|
1.
Please visit our "little" News-Ticker in
User Files, Engines
selection.
Here can be found information about "NEW" User files / updates too!
Go to
News-Ticker:
Overview ,
Main Page,
Page 1,
Page 2,
Page 3,
Page 4,
Page 5,
Page 6,
Page 7, Page 8,
Page 9
April 29th, 2005
Ktulu 7.0 results by
Klaus Wlotzka
122,
FQ
Klaus Wlotzka started testing with
Ktulu 7.0 for his own ratinglist (called CSS ratinglist). Klaus is playing
on "only one" machine. The rating list based on an AMD Athlon 2800+
system, Fritz 8 GUI with the time control: 10 minutes per game + 10
seconds (Bobby's Fischer time control). With his own rules Klaus cann't test
more as 25 programs, this isn't possible on "only" one system. Each engine have
to play 480 games!
The questions are:
1. Is "only" one system enough to find
out balanced ratings? Today we know that results based on AMD and Intel
processors are quiet different for many of the available engines. After all
results we have from Ktulu 5.1 we can see, that Ktulu 5.1 like Intel processors.
Ktulu 7.0 have very balanced results on Intel
and AMD processors. This could be the reason that Ktulu 5.1 lost
around 40 ELO in the CSS list. A good example can be found on the new
Patriot Forum.
For Patriot 2.0 the results on Intel processors are clearly stronger and the
programmer Vladimir Yelin (Belarus) is working on it to fix the problem.
Patriot 1.3.0 are clearly stronger on AMD processors :-). Gandalf
is clearly stronger on Intel processors and need a little bit more time than the
others. In the rating list of Klaus is Gandalf after 480 games "only" on place
11, in CEGT / ATL with longer time controls on place 4-5 after around 1.600
games. This have nothing to do with the rating list of Klaus, only with the
rules of the tournament directors. Often, computer chess isn't easy to
understand and we all have much more fun if we look on the work others do.
To look only on the own work is very, is
very very boring.
The newest example (Patriot 2.0):
http://f27.parsimony.net/forum67213/messages/4439.htm
2. Today we know that Ktulu 7.0
is around 85 ELO stronger as the preview version Ktulu 5.1. Tested on 10
systems with slower time controls (40 moves in 20 minutes) in CEGT / ATL. This
rating list is in my opinion the reference because this list based on 10 systems
(5x Intel and 5x AMD processors). Ktulu 7.1 is with this playing level on
place 4-9 in the World, that means that we have 6 programs which have a
comparable playing strenght (Gandalf 6.01, TheKing 3.33, Hiarcs 9.0, Ktulu 7.0, Ruffian 2.1.0 ChessTiger
15.0 and ProDeo 1.1,
). The potential Rahman Paidar (programmer
of Ktulu) must have is enormous. Only the really experts we have today saw
it in the preview versions of Ktulu. It's possible that Ktulu is in a really
short time the number 4 or better. Of course, we will inform our customers
automatically for updates of Ktulu! Rahman Paidar is working on an update
and this need time because the programmer of IRAN is testing very exactly each
changes he made.
After 1.600 Ktulu 7.0 games I have from games
of customers, CEGT and ATL I am to 98,5% sure the prognosis on Athlon 2800
can be +115 (+-18) to the preview version Ktulu
5.1 in the list of Klaus. We have to wait of his final result.
His statistics are the
best you can find in WWW since many years.
Klaus spent a lot of time in his statisics and "we users" like this
work.
I wish me Klaus will test in the future chess engines on a second, an
Intel system.
In this case the final results are more balanced and the list is more
interesting.
Furthermore, Klaus have a fantastic service. He
sent after each of the 10 rounds a PDF with Ktulu 7.0 results with his
statistics to me and main tester Jan Kiwitter.
This is really thrilling, my compliment !!
Ktulu 7.0 results by Klaus Wlotzka (actual results after 240 of 480 games)
CSS ratinglist
(you can see the great work Klaus do for us).
April 27th, 2005 / April 29th, 2005
Search testers:
Movei and Anaconda
121,
FQ
The programmers of Movei and Anaconda
search testers. So I will used our webpage to make this informaiton public for
all Gladiators.
If you have interest and time for a good test
please contact the programmer of Movei (Uri Blass, Israel) and Anaconda (Frank
Schneider and Kai Skibbe). After all I know the programmers of
Anaconda will test UCI support and Uri improved his program again.
Anaconda (home of Anaconada, Pocket Grandmaster)
Anaconda (second www address)
Movei (no webpage available).
You can find the information in CCC and WinBoard forum.
Sorry, I don't public mail address on playwitharena.
April 28th, 2005
NON official CEGT / ATL ratinglist
120,
CEGT testers, FQ
Message 120a is cancelt!
April 27th, 2005
Interview with the programmer of Thinker
Lance Perkins (Canada)
119,
AM, FQ
v. 0.92
WB compatible, own GUI, tool for opening books
Home of
Thinker
Interview with Lance Perkins
Hint: Thinker is 12 in ATL-1 Rating-List
01. Frank Quisinsky
Lance, this interview will be not easy for me. I don't have enough information
of yourself and Thinker development. In the latest days I replayed Thinker games
from CEGT and my ATL-1 tournaments. Thinker seems to be a very strong positional
engine. I believe Thinker likes the late middle game more as most of other
stronger amateur chess programs. I believe this can be the "trademark" of
Thinker?
Lance Perkins:
I won't claim that Thinker plays good positional game. The reason is that
Thinker really has very little chess knowledge. The current binary is less than
74K. The evaluation code is very simple. In fact, Thinker performs very poorly
in test positions.
02. Frank Quisinsky
Most of the problems I notice Thinker have in endgames. I made some statistics
and believe Nalimov table bases support will make Thinker 30-40 ELO points
stronger. But only Nalimov support isn't everything to make an engine stronger
in endgame. Is the endgame / Nalimov table base support on your to-do list? In
my opinion the endgame programming is pretty difficult. Do you have time and
knowledge to realize it or do you wish a team with stronger chess players which
can help you. The most of the readers don't have any information how the
programmers added chess knowledge. Could you give us an example of the source
code of Thinker?
Lance Perkins:
Support for endgame table bases has always been in my to-do list. However, it is
way at the bottom of that long list. The reason is that it takes a lot of code
to support them. The alternative would be to simply use someone else's code,
which I'm not really comfortable in doing. It amazes me as to how other new
engines were able to support endgame table bases so easily. What Thinker
currently has is a set of code to deal with different material combinations in
an endgame (an idea that I stole from the endgame recognizer of the Amy engine).
|
Here is an
example of
Thinker's efficient move generator
TMove *CPosition::GeneratePieceMoves
(TMove *pmoves, TBitBoard target)
{
TBitBoard pieces = AllPieces (wtm) ^ Pawns (wtm);
while (pieces)
{
TBitBoard moves;
TSquare from = FindBitAndClear(pieces);
TPieceType pct = PieceType (BoardPiece (from));
switch (pct)
{
case Knight:
moves = KnightAttacks (from) & target;
break;
case Bishop:
moves = BishopAttacks (from) & target;
break;
case Rook:
moves = RookAttacks (from) & target;
break;
case Queen:
moves = QueenAttacks (from) & target;
break;
case King:
moves = KingAttacks (from) & target;
break;
}
if (moves)
{
TMove mv = MvMkFrom (from) + MvMkPiece (pct);
do
{
TSquare to = FindBitAndClear(moves);
*pmoves++ = mv | MvMkTo (to) | CaptureOnSq (to);
}
while (moves);
}
}
return (pmoves);
}
#define EmptySquares() (~(AllPieces(White)|AllPieces(Black)))
TMove *CPosition::GenerateCaptureAndPromotionMoves (TMove *pmoves)
{
pmoves = GeneratePieceMoves (pmoves, AllPieces (Opponent (wtm)));
return GeneratePawnMoves (pmoves, AllPieces (Opponent (wtm)) |
c_PromotionSquares);
}
TMove *CPosition::GenerateNonCaptureMoves (TMove *pmoves)
{
pmoves = GeneratePieceMoves (pmoves, EmptySquares());
return GeneratePawnMoves (pmoves, EmptySquares() |
c_NonPromotionSquares);
}
TMove *CPosition::GenerateAllMoves (TMove *pmoves)
{
pmoves = GenerateCaptureAndPromotionMoves (pmoves);
return GenerateNonCaptureMoves (pmoves);
} |
03. Frank Quisinsky
Against tactical stronger opponents Thinker has sometimes problems after the
first calculated own moves. Your program find his own style. Perhaps a problem
with the opening book? Gandalf seems the most feared opponent. Is this the price
stronger positional engines have to pay? Perhaps I am wrong with my opinion. I
think with faster hardware the positional programs have an advantage. What do
you want to force in your program?
Lance Perkins:
The reality is that Thinker is not at all a good positional engine. There is
just so little chess knowledge for that to be possible. Add to that is the other
simplistic code that Thinker has for opening books. Thinker currently does not
do book learning or game learning. Thinker also does not do transpositions in
opening lines (again, because it is simpler to code that way). I already have
ideas on efficiently implementing such features. I just can't promise on when
they will get delivered.
04. Frank Quisinsky
I can write six questions more or I can make it all in one?! Why is your program
so strong? Thinker is one of the TOP 8 free available WB / UCI engines. Which
programming technique could be the reason for such a playing level Thinker has.
