ZChess getting much stronger...

Archive of the old Parsimony forum. Some messages couldn't be restored. Limitations: Search for authors does not work, Parsimony specific formats do not work, threaded view does not work properly. Posting is disabled.

Re: ZChess getting much stronger...

Postby Franck ZIBI » 10 Nov 2000, 14:19

Geschrieben von: / Posted by: Franck ZIBI at 10 November 2000 14:19:54:
Als Antwort auf: / As an answer to: ZChess getting much stronger... geschrieben von: / posted by: CL iebert at 10 November 2000 10:08:08:

Hello Christian,
During the last French championship, I discussed with Bernard Brioit
of the latest improvement he has made on its engine ( BBChess
drew with ZChess and Chess Wizard, and had a very nice game with Tiger,
actually BBChess fully deserved a better final ranking).
Bernard told me that he was using R=3 as the reduction factor for Null move,
instead of the more common R=2.
So the first think I did after the tournament was to try it ...
to realize that it had *NO* impact on ZChess ?!
That was very strange because such a change should change the
behavior of the engine.
Then I looked at my Null Move code to find the problem :
a 2 character changes ( adding '//' at the beginnig of a line)
was enough to fix it, and to realize that the new ZChess (Zchess >= 2.2)
becomes suddenly between 2 and 3 time faster at slow time control then the
previous ones (from ZChess 1.51 to ZChess 2.1) !
The tricky think was that this bug had nearly no impact at speed time control,
and most of my testing are done at blitz...
Best regards.
I just did a few tests with an fix-rel. of zchess.
This version plays surprisingly well, much metter than 2.0.
It looks like an increase of may be 80 BfF-Points. Totally on par
with sos and crafty now!
I was told this change was a result of a forgotten NM-Algo?!
Funny thing, isn´t it...
Regards
Christian


ZChess Home Page
Franck ZIBI
 

Re: ZChess getting much stronger...

Postby Paul Bedrey » 10 Nov 2000, 15:22

Geschrieben von: / Posted by: Paul Bedrey at 10 November 2000 15:22:13:
Als Antwort auf: / As an answer to: ZChess getting much stronger... geschrieben von: / posted by: CL iebert at 10 November 2000 10:08:08:
I just did a few tests with an fix-rel. of zchess.
This version plays surprisingly well, much metter than 2.0.
It looks like an increase of may be 80 BfF-Points. Totally on par
with sos and crafty now!
I was told this change was a result of a forgotten NM-Algo?!
Funny thing, isn´t it...
Regards
Christian
I just got done with my match game play and Zchess 2.0 was very strong so your
news is great.

Program Elo + - Games Score Av.Op. Draws
1 crafty : 2506 39 44 208 64.2 % 2405 28.4 %
2 goliath : 2486 41 47 201 61.2 % 2407 21.9 %
3 zchess : 2443 46 39 201 54.5 % 2412 28.4 %
4 sos : 2441 46 40 202 54.2 % 2411 26.2 %
5 comet : 2418 49 39 201 50.5 % 2414 24.4 %
6 anmon : 2393 40 47 202 46.5 % 2417 23.8 %
7 gromit : 2391 41 45 207 45.9 % 2420 23.2 %
8 yace : 2386 40 46 201 45.5 % 2417 25.4 %
9 tcb : 2378 43 45 200 44.2 % 2418 21.5 %
10 phalanx : 2364 44 44 200 42.0 % 2420 23.0 %
11 amy : 2356 47 42 203 40.9 % 2420 17.7 %
Paul Bedrey
 

Re: ZChess getting much stronger...

Postby Aaron » 10 Nov 2000, 18:04

Geschrieben von: / Posted by: Aaron at 10 November 2000 18:04:26:
Als Antwort auf: / As an answer to: Re: ZChess getting much stronger... geschrieben von: / posted by: Franck ZIBI at 10 November 2000 14:19:54:
Hello Christian,
During the last French championship, I discussed with Bernard Brioit
of the latest improvement he has made on its engine ( BBChess
drew with ZChess and Chess Wizard, and had a very nice game with Tiger,
actually BBChess fully deserved a better final ranking).
Bernard told me that he was using R=3 as the reduction factor for Null move,
instead of the more common R=2.
So the first think I did after the tournament was to try it ...
to realize that it had *NO* impact on ZChess ?!
That was very strange because such a change should change the
behavior of the engine.
Then I looked at my Null Move code to find the problem :
a 2 character changes ( adding '//' at the beginnig of a line)
was enough to fix it, and to realize that the new ZChess (Zchess >= 2.2)
becomes suddenly between 2 and 3 time faster at slow time control then the
previous ones (from ZChess 1.51 to ZChess 2.1) !

Wow! That's quite a lot of difference.. Too bad you didn't discover it untill after the tournament..
Looking forward to your next release..
Aaron
 

Re: ZChess getting much stronger...

Postby Pavel » 10 Nov 2000, 18:21

Geschrieben von: / Posted by: Pavel at 10 November 2000 18:21:17:
Als Antwort auf: / As an answer to: Re: ZChess getting much stronger... geschrieben von: / posted by: Aaron at 10 November 2000 18:04:26:
Hello Christian,
During the last French championship, I discussed with Bernard Brioit
of the latest improvement he has made on its engine ( BBChess
drew with ZChess and Chess Wizard, and had a very nice game with Tiger,
actually BBChess fully deserved a better final ranking).
Bernard told me that he was using R=3 as the reduction factor for Null move,
instead of the more common R=2.
So the first think I did after the tournament was to try it ...
to realize that it had *NO* impact on ZChess ?!
That was very strange because such a change should change the
behavior of the engine.
Then I looked at my Null Move code to find the problem :
a 2 character changes ( adding '//' at the beginnig of a line)
was enough to fix it, and to realize that the new ZChess (Zchess >= 2.2)
becomes suddenly between 2 and 3 time faster at slow time control then the
previous ones (from ZChess 1.51 to ZChess 2.1) !

