am
I wanted to mention that the ELO number displayed in the ELOStat output are not real. I just inputed a starting ELO of 2000. I was really interested in the winning percentage and delta elo.
Regards,
Vlad.
-
swami
- Posts: 6535
- Joined: Thu Mar 09, 2006 3:21 am
Post
by swami » Sat May 15, 2010 4:29 am
50 elo improvement is about right estimation for a 10+ improvement out of 1000 in STS.
Keep up the good work, Vlad. I'm happy to see the improvements with recent versions.