Lance Perkins:
I myself am surprised at the way Thinker performs, considering that there is
really nothing special about it. Its simplicity will baffle many. Its a basic
implementation of what Bruce Moreland and Ed Schröder have on their sites. One
would think that if everyone implements from the same documents, then the
resulting engines would behave similarly. Maybe its because I'm not doing what
others are doing, and in this case, less is more.
05. Frank Quisinsky
Lance, please give us some information about your person: Age, job / profession.
How long are you working in chess programming? What was the first you read about
chess programming? Why do you choose out chess as your hobby? Do you have any
other interests in your free time?
|
Lance Perkins:
I'm a software developer by profession. I have been in this field for
more than 15 years. Chess and chess programming is actually not at the
top of my past times.
There's
photography
http://www.photo.net/photodb/top-rated-photographer-photos?user_id=851055
basketball, table tennis, Halo (XBox),
and music (geocities.com/simplelance). I started playing chess at 5 when
an older cousin taught me the game. Almost everyone in my father's clan
plays chess - some, more serious than others. I fall into the
not-so-serious category since I can never win against my older brother. |
by Lance Perkins |
06. Frank Quisinsky
The history of Thinker is very interesting. I am surprised that a Thinker
version in the past was improved with more than 500 ELO. I hope I have the right
information in my brain. The most people are thinking that a new clone is out
and I followed messages about it. After all I believe you changed an important
part in Thinker and the result was 500 ELO more. Could you give us the
information about this high ELO jumping?
Lance Perkins:
I don't recall that Thinker ever made a 500 Elo climb in one shot. Its a
progress through many versions. Thinker has a long history dating back to when I
first saw Borland's Turbo Chess. That got me started with a chess program which
never made it to the public except to a few friends. The GUI and the engine is
in one program. For chess beginners, it was quite useful as it is. Many years
later, I finally decided to bolt a WinBoard interface to it and to separate the
GUI. I submitted it to Leo's tournament, and that one could barely make it to
TSCP's level. That version is a pure searcher. It has no chess knowledge at all
- it just counts the material values and treats it as the evaluation score. It
does not even have q-search. After adding primitive chess knowledge and q-search,
plus a host of bug fixes, the engine became descent enough for people to take
notice. Search has improved so much since then. On each new version, there is
some new code for pruning, move ordering (mostly from Ed Schröder's documents)
and extensions. Evaluation too has improved on each new version.
07. Frank Quisinsky
At the moment the main topics of computer chess: Chess960, UCI and free
available sources. Do you have any interest to add Chess960 and / or UCI support
in your engine? Do you have a first look on UCI and what do you think about it?
Perhaps you can give Stefan Meyer-Kahlen a hint to improve the UCI protocol?
What is your opinion about free sources? Do you have a look in sources of other
chess programmers? Is it really possible to get new ideas by studying free
sources of other programmers?
Lance Perkins:
I am not a big fan of the UCI protocol. I have posted my thoughts about this on
CCC. It is mainly due to the fact that the engines are treated as if they are
stateless, when in reality, engines are stateful - engines understand a complete
chess game and are not just a position solvers. There are adapters between WB
and UCI, so for now, I don't need to worry about supporting another UI-to-Engine
protocol. The new lines of code that I will write can be better spent on
improving the engine's chess-playing abilities. The main reason why I am able to
maintain Thinker with very little effort is because the code is clean, simple
and well-written. There is nothing redundant in the code. Every line is meant to
do something useful that no other line of code is doing. I believe I don't need
to support another protocol when Thinker already supports one. It buys the
engine nothing, and that's one more thing for me to maintain. Maybe if I am
writing my first chess engine, I probably would consider using UCI. As for open
source codes, I believe is possible to get ideas from them. I myself got the
idea of the end-game recognizer from Amy (it makes for a reasonable end-game
knowledge without having to write the huge code for table bases). However, I
believe that one should always think of better ways of doing things and not be
limited with what they see from available source codes. I've always wondered,
why is it that almost every open source
bitboard engine out there has an SSE/Swap function that looks eerily similar to
that of Crafty?
08. Frank Quisinsiky
I think that the Gladiators / Winboarders / Chessbaslers have to discover your
engine. The reason is that your engine doesn't give much information about the
search. A blank engine analyze output (search depth, nodes, nodes per second,
value and mainline) will give us not that what we like it to see. I am sure the
biggest wish of all the users is to see the complete analyze output. Lance,
could you ...
We can find
the Gladiators in each country ...
Lance Perkins:
Thinker does output its search information. However, it will only output a new
line if the best move changes. The PV display is often cut short by a hash table
match. These two things make Thinker look as if its putting out very little
information.
09. Frank Quisinsky
Very interesting is your own GUI. How important is this development for you? We
have very good available interfaces, like Shredder Classic, WinBoard and of
course Arena. What is the reason to develop an own GUI? A question I often hear.
On the other hand, a GUI is every time welcome, I like it :-)
Lance Perkins:
I have my own GUI because I've always had a GUI. My first chess program is a GUI
and an engine, combined. But the reason why I'm still developing it is so that I
can add features which I myself use frequently. One example is the 'tournament'
mode. You can drag-and-drop programs to it - something you can't do with any GUI
out there. You can limit both the number of moves for the games and the time
that programs can take for a move (so you can end engines that have hung). This
is probably the only GUI that has a true 'skin'. Whereas all other GUIs operate
on rectangular windows, with ThinkerBoard, the display can be any shape. However,
it is still true that the ThinkerBoard GUI lacks many things (e.g., you can't
edit the board, you can't analyze positions, etc.). When I do get the spare
time, support for those features should come in. With ChessThinker, I think I
have introduced new ways of doing things, not just with the GUI. Another is
BookThinker - probably the only one of its kind. I'm even about to add a new
feature to BookThinker that would be another first in the WinBoard community -
watch out for another surprise.
10. Frank Quisinsky
Could you give us information about the future of Thinker? Do you want to
release a new version in the next time? Do you think that your actual version is
improved in playing strength? How do you test the playing strength in self work?
I wonder that in ~ 80% each new version of chess engines are improved in playing
strength. After all I know it's possible that a program comes out with many new
chess knowledge but it's weaker than the preview version. Perhaps you want to
speak about your bag of tricks?
|
 |
Lance Perkins:
There will be a new version of Thinker soon. I just need to be convinced that it
is stable enough. The improvement may be small - I have not verified that yet.
As you have stated, its even possible for new versions to come out weaker than
older versions. I hope that does not happen with Thinker. I have made 64-bit
builds of the engine. It seems to be working well, and I have it tested by
playing at ICC. However, the engine would occasionally lose communication with
WinBoard, causing the engine to be unresponsive. I'm still trying to figure out
the cause of that. I'm also adding support for parallel search. I'm progressing
quite slowly on that one. |
 |
The pics from
question 10 I added. Of course, some nice pics from Greece are in the interview
with Anastasios available. Canada is a very nice country too. A dream from each
European to visit this country.
THANKS Lance
More interviews can be found under:
Reviews, Interviews
April 27th, 2005
Interview with the programmer of AICE
Anastasios Milikas (Greece)
118,
AM, FQ
v. 0.92
UCI / WB / Chess960 compatible
Home of
Aice (Arena Partner)
Interview with Anastasios Milikas in cooperation with Alexander Schmidt
Hint: Aice is one of the most improved chess
engines with great engine support!
01. Alexander Schmidt
AICE is not your only program. When I look on your webpage I have the impression
you are an experienced programmer. What is (beside the time) most important to
create a strong chess engine, to be a talented chess player or to be a talented
programmer?
Anastasios Milikas
In my opinion none of them! To be precise, a minimum programming skill is
required but more importand is basic chess knowledge. Being a descent programmer
helps to do things fast and avoid trivial bugs but nothing more. The most
difficult part in creating a strong chess engine is the collection and analysis
of program's games in order to find its weaknesses and correct them. On the
other hand, after many conversations with fellow strong players (two of them are
GMs!) I realized that it is very difficult for a strong player in this level to
categorize his thoughts in the level of detail required to encode them in a
chess program. They see a middle-game position and they say 'and now white is
positionally lost because this bishop does not belong here in this type of
position' (especially the GMs). They think with patterns and is very difficult
to encode this type of knowledge in a chess program of today.
|
02.
Alexander Schmidt
You are also working "on a templated C++ library for artificial neural
networks", though I don't exactly know what that means :-), I am
wondering if you thought about implementing neural networks in AICE.
Could you further explain an outstanding person like me why neural
networks are not successful in chess programming so far?
Anastasios Milikas
C++ templates is a very elegand and portable way to create functions and
classes when we must use various data types (char, int, float...) to
represent and handle data. My library will be full inlined and will
allow someone to use char or double or anything else as data types for
neuron weights, activations etc easily. My first intention with AICE was
to use neural networks and genetic algorithms in order to optimize its
evaluation weights. Soon I found that this was not productive. It
requires a lot of time to even find acceptable weights. Then I read
papers on the subject that noticed the same findings. Roughly speaking,
neural networks are nice (but far away of perfection) function
estimators not suitable for a chess evaluation function. They might be
useful in other fields such as time management, controling extentions,
controling reductions but I haven't used them in such a way yet. I
currently try to gain in experience in chess programming before continue
with these techniques.
|
Anastasios
Milikas (Greece)
His engine AICE is more and more popular
the users like him and his work he do for us.
|
03.
