Author: Dann Corbit
Date: 14:50:30 04/05/04
Go up one level in this thread
On April 05, 2004 at 16:43:58, Johannes Zwanzger wrote:
>Hello,
>
>since today you can download my chess programme "Jonny" in version 2.61
>for Windows and Linux. The links are
>
>www.schachclub-forchheim.de/jonny/jonny261_win.zip (Windows)
>
>www.schachclub-forchheim.de/jonny/jonny261_lin.tar (Linux)
>
>Any constructive feedback is welcome. I´m sorry that the readme.txt is
>currently only available in German, ´m going to translate it when I will have
>the time. If anybody plays test series, I would be very interested in the
>results. Especially if somebody finds some better parameter settings, please let
>me now.
>
>Hope everything works, especially with Linux, as I had my first serious contact
>with that OS just two weeks ago...
Readme.txt, crude translation. First, by machine, then a little cleanup in the
top half until I got tired:
*** chess program "Jonny" short description ***
0. Non-liability
1st which is "Jonny"?
2. Installation
2,1 hardware requirements
2,2 installation under Windows
2.2.1 installation under Winboard
2.2.2 installation under Fritz
2,3 installation under Linux
3rd short reference for operation in the console window
4th parameter in the file "settings.ini"
5. developing history
6th tournament participation, estimated play strength
7th other
7,1 from where the name?
7.2 thanks on...
7,3 suggestions, criticism, etc..
0. Non-liability short the usual: Even though "Jonny" was very carefully
developed and tested in detail, I do not take over any liability for damage any
kind, which develops in connection with the use of the program. {Ed.: "Not even
if you should experience and infestation of hamsters in your armpits.}
1. What is "Jonny"? "Jonny" is a Freeware chess programm for Windows and Linux.
However prerequisite condition for use is to forward the file (readme.txt) to to
each and every copy. Although one operate "Jonny" as a console program (see 3.),
I strongly suggest the use of a graphic user interface. Since Jonny is a
Winboard engine, all GUIs, which support this interface, should be suitable; I
have tried it out with "Winboard", "Arena" and "xboard". However I tested under
only these distinct GUIs, here can quite still occur a few problems; also not
all Winboard commands are implemented. If you like to use the "Fritz" GUI, Jonny
can be use there in combination with the Winboard adapter. Details for
installation under (2.2) and/or. (2.3).
2. Installation
2,1 hardware requirements "Jonny" use a Pentium or better with at least 16 MT
RAM (with according to small adjusted Hash and Pawntables) under Windows (9x,
ME, 2000, NT, XP) and/or Linux to running.
2,2 installation under Windows
2.2.1 installation with Winboard Winboard is a freely available GUI, into
which arbitrary chess programs can be inserted, which support the appropriate
interface. A Download link exists at http://www.tim mann.org/xboard.html. After
Winboard installation necessarily one needs to copy only the files JonnyXYZ.exe
(XYZ=Versionsnummer) and Settings.ini contained in the Zip archive in the
Winboard listing. Before beginning it is advisable to adapt the Hash and Pawn
tables amount of the main memory of the own computer (see for this 4.). In the
starting dialogue by Winboard now only still the option "Play is to be selected
against A chess engines or match two engines" and be entered in the first line
the name of the exe file, thus "JonnyXYZ". All further is to be taken then from
the Winboard documentation; however cannot quite function a few things, which
are indicated there, with "Jonny", since like all Winboard interfaces are said
not supported.
2.2.2 installation with Fritz the integration under Fritz runs using the
Winboard adapter, which I may attach with friendly permission of Chessbase here.
For this you must unpack only the entire Zip file preserving the directory
structure into the "Engines" folder of your Fritz installation. In case the
folder structure should get lost, for any reason, here is where each file
belongs: jonnyXYZ.eng: \Engines jonnyXYZ.exe: \Engines\JonnyXYZ settings.ini:
\Engines\JonnyXYZ jonnyXYZ.init: \Engines\JonnyXYZ With the next start of Fritz
then a new engine should be selectable named "JonnyXYZ".
2,3 installation under Linux who in the console to play wants, must the two
files jonnyXYZ and settings.ini into any listing to only unpack. Substantially
more comfortably however xboard, Linux correspondence is to Winboard. After
installation of xboard "Jonny" is merged there with xboard fcp PFAD/jonnyXYZ,
whereby "PATH" is the listing, into which the files unpacked became. More
details supplies the xboard assistance; here the same applies as already in
2.2.1: probably everything that stands there, is not supported also by "Jonny".
