Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: 1-0 The Miracle of the Little Beast CTP 14.9 vs SJeng 12.11

Author: Jorge Pichard

Date: 11:10:44 04/07/02

Go up one level in this thread


On April 07, 2002 at 13:15:53, Christophe Theron wrote:

>On April 06, 2002 at 23:39:16, Jorge Pichard wrote:
>
>>On April 06, 2002 at 23:06:26, Uri Blass wrote:
>>
>>>On April 06, 2002 at 22:55:19, Jorge Pichard wrote:
>>>
>>>>On April 06, 2002 at 22:36:00, pavel wrote:
>>>>
>>>>>On April 06, 2002 at 22:27:24, Uri Blass wrote:
>>>>>
>>>>>>On April 06, 2002 at 22:13:17, Uri Blass wrote:
>>>>>>
>>>>>>>On April 06, 2002 at 21:43:08, Jorge Pichard wrote:
>>>>>>>
>>>>>>>>SJeng 12.11 is playing without TB to make it fair since Chess Tiger 14.9 doesn't
>>>>>>>>use TB. Even with TB SJeng would have lost this game regardless. Uri is right
>>>>>>>>this SJeng is a tought opponent with an estimated rating by its author close to
>>>>>>>>2400 in standard time control.2. What is Sjeng's status ?
>>>>>>>>
>>>>>>>>PS: Simply consider this a Pearl of a game, Simply an Amazing Performance!
>>>>>>>>
>>>>>>>>
>>>>>>>>It has successfully beaten players rated over 2500 points, including several
>>>>>>>>titled GM's and IM's. One version of Sjeng even attained the first place in the
>>>>>>>>crazyhouse players list on the Internet Chess Club (ICC), and the first place on
>>>>>>>>the Free Internet Chess Server (FICS). Its crazyhouse rating currently
>>>>>>>>fluctuates around 2300 points.
>>>>>>>>Sjeng is probably a bit weaker in bughouse than it is in crazyhouse. Most
>>>>>>>>Sjeng's on the FICS have a rating of about 2000 points in bughouse, with a top
>>>>>>>>of over 2200 points.
>>>>>>>>
>>>>>>>>For standard chess, Sjeng's rating is probably about 2400 ELO. In suicide and
>>>>>>>>losers chess Sjeng beats all but the very best humans, and will beat
>>>>>>>>non-specialized programs easily.
>>>>>>>
>>>>>>>It seems strange to me that sjeng gets only depth 4 or 5.
>>>>>>>I know that we cannot compare depthes of different programs but
>>>>>>>programs usually get bigger depths and even the weak movei on celeron gets
>>>>>>>usually depth 7-9 at that time control.
>>>>>>>
>>>>>>>Uri
>>>>>>
>>>>>>I can add that sjeng can search clearly deeper in the middle game based on the
>>>>>>following post.
>>>>>>
>>>>>>http://www.it.ro/ccc_search/ccc.php?art_id=198392
>>>>>>
>>>>>>I know that the hardware is better and that sjeng is an older version and the
>>>>>>positions are different but still sjeng could get depth 8 in a few seconds.
>>>>>>
>>>>>>I suspect that your celeron was slowed down by a significant factor so sjeng did
>>>>>>not enjoy from a significant speed advantage that the celeron should give it.
>>>>>>
>>>>>>Uri
>>>>>
>>>>>At this time control (40/40) the depth does seem very odd, even for the most
>>>>>slowest program out there. At this depth, not is it right to compare sjeng with
>>>>>CTP 14.9, Sjeng actually more crippled than CTP in a handheld.
>>>>>
>>>>>pavs
>>>>
>>>>I downloaded the SJeng v12.11 now what version do you suggest?; there doesn't
>>>>seems to be nothing wrong with my Celeron since it is the same one that I have
>>>>been using with Movei From Uri, Bestia, Faile, and Gaviota. If you can suggest
>>>>the best version of SJeng then I will use it. Just give me the website where I
>>>>can find the best Version.
>>>>
>>>>Pichard.
>>>
>>>I believe that 12.11 is good and I do not know the problem(A possible reason
>>>that I can think about is if sjeng is using too much hash tables that the
>>>celeron does not have and I know that using big hash tables that the computer
>>>does not have can cause programs to use the harddisk and be significantly
>>>slower)
>>>
>>>Movei does not use hash tables so there is no problem with it.
>>>Faile's default hash tables is 8 Mbytes so I believe it has no problem.
>>>Maybe Sjeng's default hash tables is too high for your celeron.
>>>
>>>Uri
>>
>>That could be the problem, since the hashtable size by default was set for 12
>>Mbytes. I will replay the same opening up to 4 moves with the Hashtables set for
>>6 Mbytes and this time I will use the endgame tablebase with a cache set for 4
>>MB.
>>
>>Pichard
>
>
>
>If the hash table has been set too big, then you should have heard the
>computer's hard disk going crazy during the game.
>
>Did you notice that?
>
>
>
>    Christophe

Yes I did noticed a strange sound, I just bought more memory for my Solo 1100
for 42 dollars, I will replace it this afternoon from 32 MB to 128MB SODIMM 144
PIN, PC100 PNY.

Pichard



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.