Alexander Schmidt
What is the future in chess programming, is it limited in refine the
common techniques with small improvements or is it possible that
completely new and different ideas will pass the old techniques?
Anastasios Milikas
I am not experienced enough to answer this question but I can list my
thoughts. I think that there will be another one major strength increase
but not in the spirit of null move pruning. Null move added 150+ elo
points to programs with the simple addition of 5 lines of code. I have
the feeling that the next increase will require a totally new design of
a chess program: A non numeric evaluation function and a very selective
search different from alpha-beta. A fuzzy evaluation function looks
promising and I like the main idea of B* search. Chess programming is a
free time hobby for me and I will consecrate in these fields when I have
more time.
04. Alexander Schmidt
There are so far around 250 engines available and after my impressions I would
say AICE is in the upper midfield. The most engine tournaments are done with the
top engines. Some are using also other engines but it is for a user impossible
to take a close look on all the available engines. Are there special people who
help you with testing and impressions of your engine and are you in general
satisfied with the amount of feedback of the users?
|
 |
Anastasios Milikas
Yes and I thank them all! Leo, Guether, Olivier, Lars, Patrik and others (sorry
if I forgot someone!) I have alot of feedback from them and of course from other
programmers like Richard, Andrew, Volker, Jaime, Reinhard and others and of
course of my fellow strong players (they usually play blitz games against AICE
and they comment 'your program wins lost positions'!). |
05. Alexander Schmidt
One of the most interesting engine tournaments is Leo's WBEC, engines of all
strength levels are fighting with long time control and with pondering. How
often will you visit Leo's site when the next edition is running and how
important are such tournaments for your personal motivation to keep up the good
work on your engine?
Anastasios Milikas
I am visiting Leo's site very often. Leo is a remarkable guy with a lot of
energy and in my opinion he is one of the base stones in the building of amateur
computer chess. Leo's tournaments are for me an excellent motivation for
improving AICE and also a reference for the strength of other engines.
06. Alexander Schmidt
AICE is beside Skaki and PyotrCE the 3rd Greek chess engine. In central and
northern Europe there seem to be much more interest in computer chess. Would you
say this is caused by the mentality of the Greek people or just by the sun and
the sea?
|
 |
Anastasios Milikas
I think that there will be more engines in the near future. A friend of mine (he
is also experieced AI programmer) is working on his own engine in his free time
and I think that in a year from now he will be in place to release his engine. |
07. Frank Quisinsky
Again, thanks Alex for your cooperation. Your forth questions give me the idea
for an additional questions. Anastasios, some programmers loosed their interest
with a small feedback for a free available engine. I assess that more as 500
people are working today in engine programming. What's your personal wish to try
to make the situation better for the programmers?
Anastasios Milikas
More tournaments! Especially for mid and weak engines. I see no point for
testing and testing List, Aristarch, TheKing, Shredder etc. They are too strong
already! Another thing that helped me was the online interaction with other
chess programmers with Tom's live chess viewer. We are watching the games of our
engines and we discussing techniques, ideas etc. It was very nice! Plus that the
guys have intelligence and humor - a nice combination!
08. Frank Quisinsky
I followed the results from many of available WB / UCI engines. After all I can
see Aice 0.86 is around 150-200 ELO improved to older versions. The playing
level for version 0.86 is in my opinion ~ 2.300 ELO. Which changes made your
engine 200 ELO stronger?
|
 |
Anastasios Milikas
In 0.8x I added Fail High Reductions and in combination with Adaptive Nullmove
pruning they made my engine more selective. I also improved the time management
(e.g. easy moves). But I gained more strength by fixing logical bugs: evaluation
assymetries, hash table bugs (hash table and search reductions are not good
fellows!). |
09. Frank Quisinsky
Since March 20th, 2005 is Aice 0.91.1 available. We can be sure that you
improved Aice again?! 100 ELO more and your engine is on TOP 40 level. It's now
very interesting to know more about the ideas you have for the further steps.
The actual development phase is perhaps a quantum jump? This is the most
important time for a development. It's not easy to improved an engine which
around 2.350 ELO. I am sure you are looking for a bigger add-on. MtD, q-search
or another techniques? Give us a tip Anastasios!
Anastasios Milikas
In the 0.9x series I added pondering which improved the strength of AICE. This
is the reason that AICE performed well in Leo's promotion tournament, where
pondering is on. My next step will be to rewrite my evaluation function. My
current evaluation function is a mess of unbalanced chess knowledge! With a
better evaluation and a better static threat evaluation, the FHR will be safer
and more error prune. I think this will give another strength push (I hope). I
wish but I don't think that I'll manage to be in the top 40!
10. Frank Quisinsky
Arena Beta Tester Nicolas Mantides are speaking with very friendly words about
your persons. After a mail I got from him I search for more information about
Aice. I install Aice and after the first games I like your engine. I am very
suprised about your engine support. Not many engines are compatible to WB and
the UCI protocol. The users will like it. With a good protocol support we have
more possibilities. What's your favourite engine protocol and why? Do you have
wishes for protocol programmers?
Anastasios Milikas
I prefer xboard protocol over the UCI. UCI gives me the impression that it was
designed with position analysis in mind. It doesn't come as a suprise that they
didn't initialy offer a command that indicates the engine that a new game was
started. In AICE I use a work around for this by counting the number of moves
after the 'position' command. Less moves or a different starting position means
a new game. Generally speaking I don't like stateless protocols, they feel too
artificial for me. If I had to write a comunication protocol for chess engines,
I think that my protocol will be more like xboard. Also, xboard has the
advantage that is extensible via the 'feature' command.
11. Frank Quisinsky
Another good example for your excellent work is your Chess960 support! I am sure
you have interest to participant on the first Computer Chess960 World
Championship in Mainz (Germany) 2005. Or do you search an operator for this
event?
Anastasios Milikas
Chess960 was an easy addition, so why not? :-) Of course I'd like to be in Mainz
and fortunately the organizers approved my participation so -god first- i'll be
there!
12. Frank Quisinsky
Why you like Chess960 and why you make your engine compatible to Chess960? It's
very interesting to know your reason. I wish me that we can animate other
programmers. After all I know, 17 programmers are working at the moment on
Chess960 support. Since which time you become interest on Chess960? For what a
reason?
Anastasios Milikas
As I ecplained in #11, Chess960 was an easy addition so when I saw the
announcement I decided to make my engine compatible with Chess960. This
tournament will be a great experience for me for several reasons: It will be my
first CC tournament and the super strong engines (like Shredder, Fritz, etc)
will not participate, so the tournament will be more balanced for me. Secondly
it will be a great opportunity to meet the other fellow programmers. And finaly
I'll get the oportunity to visit Germany and buy some accesories for my car!
Just for you, my
suggestion
(for the readers, this pic comes from myself):
This car has about 3.000 ELO so
everything is possible!
Of course with a little bit pruning!
13. Frank Quisinsky
The last question is more off topic. Some years ago I was on holiday in Greece (Creta).
This was a very nice time and I like the friendly Greek people and all the nice
chess boards and pieces I bought. The history of Greece and Egypt are for me the
most interesting. The sightseeing tour of Creta was really one of the important
events in my life. One day I was playing a chess game with the hotel director. I
won this game but the hotel director was a very strong opponent. He was not
happy about loosing the game. He told me ... "Frank, in Greece is very important
to have Olive trees. If you like and you wish to live in Greece I will give you
one of them. Later if you are a good chess player you might gain 100 or more
Olive trees." Each day we can play chess. Thinking about it, you are welcome!"
But this was my latest holiday day and I never understand the sentence from him.
Do you have an idea what he mean with the Olive trees?
|
 |
Anastasios Milikas
This may be an idiom of Creta. We have different idioms in the different places
in Greece. In Creta, the oil production and olive export is a very popular
engagement. So an olive tree means a good start of a native individual. Olive
trees were also the symbol of victory in ancient Greece. So the director might
think that he will loose 100 games from you if you had stayed a little more time
there! |
Frank Quisinsky
Oh, now i understand him. He sayed it with a very overearnest facial
expression.
I was not at any time sure he is joking or not. But in combination with your
words I believe I understand him now ... years later :-)
Thanks Anastasios!
Anastasios Milikas
Thanks Frank and Alex for the interview, I hope I'll meet you in Mainz!
Frank Quisinsky
Last year our brand new car stopped on the middle of the way with problems
in handling (no joke).
So Lisa and myself cann't follow the games TheBaron played. Of course, we will
try it this year again.
ChessTigers (Mainz,
Chess960 World Championship for programs)
More interviews can be found under:
Reviews, Interviews
April 26th, 2005
Ktulu 7.0 on TOP ranking!
117,
RK
15 moves in 40 minutes,
Fritz 8 GUI
Tournament by Reinhard Kruse
kruse-01 (PGN file, 0.34Mb)
Without any
words ...
