Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Winboard-tournament

Author: Heiko Mikala

Date: 03:25:00 12/28/99

Go up one level in this thread


Hi James!

On December 27, 1999 at 21:19:58, James Robertson wrote:

>On December 27, 1999 at 16:52:44, Heiko Mikala wrote:
>
>>Hi Christian!
>>
>>On December 27, 1999 at 08:35:18, Christian Koch wrote:
>>
>>>Hi,
>>>
>>>I started to play a small tournament........
>>>
>>>
>>>2x Celeron 562 MHz (7,0x75),256 MB, Windows2000
>>>
>>>A) Time control: 40 moves in 120 minutes (ponder on!!)
>>>B) Winboard 4.05
>>>
>>>Programs in the tournament
>>>--------------------------
>>>01. Zarkov 4.5L             Hash = 16384   kB   --
>>>02. Crafty 17.05            Hash = 24/04   MB  N-TB
>>>03. Patzer 3.09             Hash = 36      MB  N-TB + BitBases
>>>04. LGoliath 2000 2.0       Hash = 32      MB   --
>>>05. Nimzo 2000b             Hash = 32768   KB   --
>>>06. SOS 03.11.99            Hash = 20      MB  N-TB
>>>07. Insomniac 0.61          Hash = 32/04   MB  N-TB
>>>08. AnMon 5.03              Hash = 32      MB  N-TB
>>
>>That looks *very* interesting!
>>
>>I think this is the first tournament I see, in which programs like
>>Crafty, LGoliath, Nimzo and Zarkov play in their "natural" environment
>>on two computers with ponder on and long time controls. I'm very
>>interested in the results!
>>
>>Btw, do you play the games by hand, or did you find a way to let two
>>Winboards play against each other on two computers? I know this is
>>possible with XBoard, but I couldn't find a solution for Winboard
>>yet...
>>
>>
>>Greetings,
>>
>>Heiko.
>
>It is possible under Fritz, I think.

Yes, but you know about Fritz implementation of the Winboard protocol. The "new"
command before each move is a killer for some programs. For my program for
example this means not only that the hash table is cleared, but also the
evaluation table (2 MB), some other tables, a lot of initialization code is
performed, killer tables and history tables are cleared (and the information
lost), the learning file (1 MB) is read in and 65000 entries are stuffed into
the hash table, the opening book is probed for every move and a lot of other
stuff. I know some programmers have written some code to prevent the worst
things from happening when their program is run inside of Fritz, but I guess a
lot of programmers haven't done that. That's why I said that I find it
interesting to see the programs battle each other in their "natural" environment
(real Winboard).

That said, I should mention that I love Fritz and it's GUI. And I've used it a
lot as a testbed for my own engine and to run tournaments myself.


>I am happy to see Insomniac is in the tournament, too. :)

Yeah, smash them! :))


>BTW, I haven't seen HMChess on FICS lately....

You know, I saw Insomniacs success story during the last months, and thought I'd
better not return to FICS... ;-)

If I do a "pstat HMChess JRCP"(Insomniac) on FICS I get a result of 4-3
currently. I fear that would change drastically, if I'd return, so I decided to
retire from FICS and chess programming and let you become the new king. ;-)

No, just a joke :) I'll continue the work on my program some day, and I'll
return to FICS then. I haven't done much on the program this year (nearly
nothing), but I hope to be able to work hard on it again next year!


See you, greetings,

Heiko.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.