NEW Xpdnt_060821

[David Weller]

http://free-chess.home.comcast.net

[Volker Pittlik]

http://free-chess.home.comcast.net

Unfortunatly it does not compile here (Linux).

The compiler complains:

treewalk.c: In Funktion ?treewalk?:
treewalk.c:83: Fehler: ung?ltige Operanden f?r bin?res -
treewalk.c:97: Fehler: ung?ltige Operanden f?r bin?res -
treewalk.c:111: Fehler: ung?ltige Operanden f?r bin?res -
treewalk.c:114: Fehler: ung?ltige Operanden f?r bin?res -
make: *** [treewalk.o] Fehler 1

Translation: invalid operands for binary -

Example:

Code: Select all

pos -= 106;

Also in helper.h

Code: Select all

#ifdef _MSC_VER // G_OS_WIN32
#include <time.h>
#else
#include <sys/time.h>
//#include <sys/timeb.h>
#endif


sys/timb.h was a comment again. Is there a reason for that?

Volker

[Pablo]

http://free-chess.home.comcast.net

Hello David,
I read you increase fixed hash size of this new version to 64MB. My PC as many others of Winboarders do not have big RAM memories. I wonder if you can compile same version with a hash size of 16 and/or 32Mb.
Cheers,
Pablo Urz?a

[David Weller]

Hi Volker,

There _WAS_ a good reason why I _HAD_ linux installed on my old laptop!!

Sorry. I will look into those errors and get back ...

-David

[David Weller]

Hi Pablo,

Perhaps I _SHOULD_ make it dynamic.

Will get back soon.

-David

[David Weller]

Hi again Volker,

It seems the sys/timeb.h should NOT be commented out. [This adds the timeb structure used in helper.c]

[I seem to remember someone else having to comment this out, I dont remember why ??]



Also, the whole treewalk.c file can be eliminated from the compile, as long as the 2 references in main.c are commented out.
and remove the treewalk.o from Makefile ...

I'll try and learn why it is trouble in the mean time

-David

[Jim Ablett]

I was getting and undefined reference to'show_pv'.

I added 'show_pv.o' to makefile, which was missing.

Jim.

[David Weller]

OK,

treewalk() has been eliminated from this version

also, sys/timeb _IS_ included

and

_WIN32 is now used to conditionally compile

no change to version number however

except a xpdnt_060821a.exe has been included which uses 32M hash

-David

[David Weller]

Hi Jim,

Thanks.

showpv() isnt working anyway [its for debugging] and I removed it also from the sources

-David

[Guenther Simon]

Hi David,

An option for reducing Xpdnts' stats output, which is as talkative
as UCI programs, would be nice too
I noticed that it is impossible to resync a game with the current
broadcast server once it misses a move or stuck in a game with
Xpdnts' participation.

Best regards,
Guenther

[David Weller]

Hi Guenther,

An option for reducing Xpdnts' stats output, which is as talkative
as UCI programs, would be nice too

Heh, and to think how many times i complained about that myself!
It's out-a-here!

I noticed that it is impossible to resync a game with the current
broadcast server once it misses a move or stuck in a game with
Xpdnts' participation.

I'm sorry, I dont understand this??

---

I guess I must seriously consider adding an option.ini !!!

-David

ps - I discovered a couple time control errors which will allow x to lose on time if winning and ahead in time ....

[Volker Pittlik]

Hi Jim,

Thanks.

showpv() isnt working anyway [its for debugging] and I removed it also from the sources

-David

I have added an

Code: Select all

printf("feature myname="Xpdnt-060821"\n");

in respond to the "protover" command to the source.
With all the changes it compiles and runs. Is it OK to start testing
it or is a somehow updated version to expect soon?

Volker

[Guenther Simon]

I noticed that it is impossible to resync a game with the current
broadcast server once it misses a move or stuck in a game with
Xpdnts' participation.

I'm sorry, I dont understand this??

Hi David,

I was talking about the broadcast server from Thomas McBurney,
which is used e.g. by Olivier, Leo, myself and several other people.
It reads the debug file from a running game and allows to watch
it in real time via clients over the net.
There are 3 clients currently, one from Thomas and 2 from Andrew
Fan, inn case you didn't know. With those clients you can watch
broadcasts from above mentioned people.
Currently it uses UDP and thus it sometimes misses packets
which can result in missing a move in the client and the board
isn't updated then. For this there is an restart/reload option,
when the server reads the debug again from the beginning.
If Xpdnt now is one of the participants of such a game the server
isn't able to reload the game in a reasonable time, because of the
enormous amounts of 'stats' lines from Xpdnt.
(Note that, if it cannot be loaded in a certain time, all observers
will be disconnected due to no response too)

Best regards,
Guenther

[David Weller]

Hi Guenther,

Ok, I got it.

The stats have been removed.

I am running a small touney now to check all is well as is ...

http://free-chess.home.comcast.net/xpdnt_060824.exe
http://free-chess.home.comcast.net/xpdnt_060824.zip

-David

[David Weller]

Hi Volker,

I uploaded newest binary and sources to http://free-chess.home.comcast.net

[Volker Pittlik]

Hi Volker,

I uploaded newest binary and sources to http://free-chess.home.comcast.net

Thank you very much. I'm going to test it after the test which is runing right now is finished (tomorrow morning).

Volker

[Fernando JP Vasquez]

Hi David

Seems to be a problem on the link for
Xpdnt_060824a.exe [win32 w/32M hash]
allways give this message

Page URL Not Found!!


The requested page does not exist on this server. The URL you typed or followed is either outdated or inaccurate.


Fernando JP Vasquez

[Volker Pittlik]

Hi Volker,

I uploaded newest binary and sources to http://free-chess.home.comcast.net

I tried to make a test with this version and the version of 060602 under these conditions.

I had to remove the version of 060602 because it lost on time almost 50% of the games.

Version of 060624 does better but around 15% were still lost on time. In addition it exited three times in lost positions (sure mate for the opponent). Therefore a rating does not make much sense. But I look to find the reasons for the losses on time. My computer seems to be good for provoking that errors.

However, from what I have seen the new version seems to be stronger. I'm going to post a game of it vs. Baron in another thread because there is a very strange bug.

Volker