01. Shredder 9
UCI ** 11 0½ 01 01 1½ ½1 ½1 11 00 01 1½ 11 01 11 ½1 01
00 11 ½1 11 1½ 11 11 11 11 1½ ½1 10 1½ 1½ 11 45.0/62
02. Ktulu 7.0 00 **
½½ 0½ 10 00 ½0 11 01 ½1 1½ 1½ ½1 11 10 01 11 11 11 ½1 ½½ ½0 11 11 1½ 11 11 11 1½
11 ½1 10 42.5/62
03. Toga II 0.93 1½ ½½ ** ½1 0½ 01 ½0 ½0 11
½1 10 ½1 0½ 10 ½1 11 00 1½ 11 10 11 11 11 01 1½ 01 ½½ ½½ 11 11 01 11 40.5/62
04. Junior 9 10 1½
½0 ** 0½ ½½ 01 11 0½ ½1 ½0 11 11 11 1½ ½0 1½ 10 ½1 1½ 10 00 01 ½1 1½ 11 1½ 01 01
11 1½ 11 39.5/62
05. Fritz 8 10
01 1½ 1½ ** 1½ ½½ 0½ 00 1½ ½1 01 00 10 10 ½½ 0½ 1½ ½1 01 ½1 11 11 1½ 1½ 11 11 11
11 11 ½0 ½½ 39.5/62
06. List 512 0½ 11
10 ½½ 0½ ** ½½ ½½ 11 01 11 0½ 1½ 01 10 1½ 01 01 ½0 1½ ½1 ½1 10 01 1½ 1½ 11 1½ 11
0½ ½1 11 39.0/62
07. Chess Tiger 15.0 ½0 ½1 ½1 10 ½½ ½½ ** 0½ 11 00 11 ½1 10 10 01 00
½½ 01 11 ½1 11 1½ ½1 1½ ½0 1½ ½½ ½0 1½ 1½ ½1 11 37.5/62
08. Hiarcs 9 ½0 00
½1 00 1½ ½½ 1½ ** 10 10 ½½ 10 ½1 0½ ½1 1½ 11 00 01 ½1 11 01 1½ 1½ 11 ½0 11 11 10
½½ 11 10 37.0/62
09. SOS 5 for Arena 00 10 00 1½ 11 00 00 01 ** 1½ ½0 11
11 ½0 ½0 11 11 ½1 11 01 ½½ ½0 ½1 ½1 ½½ ½½ ½½ 1½ ½½ 0½ 11 11 35.5/62
10. The King 3.33 11 ½0 ½0 ½0 0½ 10 11 01 0½ ** 00
0½ 10 ½0 ½1 0½ 00 11 0½ 1½ ½0 11 1½ ½1 11 1½ 11 ½1 10 1½ 11 11 35.5/62
11. Pro Deo 1.1 10 0½ 01 ½1 ½0 00 00
½½ ½1 11 ** 00 00 01 ½1 ½0 00 11 ½0 11 ½1 ½½ 11 1½ 01 11 01 ½1 11 11 1½ ½½
34.5/62
12. Ruffian 2.0.2 0½ 0½ ½0 00 10 1½ ½0 01
00 1½ 11 ** 11 11 ½0 ½1 11 11 ½½ 1½ ½½ ½1 ½1 00 10 ½½ 00 10 ½0 1½ 01 1½ 33.0/62
13. Anaconda 2.0.1 00 ½0 1½ 00 11 0½ 01 ½0 00 01 11 00
** ½1 0½ 10 01 ½½ 1½ 00 01 11 11 01 ½½ ½1 1½ ½1 ½1 11 0½ ½1 32.5/62
14. Glaurung 0.2.1 10 00 01 00 01 10 01 1½ ½1 ½1 10 00
½0 ** 01 10 01 11 ½0 ½1 ½1 11 00 01 10 11 0½ 0½ 11 01 1½ 01 32.0/62
15. Gandalf 6.0 00 01 ½0 0½ 01 01 10
½0 ½1 ½0 ½0 ½1 1½ 10 ** 0½ 01 11 11 ½1 01 ½0 00 ½0 ½½ 10 ½1 ½1 01 10 11 11
32.0/62
16. Naum 1.7 ½0 10
00 ½1 ½½ 0½ 11 0½ 00 1½ ½1 ½0 01 01 1½ ** ½0 1½ 11 1½ ½1 11 ½0 00 0½ ½0 10 10 ½0
½½ ½½ 1½ 30.5/62
17. Thinker 4.7a 10 00 11 0½ 1½ 10 ½½ 00 00
11 11 00 10 10 10 ½1 ** 01 10 00 00 10 0½ ½½ 1½ 01 0½ 10 11 01 11 ½1 30.5/62
18. Pharaon 3.2 11 00 0½ 01 0½ 10 10
11 ½0 00 00 00 ½½ 00 00 0½ 10 ** 11 ½1 11 ½½ 11 ½1 1½ 0½ ½½ 01 ½1 ½½ 0½ 0½
28.5/62
19. Aristarch 4.50 00 00 00 ½0 ½0 ½1 00 10 00 1½ ½1 ½½
0½ ½1 00 00 01 00 ** 11 ½1 0½ 1½ ½1 11 0½ ½1 01 11 ½½ ½½ 11 28.5/62
20. Slow Blitz WV ½0 ½0 01 0½ 10 0½ ½0 ½0 10 0½ 00
0½ 11 ½0 ½0 0½ 11 ½0 00 ** 11 0½ 10 10 1½ 01 0½ 1½ ½1 ½0 ½1 11 27.5/62
21. Nimzo 8 00
½½ 00 01 ½0 ½0 00 00 ½½ ½1 ½0 ½½ 10 ½0 10 ½0 11 00 ½0 00 ** 1½ 0½ 11 1½ 00 01 ½½
11 11 1½ 11 27.5/62
22. Spike 0.9 0½ ½1
00 11 00 ½0 0½ 10 ½1 00 ½½ ½0 00 00 ½1 00 01 ½½ 1½ 1½ 0½ ** 01 01 0½ 0½ 10 ½½ 0½
½1 11 11 27.0/62
23. AnMon 5.51 00 00 00
10 00 01 ½0 0½ ½0 0½ 00 ½0 00 11 11 ½1 1½ 00 0½ 01 1½ 10 ** 01 ½½ 10 01 11 ½1 ½0
11 ½½ 26.5/62
24. Ktulu 5.1 00 00 10
½0 0½ 10 0½ 0½ ½0 ½0 0½ 11 10 10 ½1 11 ½½ ½0 ½0 01 00 10 10 ** 1½ 01 0½ 10 ½½ 00
1½ 10 25.5/62
25. Deep Sjeng 1.6 00 0½ 0½ 0½ 0½ 0½ ½1 00 ½½ 00 10 01
½½ 01 ½½ 1½ 0½ 0½ 00 0½ 0½ 1½ ½½ 0½ ** 1½ 01 11 10 ½½ 1½ 01 25.5/62
26. Crafty Cl. 2004 00 00 10 00 00 0½ 0½ ½1 ½½ 0½ 00 ½½ ½0 00
01 ½1 10 1½ 1½ 10 11 1½ 01 10 0½ ** ½1 01 00 10 ½0 ½1 25.5/62
27. Patriot 1.3.0 0½ 00 ½½ 0½ 00 00 ½½ 00
½½ 00 10 11 0½ 1½ ½0 01 1½ ½½ ½0 1½ 10 01 10 1½ 10 ½0 ** 0½ 0½ 01 01 01 25.0/62
28. DanChess CCT7 ½0 00 ½½ 10 00 0½ ½1 00 0½ ½0 ½0
01 ½0 1½ ½0 01 01 10 10 0½ ½½ ½½ 00 01 00 10 1½ ** 11 01 ½½ ½½ 24.5/62
29. Delfi 4.5 UCIa 01 0½ 00 10 00 00 0½ 01 ½½ 01 00 ½1
½0 00 10 ½1 00 ½0 00 ½0 00 1½ ½0 ½½ 01 11 1½ 00 ** ½1 1½ 11 24.0/62
30. GLChess 3.00.34 0½ 00 00 00 00 1½ 0½ ½½ 1½ 0½ 00 0½ 00 10
01 ½½ 10 ½½ ½½ ½1 00 ½0 ½1 11 ½½ 01 10 10 ½0 ** 00 ½½ 23.0/62
31. Naum 1.6 0½ ½0
10 0½ ½1 ½0 ½0 00 00 00 0½ 10 1½ 0½ 00 ½½ 00 1½ ½½ ½0 0½ 00 00 0½ 0½ ½1 10 ½½ 0½
11 ** 11 22.0/62
32. Movei v0.08.295 00 01 00 00 ½½ 00 00 01 00 00 ½½ 0½ ½0 10
00 0½ ½0 1½ 00 00 00 00 ½½ 01 10 ½0 10 ½½ 00 ½½ 00 ** 15.0/62
Ktulu 7.0 made 15.5 points
more as the preview version Ktulu 5.1 !!
