Author: Heiner Marxen
Date: 07:28:36 10/04/01
Go up one level in this thread
On October 04, 2001 at 04:17:27, Paul wrote:
>On October 03, 2001 at 20:43:26, Heiner Marxen wrote:
>
>>Hello Paul, Hello Leonid!
>>
>>There are 2 different ways to compile Chest, either with a fixed amount of
>>hash table (42 M by default) or with hash memory allocated during run time.
>>Only this second version does accept the -M switch for obvious reasons.
>>It looks like Dann did offer the first kind instead of the second.
>>Sorry.
>
>Ok, thanks for the explanation ... should have figured that out myself ... but
>asking is always easier :)
>
>>If you can compile it yourself, make a "dchest" instead of an "achest",
>>and you can use the -M to use larger hash tables.
>
>Yep, I could ... but chose the Dann-version instead, because I only have Borland
>C++Builder to compile with, and it seems to generate executables that are quite
>a bit slower than the ones coming out of MS VC, or the Intel compiler.
>
>For Pretz I don't care, because I don't have an optimized program anyway ... but
>for yours I did ... :) I'll send Dann an email asking for a recompile if he has
>some spare time left.
>
>>Ha! You managed to use it!
>>I will immediately stop working on a GUI :-))
>
>I could integrate Chest into Pretz's GUI :) ... hey, maybe I'll do that!! Add
>some listboxes for output, etc ... errrr, don't tell anyone!
Hmmm... if it works... wouldn't that be great?
>>You seem to have interpreted the results correctly. Chest gives the complete
>>list of key moves as "bm".
>
>Yep ... the output is pretty clear. One other question: when you tackle one of
>these problems, do you startup Chest multiple times using -Z4, then 5, etc?
>Guess not, right? Probably you use the built in command interpreter?
Yes. For every single problem (mostly from Leonid) I create a single file
(using a text file editor (like notepad?)) with contents like this:
# comment containing posting date & URL
LE
F 8/K6k/8/8/8/8/8/8 w - -
z9w
..
Explanation:
The "#" introduces a comment line (empty lines are also permitted).
The "LE" causes Chest to expect English (p = pawn etc).
The "F" introduces a FEN (with 4 fields).
The "z" is followed by the depth (option -z (lower case) can override that),
and the side to move (must be repeated, here).
The double dot is a logical EOF within the file.
Then I run Chest like this (UNIX command line):
chest319 -sSlllU input_file >> input_file &
(appends the results to the input file).
You may like the simple:
chest319 -r input_file > result_file
Unfortunately, Chest 3.19 does not have an option to print PVs, except
in "batch mode" (option -b). The next release will have such an option.
Sometimes we can afford the complete solution tree:
chest319 -LS input_file > result_file
but that may produce _huge_ outputs, not just the PVs.
>>Explanation above.
>>If you cannot recompile (or get Dann to do so) you have to live with
>>the fixed 42MB version. Sorry. Otherwise, -M350 is what you want.
>>
>>Sorry for answering so late, during 1BELCT my PC was disassembled (monitor
>>was used on the tournament).
>
>Wish I could have seen it ... want to buy a TFT myself one of these days. But
>checked and Berlin meant a 6 hour train ride ... and that's just one way! ;)
May be next year at 2BELCT.
Cheers,
Heiner
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.