3. Short reference for operation in the console window: Since Jonny is actually
not for the operation on console level meant, only the most important commands
are described here: "sd x": the search depth, up to which Jonny counts, sets x
to the value. Possibly with the "time" command made the attitude is then
ignored. "time x": the remaining time sets x to the value (in 1/100 s); "Jonny"
does not update this time, so that she must again in principle be set before
each course, otherwise with each course from the remaining time x is proceeded.
Possibly with the "sd" command made the attitude is then ignored. "edit": Starts
the Edit mode (with the course right available in the last position). From this
the following commands are available: "#": empty the board "Kxy", "Qxy", "Rxy",
"Bxy", "Nxy", "Pxy": king/lady/tower/runner/Springer/farmer of the party, for
which momentarily the course right is set, sets XY on the field. "C": the course
right changes ".": leaves the Edit mode with the course right stopped last; an
examination of the position on schachliche correctness does not take place at
present, here can thus easily crashes be produced... "force": permits it to
enter courses without the engine answers to it. A second input waives the
command again. "undo": back "remove" takes a section: takes two sections back
"xboard": switches into the Winboard mode, in which the announcements are
somewhat more compactly represented to the search and which is switched off
board representation. Repeated input takes the command back. "hard": from this
"Jonny" begins to think after each out-played course automatically on the time
of the opponent. "easy": the "hard" command takes back "quit": terminates the
program "new": a new portion starts "go": does the engine demand for thinking on
"?" : breaks off the computing method; momentarily for course best held one
out-plays "abcd": Input of an own course in the form "e2e4", if the figure from
the field e2 is to pull after e4
4. Parameters in the file "settings.ini" modifications in this file make it
possible to the user to change the play style of "Jonny" within certain borders.
Here a short description of the individual parameters: more //Parameter for
table size of nHashTables: indicates, how "Jonny" uses many
Hashtabelleneintraege. This variable should be adapted the main memory of the
computer (one million entries need about in each case 12 MT memory). If no
power-of-two number is indicated, then "Jonny" rounds off to the nearest.
nPawnTables: dto. for the Pawntables, for instance 5MB needs hundredthousand of
entries. more //Parameter for table size of nHashTables: indicates, how "Jonny"
uses many Hashtabelleneintraege. This variable should be adapted the main memory
of the computer (one million entries need about in each case 12 MT memory). If
no power-of-two number is indicated, then "Jonny" rounds off to the nearest.
nPawnTables: dto. for the Pawntables, for instance 5MB needs hundredthousand of
entries. more //Parameter for figure values pawnValue etc.. : Materials values
for farmer, Springer, etc.. //positionelle parameters endgameMaterial: Total
material, at which for "Jonny" the final game begins; that has among other
things. to the consequence that king security is hardly still considered
rookOn7th: Worth for a tower on the 7. Row pairOfBishops: Bonus for possession
of the pair of runners doublePawn: Departure for each double farmer
isolatedPawn: Departure for each isolated farmer Outpost: Worth for a Vorposten,
a Springer or Laeufer in the board center, the no more by opposing farmers to be
thus attacked can centerPawn: Bonus for center farmer UncastledKing: Departure
for a king NoShortCastling gotten stuck in the center: Departure, if the king
stands in the center and no short castling is more possible NoLongCastling: dto.