Program Elo + - Games Score Av.Op. Draws
1 Shredder 9 UCI : 2750 88 84 62 72.6 % 2581 19.4 %
2 Ktulu 7.0 : 2717 81 79 62 68.5 % 2582 24.2 %
3 Toga II 0.93 : 2693 78 76 62 65.3 % 2583 27.4 %
4 Fritz 8 : 2681 75 74 62 63.7 % 2583 30.6 %
5 Junior 9 : 2681 77 76 62 63.7 % 2583 27.4 %
6 List 512 : 2675 74 73 62 62.9 % 2583 32.3 %
7 Chess Tiger 15.0 : 2658 71 70 62 60.5 % 2584 37.1 %
8 Hiarcs 9 : 2652 75 74 62 59.7 % 2584 29.0 %
9 The King 3.33 UCIa : 2635 76 75 62 57.3 % 2584 27.4 %
10 SOS 5 for Arena : 2635 72 71 62 57.3 % 2584 33.9 %
11 Pro Deo 1.1 : 2624 75 75 62 55.6 % 2585 27.4 %
12 Ruffian 2.0.2 : 2608 72 72 62 53.2 % 2585 32.3 %
13 Anaconda 2.0.1 : 2602 75 75 62 52.4 % 2585 27.4 %
14 Gandalf 6.0 : 2597 76 76 62 51.6 % 2586 25.8 %
15 Glaurung 0.2.1 : 2597 81 81 62 51.6 % 2586 16.1 %
16 Thinker 4.7a UCIa : 2581 80 80 62 49.2 % 2586 17.7 %
17 Naum 1.7 : 2581 69 70 62 49.2 % 2586 37.1 %
18 Pharaon 3.2 : 2559 73 73 62 46.0 % 2587 30.6 %
19 Aristarch 4.50 : 2559 73 73 62 46.0 % 2587 30.6 %
20 SlowChess Blitz WV : 2548 73 74 62 44.4 % 2587 30.6 %
21 Nimzo 8 : 2548 73 74 62 44.4 % 2587 30.6 %
22 Spike 0.9 : 2542 72 73 62 43.5 % 2587 32.3 %
23 AnMon 5.51 UCIa : 2537 77 78 62 42.7 % 2588 24.2 %
24 Deep Sjeng 1.6 : 2526 68 68 62 41.1 % 2588 40.3 %
25 Ktulu 5.1 : 2526 75 76 62 41.1 % 2588 27.4 %
26 Crafty Classic 2004 UCIa : 2526 75 76 62 41.1 % 2588 27.4 %
27 Patriot 1.3.0 : 2520 72 73 62 40.3 % 2588 32.3 %
28 DanChess CCT7 UCIa : 2514 72 73 62 39.5 % 2588 33.9 %
29 Delfi 4.5 UCIa : 2509 76 78 62 38.7 % 2588 25.8 %
30 GreenLightChess 3.00.34 UCIa : 2497 71 72 62 37.1 % 2589 35.5 %
31 Naum 1.6 : 2485 71 72 62 35.5 % 2589 35.5 %
32 Movei v0.08.295 UCIa : 2394 81 84 62 24.2 % 2592 25.8 %
You can order the number 1 Shredder 9.0 on the webpage of Stefan Meyer-Kahlen
http://www.shredderchess.com/
= Price 49,95 EUR with a nice GUI ... clearly the number 1
You can order the number 2 Ktulu 7.0 by Rahman Paidar on Gladiatorshop
http://www.gladiatorshop.com/
= Price 19,95 EUR
The number three Toga 0.93 based on Fruit 2.0 is free available.
Read more about Toga (by Exacta Chess)
April 26th, 2005
Patriot 2.0 Forum open
116,
FQ
Message 116 is cancelt!
April 26th, 2005
ATL-2, final results
115,
FQ
1. All games for each of the 135 matches
(3x 45 matches) with 20 games can be donwload on ATL League.
Included are PGN and Arena log files.
2. I played ATL-2 on three machines with the same time control 40 moves in 20
minutes.
Pentium IV 2.67 Mobile, Athlon64 3800, Dual Xeon 2.8.
More information about the rules can be found on the detail page.
ATL League
Shredder 9.0 always with 2.750, calculation with ELOstat 1.3! Rating
1: Games from ATL-2
Systems 1 - 3
After the last engine Ktulu 7.0
Program Elo + - Games Score Av.Op. Draws
1 Shredder 9.0 : 2750 26 26 540 71.4 % 2591 29.8 %
2 Gandalf 6.01 : 2653 25 24 540 57.3 % 2602 31.3 %
3 Ktulu 7.0 : 2636 24 24 540 54.7 % 2603 31.7 %
4 TheKing 3.33 Schumacher : 2630 24 24 540 53.8 % 2604 30.9 %
5 ProDeo 1.1 : 2623 24 24 540 52.6 % 2605 31.5 %
6 Ruffian 2.1.0 : 2619 24 24 540 52.0 % 2605 31.5 %
7 SOS 5.1 for Arena : 2566 25 25 540 43.5 % 2611 30.4 %
8 SlowChess Blitz WV : 2565 25 25 540 43.3 % 2611 28.5 %
9 Spike 0.9 : 2529 25 25 540 37.8 % 2615 31.9 %
10 Pharaon 3.2 : 2500 26 26 540 33.5 % 2619 28.5 %
Rating 2: ATL-2 System 1 (Pentium IV
2.67 Mobile)
After the last engine Ktulu 7.0
Program Elo + - Games Score Av.Op. Draws
1 Shredder 9.0 : 2750 45 44 180 70.6 % 2599 30.0 %
2 Gandalf 6.01 : 2668 43 43 180 58.6 % 2608 30.6 %
3 Ktulu 7.0 : 2650 43 43 180 55.8 % 2610 28.3 %
4 TheKing 3.33 Schumacher : 2645 43 43 180 55.0 % 2610 28.9 %
5 ProDeo 1.1 : 2640 42 42 180 54.2 % 2611 31.7 %
6 Ruffian 2.1.0 : 2612 44 44 180 49.7 % 2614 26.1 %
7 SlowChess Blitz WV : 2581 44 44 180 44.7 % 2618 25.0 %
8 SOS 5.1 for Arena : 2561 44 44 180 41.7 % 2620 27.8 %
9 Spike 0.9 : 2532 44 44 180 37.2 % 2623 27.8 %
10 Pharaon 3.2 : 2500 44 44 180 32.5 % 2627 30.6 %
Rating 3: ATL-2 System 2 (Athlon64 3800)
After the last engine Ktulu 7.0
Program Elo + - Games Score Av.Op. Draws
1 Shredder 9.0 : 2750 47 46 180 72.2 % 2584 26.7 %
2 Gandalf 6.01 : 2657 42 42 180 58.9 % 2594 33.3 %
3 Ruffian 2.1.0 : 2637 41 41 180 55.8 % 2596 35.0 %
4 Ktulu 7.0 : 2632 40 40 180 55.0 % 2597 38.9 %
5 TheKing 3.33 Schumacher : 2613 42 41 180 51.9 % 2599 33.9 %
6 ProDeo 1.1 : 2609 42 41 180 51.4 % 2600 33.9 %
7 SlowChess Blitz WV : 2553 42 42 180 42.5 % 2606 32.8 %
8 SOS 5.1 for Arena : 2550 42 42 180 41.9 % 2606 33.9 %
9 Spike 0.9 : 2508 42 43 180 35.6 % 2611 33.3 %
10 Pharaon 3.2 : 2502 45 45 180 34.7 % 2611 27.2 %
Rating 4: ATL-2 System 3 (Dual Xeon 2.8)
After the last engine Ktulu 7.0
Program Elo + - Games Score Av.Op. Draws
1 Shredder 9.0 : 2750 44 43 180 71.4 % 2591 32.8 %
2 Gandalf 6.01 : 2634 43 43 180 54.4 % 2603 30.0 %
3 TheKing 3.33 Schumacher : 2634 43 43 180 54.4 % 2603 30.0 %
4 Ktulu 7.0 : 2627 43 43 180 53.3 % 2604 27.8 %
5 ProDeo 1.1 : 2620 43 43 180 52.2 % 2605 28.9 %
6 Ruffian 2.1.0 : 2610 42 42 180 50.6 % 2606 33.3 %
7 SOS 5.1 for Arena : 2587 43 43 180 46.9 % 2609 29.4 %
8 SlowChess Blitz WV : 2561 44 44 180 42.8 % 2612 27.8 %
9 Spike 0.9 : 2547 41 42 180 40.6 % 2613 34.4 %
10 Pharaon 3.2 : 2498 45 45 180 33.3 % 2619 27.8 %
01. Shredder
9.0
by
Stefan Meyer-Kahlen (Germany)
Achilles (Shredder) lost only 2 of 27 20-game-matches vs.
Ktulu 7.0. On Dual Xeon with ponder = on Shredder have the strongest
results.
Shredder by Stefan Meyer-Kahlen is clear the number 1 in the
World.
Interview with Stefan
02. Gandalf
6.01
by
Steen Suurballe (Denmark)
97 ELO to Shredder 9.0. This is lot but Gandalf
is still the number two of all available UCI engines and the strongest available
WB engine. Steen Suurballe answered in the interview that he need again 100 ELO for fighting against Achilles.
This is to 100% correct!
Interview with Steen
03. Ktulu 7.0
by
Rahman Paidar (Iran)
Very stable results on three different
processors. Ktulu 7.0 can be the sensation of the year 2005. But the
other programmers are working too. A good example is Patriot 2.0. My
congratulation to Rahman Paidar. Without book learning and tablebase
support is such an high rating just a sensation. Perhaps the program with the
most potential for the future? At the moment Rahman is working on an update. This should be available in around 4-8 weeks. Our customers will get an
automatically mail. Ktulu seems the World's WinBoard engine number two and UCI
engine number three (speaking from available chess programs). The endgame looks
better as from other commercial programs with have tablebase (Nalimov) support.
