Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: new ChestUCI Ver.3.0 - the final version!

Author: Andy Prap

Date: 00:27:28 12/03/03

Go up one level in this thread


Hello!

How you install it in Arena GUI. Like any other engine?


On December 02, 2003 at 12:44:14, F. Huber wrote:



>Hello Chest friends!
>
>Now I´ve also implemented my last ideas for ChestUCI, and so here is
>(as an early Christmas present ;-)) the recent (and probably definite)
>version 3.0 of ChestUCI.
>
>What´s new?
>
>At first a few minor changes:
>- displays immediately, when the search process is completed
>- recognizes itself, if the start position is mate or stalemate
>- output optimal adapted to the capabilities of the Shredder GUI
>- better treatment of special cases with ´DefenderMoves´ and the
>  problem types ´self/help-mate/stalemate´
>- extended input possibilities for parameters in ´Expert mode´
>
>Additionally now I´ve also included my little test file (´Mate.epd´)
>with about 170 (from very easy to extremely hard) mate problems from
>different fora for trying.
>
>Now about the more important enhancements:
>
>1) a new option ´FullMovelist´:
>Normally (i.e. in earlier versions, and with this option switched ON)
>ChestUCI provides the complete move tree as solution for a problem,
>i.e. a solution variant for every possible opponent move.
>In rare cases (at problems with a large search depth or a lot of possible
>moves) this could lead to a very long time to produce the complete move tree,
>although the solution has already be found much earlier.
>
>A small example for this behaviour (and not even the most extreme one!):
>4k3/8/4pPp1/3pP1P1/2pP1K2/2P5/8/8 w - -
>That´s a mate in 18, where my (slow) computer needs about 15 sec to find
>the mate, but after that it takes another 2 min (!) to calculate and
>produce the full solution file (with a size of 24MB!).
>
>By switching OFF this new option, in such cases now you can avoid the
>calculation of the complete move tree and you´ll get the solution
>immediately after the successfull mate searching.
>Nevertheless you´ll still get a complete main line (PV) for every solution,
>but simply not for all possible opponent moves.
>
>2) a further new option ´ShowStatistics´:
>If this option is activated, then after each completed search depth ChestUCI
>shows detailed informations about the previous calculation, which can give
>a rather good indication to the further calculation process.
>
>Such an info line could look like this:
>d=7  t=150.22s [9.30]  n=11957kN [9.56]   Hash: spd=3.55  in=1114866 out=18447
>
>The values for depth(d), time(t) and nodes(n) are probably clear - the two
>numbers in [...] are the so-called ´effective branching factors´ (EBF),
>separately for time and node, and mean the ratio of the value for the current
>search depth to the value for the previous search depth ( in other words: the
>factor which shows, how much more this depth ´costs´ than the last one).
>Larger values indicate an already ´exploding´ search tree, what means that the
>further calculation could take a rather long time.
>
>The following 3 values after ´Hash:´ are the ´speed-up factor´ (spd) by the
>hash table usage (i.e. "how much faster was the calculation WITH vs. WITHOUT
>hash table"), the number of entries written to the hash table and the number
>of deleted (or overwritten) hash table entries.
>These values also provide an indication for the efficiency of the hash table
>administration and show, if the hash size setting is sufficient.
>
>Surely these ´statistics´ are rather ´technical´ informations, but maybe a few
>people are interested in it, since there are a lot of experts here. ;-)
>
>Now such a ´complete´ output of a problem solution by ChestUCI could look
>as following: (a copy from the analyse window of Arena 1.0!)
>
>------------------------------------------------------------------------------
>ChestUCI Ver.3.0  UCI:
>
>CPU: Celeron 400MHz
>FEN: 8/1K6/8/8/8/3N3R/n7/kb5R w - -
>Search for Mate in 10 ...  (Hash=64MB)
>d=1  t=0.05s [1.00]  n=0kN [1.00]   Hash: spd=1.00  in=1  out=0
>d=2  t=0.11s [2.20]  n=0kN [1.00]   Hash: spd=0.99  in=1  out=0
>d=3  t=0.22s [2.00]  n=1kN [14.88]   Hash: spd=0.99  in=34  out=0
>d=4  t=0.38s [1.73]  n=11kN [11.94]   Hash: spd=1.35  in=774  out=0
>d=5  t=2.31s [6.08]  n=120kN [11.24]   Hash: spd=1.92  in=11422  out=0
>d=6  t=16.15s [6.99]  n=1250kN [10.44]   Hash: spd=2.50  in=124360  out=0
> 7   01:29  M7  Rh4
> 7   01:42  M7  Rh8
>d=7  t=150.22s [9.30]  n=11957kN [9.56]   Hash: spd=3.55  in=1114866 out=18447
>Search completed ...  (Time=150.22s)
>Mate in 7 found !  (2 Solutions in 02:30)
> 7   02:31  M7  Rh8 Nb4 Ra8+ Na2 Ka6 Nb4+ Kb5+ Na2 Ka4 Nc1 Rxc1 Ka2 Kb4+
> 7   02:31  M7  Rh4 Nc3 Rc4 Ka2 Rxc3 Bxd3 Rxd3 Kb2 Rh2+ Kc1 Rf3 Kb1 Rf1+
>------------------------------------------------------------------------------
>
>Ok, that´s the end of my detailed description of the new possibilities -
>now I wish you much fun with this new verion, which is again - as usual -
>downloadable from Alex´ UCI-homepage:
>
>http://www.uciengines.de/UCI-Engines/Chest/chest.html
>
>With my best regards,
>Franz.



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.