for the long castling RightToMove: Bonus for the course right centeredKing: The
more highly this value, the more is paid attention in the final game to
centralization of the king (no absolute value) kingFarFromPassedPawn: Departure,
if the own king stands far far away from an opposing free farmer
rookBehindPassedPawn: Bonus for towers behind own free farmers
rookAgainstNoRook: If a party has still at least a tower on the board and the
opponents not, then it receives these bonuses QNAgainstQB: Bonus for the party
with Springer in the final game D+S - D+L blockedCenterPawn: Departure, if a
center farmer on d2/e2 and/or d7/e7 is blocked by a figure backstandingPawn:
Departure for a backward farmer badBishop: Departure for a bad runner
rookOnOpenLine: Bonus for a tower on an open line rookOnHalfOpenLine: Bonus for
a tower on a half-open line center control: the more highly this value, the more
strongly is weighted center control (no absolute value) spaceAdvantage: the more
highly this value, the more strongly is weighted space advantage (no absolute
value) pawnStormValue: the more highly this value, the more highly farmer storms
are evaluated on the opposing king (no absolute value) more //Parameter for the
Tablebase access TBdistance: Minimum distance up to the quiescent search, so
that the TBs one accesses; at a lower value take place more access, but the knot
throughput breaks possibly a TBCacheSize: Size of the TB Caches in byte TBPath:
Path to the TB listing
5. Developing history "Jonny" is a still relatively recent program, I works on
it since August 2002. However there were already some "predecessors": I still
programmed my first chess program on an old Atari ST under GFA basic; later I
described the weighing function in assembler, brought a clear acceleration,
could however nix to change, which was berauschend the play strength still not
too. It concerned however at that time honestly said also only to prove to me
which I can write in principle a functional chess program. Thus I was for first
content and let the thing first 4-5 years rest. Until then in the Mathe
Leisungskurs the skilled work lined up and came to me the idea, I could take the
development of a chess program nevertheless as topic. There my course chief
anything against it did not have, began I still before the final topic
assignment a basic structure in x86-Assembler to program. This "program" could
already games of chess, remained however unfortunately unfinished: there my
process card chief little of computer science understood, me as a responsible
person a lecturer of the university an attaining was assigned, which much pushed
to write the chess program nevertheless for object orientation in Java. I could
be broad-struck in addition and began again from the front in Java, however I
had rather raped thereby some rules of object-oriented programming from
ignorance, so that at the end a Java program ("Mattjoe") developed, which used
however few of the advantages of the Objektiorientierung and in addition was
quite slow. The play strength was nevertheless already completely properly,
perhaps Elo 2100-2150 on a current computer. After delivery of the skilled work
I made myself to rewrite the program code for speed reasons in C which lasted
not too for a long time; as far as I remember, "Mattjoe" ran thereafter about
3-4 times faster and played also accordingly more strongly. Next I inserted the
most important Winboard commands and had then with the Fritz GUI (more
+Winboardadapter) a beautiful GUI for testing and playing from Enginematches to
the order. A little I developed "Mattjoe further" then still, amount to the play
strength of the last version Elo 2250-2300 probably in such a way. However me
then some fundamental errors are in the beginning (board representation etc..)
been noticeable, the one high knot throughput prevented. Thus I made myself in
the time after my civil service to write from reason on new program in C:
"Jonny" evenly...
6. Tournament participation, estimated play strength so far "Jonny" participated
only in two official tournaments: on the WCCC 2003 in Graz it obtained a result
with 4,0 points from 11 portions, with which I can quite live, whereby in some
portions also more in it was. The second tournament, in which I had
participated, was the 6. Edition annually of the CCT delivered by Volker Richey
in the ICC. Here "Jonny" could cut off very properly and obtained 5.5 points
from 9 portions, under it however a combatless victory. In lightning portions
version is appropriate 2.55 about 70-100 points behind Crafty 19,10, I could
introduce itself, who perhaps catch up "Jonny" with longer considering times
something, but is safe I there naturally not. For extensive testing of long
considering times simply the time is missing to me. Thus careful estimation:
perhaps Elo 2500-2550 in the SSDF on Athlon 1200MHz?!
7. other
7,1 from where the name? When the basic structure of the program was written,
the most difficult problem placed itself to me: how should the child be called
now? I let view somewhat helpless curve mine by the room, until he remained
suddenly hanging on the small plastic steamer, which I had as a child so gladly
- better said at the name plate attached on the cabin: "Jonny"...
7.2 thanks on... - Eugene Nalimov and Andrew Kadatch for the outstanding code
to the Tablebasezugriff and permission for being allowed to use it in "Jonny" -
Chessbase for permission for attaching the Winboard adapter here - mean study
colleagues of the University of Bayreuth, Christoph Groth, which helped me with
the Portierung on Linux with large patience and authority
7,3 suggestions, criticism etc.. I am always grateful for suggestions and
references of all kinds, am it the report of a nose, the statement of salient
weakness in the engine or the like; particularly to be pleased I would test
myself, if someone has time and desire, "Jonny" times with other parameters than
the preset (gladly also lightning and high-speed chess) and then the results to
me if necessary communicates. However it should concern a meaningful number of
portions, by 20 portions can an improvement around perhaps 20-30 Elopunkte not
be proven; when "extremely positive" cutting off in tournament chess with normal
considering time me in addition, such results interest. Since Jonny does not
have at present yet an opening book, I would recommend tests from given
positions. If someone wants to work durably as beta testers, he may contact me
naturally also gladly. Who somewhere uses "Jonny" publicly (tournaments with
publication in the InterNet, ICC, etc.), please I before around short report.
Kirchehrenbach, the 05.04.2004 Johannes Zwanzger (email: johannes.zwanzger@t
online.de)
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.