Plisk 025 released

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

Moderators: hgm, Rebel, chrisw

vladstamate
Posts: 161
Joined: Thu Jan 08, 2009 9:06 pm
Location: San Francisco, USA

Plisk 025 released

Post by vladstamate »

Hi all,

I've made available Plisk 025n which is the version that played in WCRCC and shared 10th place with Arasan, Telepath and Jabba (while having won against both Jabba and Telepath). The only minor change from that is that Plisk will now show evaluations from white's point of view when playing or analyzing.

In STS tests it now gets 13 more positions correct compared to Plisk 024o (596 versus 609) in STS 1-10. The results of all STS tests as well as others are available on the website. Also in direct games, the new version scores around 55% (so about 30-36ELO) compared to previous version.

Windows (64, 32 and popcount), Linux (32) and MacOSX builds are available.

Also this build comes with an updated book (polyglot).

http://sites.google.com/site/pliskchess/

NOTE: The amd64 executable for Windows will run on Intel CPUs as long as popcount is supported - such as Core i7 and so on.

Thank you for trying it and Enjoy,
Vlad.
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: Plisk 025 released

Post by Matthias Gemuh »

vladstamate wrote: Plisk will now show evaluations from white's point of view when playing ...
That is bad for engine-engine tournaments. Automatic adjudication becomes impossible and tester's time is wasted by playing till checkmate.
GUIs expect scores from engine's point of view when playing.

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
vladstamate
Posts: 161
Joined: Thu Jan 08, 2009 9:06 pm
Location: San Francisco, USA

Re: Plisk 025 released

Post by vladstamate »

Hi Matthias,

I was under the impression that always showing eval from white's pov is desirable and most engines do that. Plisk will now show that even when analyzing a position using UCI and Winboard.

I can see the problems for testing though.

Do most engines out there output eval from engine point of view when playing? If that is the case then I'll revert that for the next release.

I just noticed that Stockfish reports scores from engines pov and Crafty from white pov. I am confused...

Regards,
Vlad.
Last edited by vladstamate on Fri Jul 23, 2010 9:21 pm, edited 1 time in total.
vladstamate
Posts: 161
Joined: Thu Jan 08, 2009 9:06 pm
Location: San Francisco, USA

Re: Plisk 025 released

Post by vladstamate »

I will go with what is correct/majority if somebody tells me what is that.

Regards,
Vlad.
Christopher Conkie
Posts: 6073
Joined: Sat Apr 01, 2006 9:34 pm
Location: Scotland

Re: Plisk 025 released

Post by Christopher Conkie »

vladstamate wrote:I will go with what is correct/majority if somebody tells me what is that.

Regards,
Vlad.
I can understand Matthias's point. Sometimes playing to the end discovers previously unknown bugs however. In Arena it is possible to adjudicate games so it is an issue in that way also

Still, this is very nice Vlad, so thank you. It is always nice to get an update to an engine that is a challenge. Plisk improves....very nice.

Chris
User avatar
Graham Banks
Posts: 41428
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Plisk 025 released

Post by Graham Banks »

vladstamate wrote:Hi all,

I've made available Plisk 025n which is the version that played in WCRCC and shared 10th place with Arasan, Telepath and Jabba (while having won against both Jabba and Telepath). The only minor change from that is that Plisk will now show evaluations from white's point of view when playing or analyzing.

In STS tests it now gets 13 more positions correct compared to Plisk 024o (596 versus 609) in STS 1-10. The results of all STS tests as well as others are available on the website. Also in direct games, the new version scores around 55% (so about 30-36ELO) compared to previous version.

Windows (64, 32 and popcount), Linux (32) and MacOSX builds are available.

Also this build comes with an updated book (polyglot).

http://sites.google.com/site/pliskchess/

NOTE: The amd64 executable for Windows will run on Intel CPUs as long as popcount is supported - such as Core i7 and so on.

Thank you for trying it and Enjoy,
Vlad.
Thanks Vlad. 8-)
gbanksnz at gmail.com
User avatar
Matthias Gemuh
Posts: 3245
Joined: Thu Mar 09, 2006 9:10 am

Re: Plisk 025 released

Post by Matthias Gemuh »

vladstamate wrote:I will go with what is correct/majority if somebody tells me what is that.

Regards,
Vlad.
Less than 5 engines from a total of more than 300 show eval from white's POV.

ChessGUI has a special feature to correct such misbehaviour by Crafty and co.

Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
Christopher Conkie
Posts: 6073
Joined: Sat Apr 01, 2006 9:34 pm
Location: Scotland

Re: Plisk 025 released

Post by Christopher Conkie »

Matthias Gemuh wrote:
vladstamate wrote:I will go with what is correct/majority if somebody tells me what is that.

Regards,
Vlad.
Less than 5 engines from a total of more than 300 show eval from white's POV.

ChessGUI has a special feature to correct such misbehaviour by Crafty and co.

Matthias.
Yes, but how do you like the chess it plays? The problem is superficial to advanced GUI's.

Should be a good opponent for BigLion.

Really good when we get these kinds of things to measure ourselves against.

Chris
Christopher Conkie
Posts: 6073
Joined: Sat Apr 01, 2006 9:34 pm
Location: Scotland

Re: Plisk 025 released

Post by Christopher Conkie »

Graham Banks wrote:
vladstamate wrote:Hi all,

I've made available Plisk 025n which is the version that played in WCRCC and shared 10th place with Arasan, Telepath and Jabba (while having won against both Jabba and Telepath). The only minor change from that is that Plisk will now show evaluations from white's point of view when playing or analyzing.

In STS tests it now gets 13 more positions correct compared to Plisk 024o (596 versus 609) in STS 1-10. The results of all STS tests as well as others are available on the website. Also in direct games, the new version scores around 55% (so about 30-36ELO) compared to previous version.

Windows (64, 32 and popcount), Linux (32) and MacOSX builds are available.

Also this build comes with an updated book (polyglot).

http://sites.google.com/site/pliskchess/

NOTE: The amd64 executable for Windows will run on Intel CPUs as long as popcount is supported - such as Core i7 and so on.

Thank you for trying it and Enjoy,
Vlad.
Thanks Vlad. 8-)
Well it beat the living crap outta me a few times.

Much fun.

:)

Chris
vladstamate
Posts: 161
Joined: Thu Jan 08, 2009 9:06 pm
Location: San Francisco, USA

Re: Plisk 025 released

Post by vladstamate »

Christopher Conkie wrote:
Graham Banks wrote:
vladstamate wrote:Hi all,

I've made available Plisk 025n which is the version that played in WCRCC and shared 10th place with Arasan, Telepath and Jabba (while having won against both Jabba and Telepath). The only minor change from that is that Plisk will now show evaluations from white's point of view when playing or analyzing.

In STS tests it now gets 13 more positions correct compared to Plisk 024o (596 versus 609) in STS 1-10. The results of all STS tests as well as others are available on the website. Also in direct games, the new version scores around 55% (so about 30-36ELO) compared to previous version.

Windows (64, 32 and popcount), Linux (32) and MacOSX builds are available.

Also this build comes with an updated book (polyglot).

http://sites.google.com/site/pliskchess/

NOTE: The amd64 executable for Windows will run on Intel CPUs as long as popcount is supported - such as Core i7 and so on.

Thank you for trying it and Enjoy,
Vlad.
Thanks Vlad. 8-)
Well it beat the living crap outta me a few times.

Much fun.

:)

Chris
Thank you for your encouraging words Chris! :)

Regards,
Vlad.