The tactic is very good. Perhaps Rahman have to improved the positional playing
strength.
Interview with Rahman
04. TheKing
3.33 Schumacher
by
Johan de Koning (The Netherlands)
I played with the strongest settings I found. I
am not TheKing expert but I am sure that this setting by Dr. Alexander
Schumacher are excellent. TheKing don't support tablesbases and the
strong part is the
positional playing style. TheKing likes endgames with queen and rocks, perhaps in this
point TheKing can be the number 1.
Settings by Dr. Alexander Schumacher
Grailmaster.
TheKing 3.33 (Schumacher): opp=100 opn=100 opb=100 opr=100 opq=100 myp=100 myn=100
myb=100 myr=101 myq=101 cc=107 mob=120 ks=150 pp=102 pw=102 avd=0 cfd=-2 sop=100
rnd=0 sel=14
I will ask Johan de Koning for an interview later
05. ProDeo 1.1
by Ed
Schröder (The Netherlands)
ProDeo 1.1 is improved with 30-40 ELO to ProDeo
1.0 and the strongest available freeware engine. For long time matches it's more
intersting to set booklearning = off. With other words ProDeo's book learning is
too good for the others. Ktulu must learned it in the matches.
Thanks to Ed Schröder for such a strong freeware. ProDeo is a startegic
engine with a good positional playing style without tablebases support.
Unfortunately, no interview with Ed available. I asked Ed but he don't have time for it.
06. Ruffian
2.1.0
by
Per-Ola Valfridsson (Sweden)
Ruffian is around 50 ELO stronger as the free available versions 1.0x.
Unfortunately, I has since July 02nd, 2004 no contacts to Per-Ola
Valfridsson and after my information no other Gladiators have contacts to
him. I hope all is OK with him. Ruffian is stronger without ponder and need a fast
processor for higher rankings. If anyone have contacts to Per-Ola, please give
me a hint.
Interview with Perola
07. SOS 5.1 for
Arena
by
Rudolf Huber (Germany)
SOS 5.1 for Arena is improved in tactic.
Positional looks version SOS 4.0 for
Arena stronger. All in all SOS could be 20 ELO stronger as the preview version. One of
the TOP 5 from all available freeware engines. SOS need a lot of hashtables. You
can see that SOS have the highest rating on Xeon Dual 2.8 with 256Mb
hash.
More information can be found in
News-Ticker Message 83
An interview with programmer Rudolf Huber is in progress and available
middle of Mai 2005.
08.
SlowChess Blitz WV
by Jonathan
Kreuzer (USA)
The programmer Jonathan Kreuzer (USA)
improved SlowChess in the latest year a lot. Slow is now the strongest American
chess program, around 75 ELO stronger as Crafty. Slow's playing style is very
balanced. Just a nice development with more and more fans.
Interview with Jonathan
09.
Spike 0.9 / 0.9a
by Volker
Böhm and Ralf Schäfer (Germany)
Again a German amateur program in headline. I
believe my ATL-2 Spike rating isn't hight enough. 540 games are not enough for a
stable rating. If I look on my private Centrino 1.6 / Athlon 2800+ database
Spike have 25 ELO more. The programmers of Spike working on tablebase support
for version 1.0. With tablebases support Spike can be 20 ELO stronger. A very
strong and balanced engine. Perhaps with problems in endgame but with a very
good positional playing style.
Example with CEGT and ATL ratings:
After 500 CEGT games: Gandalf 6.01 to Shredder 9 = 75 ELO
After 300 ATL games: Gandalf 6.01 to Shredder 9 = 125 ELO
And now after more as 1.000 CEGT and 500 ATL games, both ratings are the same,
Gandalf 6.01 to Shredder 9 = 100 ELO
Interview with Volker Böhm and Ralf
Schäfer
10.
Pharaon 3.2
by Franck
ZIBI (France)
Pharaon 3.2 are improved too. I like the work
by Franck ZIBI since many years. Pharaon need a lot of time. 40 moves in
20 minutes is BLITZ chess for Pharaon. I am thinking under regluar tournament
rules the ranking will be higher. Pharaon have problems in tactic. The strategy is
the advantage.
Interview with Fanck ZIBI
(available on ExactaChess)
This was a really nice
tournament.
Thanks to all programmers for this nice engines and the other information you
gave me (interviews, mails ...).
April 25th, 2005
Ktulu 7: CEGT Heinz van Kempen gave his statement
114,
FQ
I wrote in the last weeks some messages about
the CEGT tournament / rating list by Heinz van Kempen, Christian Koch,
Chris Tatham and Charles Smith. This group of people produced on
7-8 systems with ~ 40 moves in 20 minutes the most interesting rating list at
these times.
Clear, a good combination from Intel / AMD
systems will give us the information we search day for day.
This rating list can be for objective people, working on such places like Arena,
the new STANDARD.
It's not easy for self artists which betake
oneself to flight. Computer chess fora build the basic for these people.
Editorial journalists of German chess magazines gave comments to Ktulu 7.0
without to produce one game in self work, without to look what the young
programmer of IRAN (Rahman Paidar) created. We can see it each year if a strong
NON ChessBase engine is available. One of the reasons Gladiator's / Chessbaslers
/ Winboarders lost more and more interest on computer chess fora. I am sure that
ChessBase don't like it because the ChessBase engine / programmers have
interest on a good fight against the group of young talents like Rahman
Paidar.
To hold a bad reputation is easy!
The most computer chess fora are a kindergarten for adults.
Sorry, but with this opinion I don't stand alone!
Now back to Ktulu 7.0 by
Rahman Paidar.
This is the topic, the new sensation ...
A young programmer of IRAN
sitting with the possibilities he has in his domicile without many of
information others have and possibilities others have. Rahman gave us an engine
with the playing level of long year's developments from good known TOP
programmers. What for a potential / capability Rahman Paidar must have?
Do you think about this one?
We have to
power the young talents!
Rahman Paidar
will give the German editorial journalists the possibility to repair the World
wide bad reputation.
It's very easy to find out that this program is a SENSATION!
The CEGT team don't have to work
on a bad reputation because all is convincing.
You can learn a lot about
ratings and statistics if you read the review by Heinz van Kempen.
Ktulu 7.0 have exact the
same rating as in ATL-2 and is at the moment on place 5-6 from all available
chess programs the World have.
Heinz van Kempen wrote me:
As there seems to be a lot
of interest in the new Ktulu we try to give a first careful report for CEGT:
http://www.husvankempen.de/nunn/ktulu.doc
Completed and interim
crosstables for CEGT 3 are also available, as well as the download of all games
so far:
http://www.chessfighters.de/cegt/
http://www.husvankempen.de/nunn/
April 22nd, 2005
"Der Mond von Lisa Thul"
113,
FQ
Auch der Mond könnte eine Rolle spielen warum
so viele Menschen in Internet-Foren ausrasten. Es ist schon schade wie diese
schöne Möglichkeit von Selbstdarstellern missbraucht wird. Viele Leser kennen
vielleicht die Auswirkungen vom Mond nicht und das ist natürlich ein Grund den
"Mond" mal etwas näher vorzustellen.
Text von Lisa Thul
(female Gladiator) :-)
Einfach mal zurücklehnen und Lisa's Worten lauschen ...
Zurücklehnen habe ich gesagt und nicht zornig
in den Bildschirm schauen !!
Die Welt besteht nicht nur aus interaktiven
Engine-Engine Matches und verkrampftes frönen und feiern von Resultaten,
Personen die sich Bemühen mit ziemlich dummen Kommentaren bloß zu stellen etc..
Dieses unterste Schubladen-Niveau ist der Grund für das Aussterben von Foren. Es
gibt viele Vereinigungen in unserer heutigen Zeit "Rettet die deutsche
Buttermilch". Warum nicht ... "Rettet die Schachcomputerforenkultur".
Nun aber zum Mond ...
Ein kleiner Überblick über
seine Auswirkungen
Der Mond begleitet uns Nacht
für Nacht seit Millionen von Jahren und damit ist er einer der wenigen
Konstanten, die man in unserer heutigen unruhigen Zeit noch finden kann. Diese
Kontinuität macht ihn zu einem besonders verlässlichen Begleiter, mit dem man
auf Grund seines bestehenden Rhytmus, im wahrsten Sinne des Wortes, immer
rechnen kann.
Immer öfter haben wir böse
Naturkatastrophen zu beklagen. Eine der schlimmsten überhaupt sind sicherlich
Überflutungen. Man steht ihnen völlig machtlos gegenüber. Wissenschaftliche
Beobachtungen haben ergeben, dass sich die Wassermassen immer erst bei
abnehmenden Mond auf den Rückzug begeben. Man kennt dieses Phänomen nur zu gut
in ähnlicher Form bei Ebbe und Flut. Nur der Mond, als Naturmacht vermag dieser
anderen Naturmacht, dem Wasser tatsächlich die Stirn zu bieten und damit auch
noch erfolgreich zu sein.
Dieses gilt im übertragenden
Sinn auch für uns Menschen.