Wow! That's quite a lot of difference.. Too bad you didn't discover it untill after the tournament..
Looking forward to your next release..
yes, indeed speed is a factor in blitz games,
I was wondering does it only speed up in longer time control but not in slower time control?
speeding up in slower time control should effect the game IMO, (ie it will search deeper, and possibly better move, in certain case where it wouldnt with previous versions because of slower search in short time.)
just a thought, i would like to test it on ICS server if you dont mind. (v2.2)
and would also like to have some tests of my own...
thanks
pavs.
Pavel
 

Will ZChess 2.2 be published??

Postby Hannu Wegner » 10 Nov 2000, 22:19

Geschrieben von: / Posted by: Hannu Wegner at 10 November 2000 22:19:42:
Als Antwort auf: / As an answer to: Re: ZChess getting much stronger... geschrieben von: / posted by: Franck ZIBI at 10 November 2000 14:19:54:
Hi Frank,
I would like to know if you will publish the new version of ZChess (2.2).
I think I am not the only one who would like to get it. :-))
Best greetings
Hannu Wegner
Hannu Wegner
 

Re: Will ZChess 2.2 be published??

Postby Aaron » 11 Nov 2000, 03:59

Geschrieben von: / Posted by: Aaron at 11 November 2000 03:59:29:
Als Antwort auf: / As an answer to: Will ZChess 2.2 be published?? geschrieben von: / posted by: Hannu Wegner at 10 November 2000 22:19:42:
Hi Frank,
I would like to know if you will publish the new version of ZChess (2.2).
I think I am not the only one who would like to get it. :-))
Best greetings
Hannu Wegner
If it's a lot better , why not release it?
Aaron
 

Re: ZChess getting much stronger...

Postby CL iebert » 11 Nov 2000, 11:20

Geschrieben von: / Posted by: CL iebert at 11 November 2000 11:20:31:
Als Antwort auf: / As an answer to: Re: ZChess getting much stronger... geschrieben von: / posted by: Franck ZIBI at 10 November 2000 14:19:54:
Hello Christian,
During the last French championship, I discussed with Bernard Brioit
of the latest improvement he has made on its engine ( BBChess
drew with ZChess and Chess Wizard, and had a very nice game with Tiger,
actually BBChess fully deserved a better final ranking).
Bernard told me that he was using R=3 as the reduction factor for Null move,
instead of the more common R=2.
So the first think I did after the tournament was to try it ...
to realize that it had *NO* impact on ZChess ?!
That was very strange because such a change should change the
behavior of the engine.
Then I looked at my Null Move code to find the problem :
a 2 character changes ( adding '//' at the beginnig of a line)
was enough to fix it, and to realize that the new ZChess (Zchess >= 2.2)
becomes suddenly between 2 and 3 time faster at slow time control then the
previous ones (from ZChess 1.51 to ZChess 2.1) !
The tricky think was that this bug had nearly no impact at speed time control,
and most of my testing are done at blitz...
Best regards.
I just did a few tests with an fix-rel. of zchess.
This version plays surprisingly well, much metter than 2.0.
It looks like an increase of may be 80 BfF-Points. Totally on par
with sos and crafty now!
I was told this change was a result of a forgotten NM-Algo?!
Funny thing, isn´t it...
Regards
Christian

Hello Frank,
hope you´re fine. Zchess 2.2 played a few more games
and has up to now round about 60 BfF-Points more than 2.0.
Still amazing this character "//" ;-)
Regards to France
Christian
BfF-Ratings 11-00
SOS 2457 2107
lg2000v28 2454 171
yace 0.23 2453 250
Crafty 17.13 2444 261
zchess 2.22 2442 170
Ikarus V0.18 2425 1047
Comet B22 2422 401
The Crazy Bishop X 2404 188
zchess 2.0 2384 541
CL iebert
 

Re: Will ZChess 2.2 be published??

Postby Franck ZIBI » 11 Nov 2000, 13:03

Geschrieben von: / Posted by: Franck ZIBI at 11 November 2000 13:03:55:
Als Antwort auf: / As an answer to: Re: Will ZChess 2.2 be published?? geschrieben von: / posted by: Aaron at 11 November 2000 03:59:29:

Hello Aaron,
If it's a lot better , why not release it?
Yes why ? :-)
You can download ZChess 2.22 Beta at :
http://myweb.worldnet.net/~fzibi/zchess/beta/zchess.zip
(2.7 MB)
If you have already downloaded ZChess 2.0, you can download only
the new executable (zchess.exe) at :
http://myweb.worldnet.net/~fzibi/zchess/beta/zchexe.zip
(0.2 MB)
I plan to release the next official release within 2 months (to have more
time to test the Temporal Difference Learning)
Note that this is a beta version, so comments and games are even more welcome.
Regards from Paris.


ZChess Home Page
Franck ZIBI
 


Return to Archive (Old Parsimony Forum)

Who is online

Users browsing this forum: MSN [Bot] and 3 guests