Bekannter weise besteht unser Körper zu etwa 65% aus Wasser. Diese Tatsache
macht sich der Mond zu eigen, um unser Psyche und auch unserer Physis einen
Streich zu spielen. Die Intensität der Wirkung ist natürlich immer von der
jeweiligen Konstitution der einzelnen Personen abhängig. Bei abnehmenden Mond
zum Beispiel entzieht er uns Schwung und Elan, bei zunehmendem Mond sprühen wir
vor Unternehmungsgeist und neuen Ideen. Bei Vollmond sind manche von uns
hyperaktiv und leicht reizbar und bei Neumond beschert uns vielleicht einen
Neubeginn.
Dementsprechend können wir die
einzelnen Mondphasen auch für unsere Pflege einsetzen.
Bei zunehmendem Mond ist unser Körper auf Aufnahme programmiert. Das heißt, dass
alle Stoffe die wir unserer Haut zufügen jetzt ganz besonders gut aufgenommen
werden. Es empfiehlt sich hierzu Masken, Ampullen oder sonstige
Intensiv-Präperate in den Pflegeplan mit zu integrieren. Der Zeitpunkt ist
äußerst günstig, da er die Wirkung fast verdoppelt.
Bei abnehmendem Mond ist der
Körper auf Loslassen programmiert. Es fällt ihm in dieser Zeit wesentlich
leichter zum Beispiel Kopfhaut-Schuppen zu verdammen oder die überflüssigen
Pfunde purzeln zu lassen. Hier ist die Erfolgsquote für Entgiftungs- und
Entschlackungskuren am höchsten. Auch das Ausreinigen der Haut zum Beispiel mit
Spezialpeelings oder eine Akne-Behandlung ist in diesem Zeitraum angesagt und am
effektivsten. In diesem Zeitraum werden sie von Verletzungen kaum Narben zurück
behalten.
Vollmond und Neumond haben hier
natürlich einen Sonderstatus. Bei Vollmond sollten Sie alles in Angriff
nehmen von dem Sie sich für immer trennen möchten.
Sorry Lisa, muss Dich unterbrechen ...
Das bedeutet nicht, sein Passwort mit großen Tobowabobo offiziell
zurückzugeben um dann via Passwort-Sharing das Forum dennoch zu benutzen ...
klar?
Vielleicht ist das die
anstehende notwendige Entfernung von einer Hautpigmentierung oder eine
gewünschte Laserbehandlung zur Körperhaarentfernung. Vom Neumond wird alles
aktiviert oder auch wieder regeneriert. Hier kann man zum Beispiel stark
strapazierter Haut mit Massagen und entsprechenden Pflegekuren wieder einen Kick
geben. Die nachfolgende zunehmende Mondphase schließt dann nahtlos an die
begünstigte Wirkstoffaufnahme an.
Diese Struktur können Sie
genauso auf Haare, Nägel oder Ernährung übertragen. Die Grundsätze und auch die
14-tägigen Zeitabschnitte bleiben immer gleich.
Eigentlich sind diese alten.
traditionellen Weisheiten seit langem verschollen oder belächelt, doch in
jüngster Zeit erfahren sie wieder eine neue Renaissance. Was Früher galt, gilt
auch Heute noch und zum Jahreswechsel kann man in allen Bücherecken eine Auswahl
an Büchern oder Mondphasen-Kalendern finden, die detailliert und übersichtlich
die Auswirkungen der einzelnen Abschnitte auf Mensch und Natur beschreiben.
Ein sehr interessantes und
ergiebiges Thema, womit es lohnt sich zu beschäftigen. Es bietet Ihnen die
Chance den Kontakt zur Natur wieder aufzunehmen und sich wieder in das
natürliche System zu integrieren.
Quelle:
http://www.casalisa.com
Nutzen Sie
dieses Wissen
Viel Spaß dabei
Ihre Lisa
April 22nd, 2005
Patriot 2.0.0 /
Forum / ATL-2
112,
FQ
Note (next interviews):
1. English interview with Aice programmer Anastasios Milikas (Greece)
will be available in the next days.
Cooperation from Alex Schmidt and myself!
2. English interview with Thinker programmer Lance Perkins (Canada)
will be available in the next days (think so).
3. English interview with a commercial programmer (little secret) will be
available in the next days.
Cooperation from Alex Schmidt and myself!
4. German interview with SOS programmer Rudolf Huber (Germany) in
processing.
ATL-2 League will be finished with engine
number 10: Ktulu 7.0.
Possible that the tournament mode is interesting for the readers but it's boring
for myself.
I will see more engines and thinking about a better tournament mode.
ATL-3 will start as soon as possible.
Nice weather in Germany.
Time for stock up on some sunshine.
This athletic
computer chess player (Gladiator) is wearing shorts with the label of Arena, I
am sure!
Arena - We make it
possible!
Your spectacular and exciting computer chess place!
We have it, you know it, you get it - only here!
Have fun surfing on Arena page!
Our bug report form have a bug :-)
As scary as it gets!
Sometimes you will get an error message
if you are using the form.
With or without bug, you can be certain that I get all forms.
I don't find the mistake and I am thinking ... I can live with it.
With other words, we know if from chess engines ... these bug isn't really a bug
... it's a feature!
Means, in the name "Bugreport form" should be included a little bug ... you
understand!?
April 21st, 2005
Partner
engines updated, Spike 0.9a, Hermann 1.35, Aice 0.92
111,
FQ
New versions from our partner
engines are available!
1.
Spike 0.9a by Ralf Schäfer and Volker Böhm (Germany)
http://spike.lazypics.de/
Spike 09a is is nearly identical to Spike 09,
but has some bugfixes in its interface handling. Neither playing strength nor
playing style has been changed, its still the same Paderborn version. Because of
its change of the interface Spike 09 may be a little stronger in bullet games
with extremly short time control.
Bugfixes:
Spike now supports time settings with a fixed time for the first n moves and
additional time for the rest Spike now supports time settings with a fixed time
per move Spike´s pv length is now limited to 20 moves (40 half moves). Very long
pv strings (above 512 chars) caused a crash in winboard. Spike no longer claims
draw if one side has two knights. The problem that Spike evaluates a KNNKP as
draw is not fixed as the 09a is only an interface fix and shouldn´t play
stronger - even in this seldom cases. Spike now shows its calculations (pv, kps,
...) at once in fritz gui. Spike0.9 sometimes had waiting times of 1-5 seconds
in fritz gui before the first pv line was printed. Spike hopefully not hangs
longer after analysing a game using the Fritz-Gui.
2.
Hermann 1.35 by Volker Annuss (Germany)
http://www.nnuss.de/Hermann/index.html
Changed since the latest available
version on Arena site:
Version 1.3.5
- Zeiteinteilung geändert
- Hashtabelle vereinfacht
- Erweiterungen der Suche vereinfacht
Version 1.3.4 (privat)
- etwas Königssicherheit bei Dame im späten Mittel- und Endspiel
- Bewertung der Eröffnung geändert
Version 1.3.3 (privat)
- Erweiterungen der Suche geändert
- Stellungswiederholung nach Schach in der Ruhesuche erkennen
Version 1.3.2 (privat)
- Fehler in Ruhesuche und Mehrvariantenmodus korrigiert
- Mehrvariantenmodus liefert auch bei großen Differenzen exakte Werte
3.
Aice 0.92 by Anastasios Milikas (Greece)
http://www.milix.net/aice/
Aice is a new
partner engine. Aice is the seventh available Chess960 engine and can be found in
Arena's FRC Setup (Setup 3).
An interview with Aice progammer Anastasios Milikas will follow in the
next days.
Arena Setup 2
and Arena Setup 3 are changed!
April 21st, 2005
First
Lithuanian Arena translation
110,
FQ
Time translation- 37
days:
1. English to lithuanian- 8 days (not hard)
2. Correction Alt-keys- 5d. (hard)
3. Review all(!) kommands and dialogs for good-looking view,
more info, corrects lithuanian words - 22d. (very hard)
4. Dressing - 2d. (pleasure)
|
Romualdas Birmanas (Lithuania)
sent the first Lithuanian Arena translation.
Arena is now in 19 languages available, again a new
record!
The menue translations of Arena
are the most interesting news for us.
We like it that all the Gladiators in the World can use Arena in the own
language.
An other Gladiator (comes from Australia) is translating Arena so far in
Chinese.
This could be the next available language file.
Romualdas sent me a very friendly
mail.
Perhaps this information by Romualdas is interesting for the others:
1. My hobby (if >12 years):
Hobby number-1: absolute no narcotics (alcoholic, nicotine...).
Past years: swimming, canoe ( 21 travels: 3654km ), bicycle, skaters,
photo, running, skis, orchard, correspondence chess
Last year ( 2004) in winter: skis - 371km, skaters - 36km on lake.
Last year ( 2004) in summer: bicycle - 3464km, swimming - 25,3km in lake.
2. I am engineer-energetic.
My works was:
Installing large boiler-house - 3 years.
Designing hundreds factories - 23 years.
Building very large hydro power station (KHAE in Lithuania) - 21 years.
3. Programming
7 programmes for engineering- technical coutings.
Excel files-programmes for home live, financial, sports statistic and
etc- 19.488 strings VBA codes.
"Kalendorius 1900-2200" (in Lit)- Calendar with compute intervals times
for 8x2 conditions and with sun times for all points(!) of Lithuania
(original system equations). This programme 6 years was my training
place for VB-VBA.
4. I myself, one, only
with my hands, was build not great wooden house (began from timbers in
wood), but dismal end- now he is not my.
I was bringing up 2 children's (engineers) and growing 146 fruit-trees
for 3 orchards. |
Romualdas Birmanas
I am 73 years old.
Pensioner.
|
Text by
Romualdas Birmanas:
Hydro accumulation power station
KHAE in Lithuania.
When we were coming, there was nobody. When I, after 21 years, go out- stay
that.
I made technical control building this ferro-concrete pipes (D=8,3m, L=880m)
and all working in his place.
Work was difficult, responsible, but very interesting, among thousands fine
working peoples.
When I am looking in this photo, my workfriends comes back to me.
GREAT THANKS!
April 21st, 2005
Ktulu 7.0 SERVICE 6
817 Ktulu 7.0 games
Ktulu can be better but ...
Third CEGT / ATL rating list available
109,
FQ
Ktulu 7.0 by
Rahman Paidar (Iran) played now 464 games in ATL-2.
Enough games for a stable rating (+-25 ELO).
Download (games -pgn-, log files):
ATL-2 League
ATL-2 results:
Ktulu 7.0 - Shredder 9.0
26.5 : 33.5 (over)
Ktulu 7.0 - Gandalf 6.01
31.0 : 29.0 (over)
Ktulu 7.0 - TheKing 3.33 Schumacher 28.0 : 32.0 (over)
Ktulu 7.0 - Ruffian 2.1.0 31.5 : 28.5 (over)
Ktulu 7.0 - ProDeo 1.1 28.0 : 31.0 (one
double game, will be replay)
Ktulu 7.0 - SOS.5 for Arena
33.0 : 27.0 (over)
Ktulu 7.0 - SlowChess Blitz WV 35.0 : 23.0 (two
games to play, double games)
Ktulu 7.0 - Spike 0.9
30.5 : 15.5 (14 games to play)
Ktulu 7.0 - Pharaon 3.2
1.0 : 0.0 (59 games to play)
---
ATL-2 rating list after 2.396 games
40 moves in 20 minutes
Program Elo + - Games Score Av.Op. Draws
1 Shredder 9.0 : 2750 26 26 540 71.4 % 2591 29.8 %
2 Gandalf 6.01 : 2653 25 24 540 57.3 % 2601 31.3 %
3 Ktulu 7.0 : 2639 26 26 464 52.9 % 2619 31.2 %
4 TheKing 3.33 Schumacher : 2630 24 24 540 53.8 % 2604 30.9 %
5 ProDeo 1.1 : 2623 24 24 539 52.6 % 2605 31.4 %
6 Ruffian 2.1.0 : 2619 24 24 540 52.0 % 2605 31.5 %
7 SOS 5.1 for Arena : 2566 25 25 540 43.5 % 2611 30.4 %
8 SlowChess Blitz WV : 2564 25 25 538 43.3 % 2611 28.6 %
9 Spike 0.9 : 2528 25 25 526 37.7 % 2615 31.4 %
10 Pharaon 3.2 : 2498 27 27 481 33.7 % 2616 28.3 %
---After 031 ATL-2 games = 2.693
ELO
After 056 ATL-2 games = 2.689 ELO - 4
After 105 ATL-2 games = 2.684 ELO - 5
After 178 ATL-2 games = 2.674 ELO -10
After 236 ATL-2 games = 2.653 ELO -21
After 309 ATL-2 games = 2.646 ELO - 7
After 464 ATL-2 games = 2.639 ELO - 7
My prognosis was good!
I wrote in News-Ticker Message 100 that I believe that the engine lost 50 ELO
with the next matches!
Ktulu lost exact 50 points after the first 56 up to 464 games :-)
I collect in the latest days more Ktulu 7.0
games for the programmer Rahman Padiar (IRAN)
Dieter Eberle, Johan Havegheer, Michael Jesdinsky,
Christian Koch, Charles Smith, Heinz van Kempen, Jan
Kiwitter and Harry Schnapp sent me games.
THANKS!
If you have interest to replay the games you
can see that Ktulu need a book learning option. In one of the ATL-2 matches
Ktulu lost against ProDeo 1.1 5 of the 20 games in the same opening variation.
One game is double and I have to replay it. Dieter Eberle wrote the same
today in a mail to me. Ktulu have on Dieter's machine exact the same problem. I
think with a good book learning feature, Ktulu can be better around 20 ELO. I
will contact Rahmen in the evening. At the moment Rahman is working on an
update.
The third CEGT/ATL rating list is available!
After 464 Ktulu 7.0 games a fantastic sixth place.
5 points to the number 5 Hiarcs 9.0 and 7 points to the number 4
Gandalf 6.0.
CEGT/ATL ratings (Ktulu
seems to be improved with 80 ELO) !!
Order Ktulu 7.0
Gladiator-Shop
We will inform the customers
automtically about updates!
April 17th, 2005
Ktulu 7.0 SERVICE 5
Ktulu 7.0 with a stable rating
108,
FQ
Ktulu 7.0 by
Rahman Paidar (Iran) played now 309 games in ATL-2.
Enough games for a stable rating (+-30 ELO).
Download (games -pgn-, log files):
ATL-2 League
ATL-2 results:
Ktulu 7.0 - Shredder 9.0
26.5 : 33.5 (over)
Ktulu 7.0 - Gandalf 6.01
31.0 : 29.0 (over)
Ktulu 7.0 - TheKing 3.33 Schumacher 28.0 : 32.0 (over)
Ktulu 7.0 - Ruffian 2.1.0 31.5 : 28.5 (over)
Ktulu 7.0 - ProDeo 1.1 28.5 : 31.5 (over)
---
ATL-2 rating list after 2.396 games
40 moves in 20 minutes
Program Elo + - Games Score Av.Op. Draws
1 Shredder 9.0 : 2750 26 26 540 71.4 % 2591 29.8 %
2 Gandalf 6.01 : 2653 25 24 540 57.3 % 2602 31.3 %
3 Ktulu 7.0 : 2646 31 32 309 49.2 % 2652 34.3 %
4 TheKing 3.33 Schumacher : 2631 24 24 540 53.8 % 2604 30.9 %
5 ProDeo 1.1 : 2624 24 24 537 52.7 % 2605 31.3 %
6 Ruffian 2.1.0 : 2620 24 24 540 52.0 % 2606 31.5 %
7 SlowChess Blitz WV : 2564 26 26 480 43.8 % 2608 29.2 %
8 SOS 5.1 for Arena : 2561 26 26 492 43.2 % 2609 31.1 %
9 Spike 0.9 : 2528 26 26 480 38.1 % 2612 31.7 %
10 Pharaon 3.2 : 2499 27 27 480 33.8 % 2616 28.3 %
---After 031 ATL-2 games = 2.693
ELO
After 056 ATL-2 games = 2.689 ELO - 4
After 105 ATL-2 games = 2.684 ELO - 5
After 178 ATL-2 games = 2.674 ELO -10
After 236 ATL-2 games = 2.653 ELO -21
After 309 ATL-2 games = 2.646 ELO - 7
The ATL-2 ELO history is very interesting. The
new TOP engine from IRAN have problems against weeker engines and have top level
results vs. stronger opponents. The rating of Ktulu 7.0 is very stable now. The
final results after 540 games will be 2.640 ATL-2 ELO. I am sure that CEGT
ratings will be higher. Ktulu likes stronger opponents and made here
a lot of points (speaking from Hiarcs, Fritz and Junior).
At the moment I collect Ktulu 7.0 games for
Rahman Paidar.
Rahman is working on an update for his customers.
I asked Harry Schnapp for playing more
games vs. the "once" known feared oppoment Chess Tiger 15.0. Harry played
again 20 games and Ktulu 7.0 lost after the first match (1.5 : 8.5) the second
with 2.5 : 7.5 (Ktulu original book) and the third with 3.0 : 7.0 (book by
Denis Grafen). Against the other programs from the ChessBase product family
Ktulu made 45% (Shredder 9.0) - 60% (Hiarcs 9.0). Really a sensation and
my congratulation to Rahman Paidar. Ktulu is improved with ~ 90 ELO and
with Gandalf 6.0 and Hiarcs 9.0 on the same level (+- 20 ELO). This should be
the number 4-6 in the World.
Christian Koch, Charles Smith,
Chris Tatham and
CEGT director Heinz van Kempen started the first CEGT Ktulu 7.0 games.
The first results can be found in perhaps two weeks ... THANKS for testing !!
If customers have interest
to help a little bit ... please send me your Ktulu 7.0 games.
I have interest to public your results.
We can Rahman helps with such PGN files. I am
playing with Arena's blunder check. This feature produced an EPD file
with all played Ktulu blunders. Such an options helps a lot to improved an
engine. Do you know this option?
Arena Menue:
Engines / Tournament / Blunder-Check / Enable Engine Blunder Check
My setting:
If value decreases 100 centipawns
within 2 moves
No blunder if already 400 centipawns down
Included are games from Harry Schnapp,
Christian Koch, Heinz van Kempen, Charles Smith, Michael
Jesdinsky, Jan Kiwitter and my 309 ATL-2 games. Please replaying the games, you will have a lot of fun with the playing style of Ktulu
7.0 ...
Order Ktulu 7.0
Gladiator-Shop
We will inform the customers
automtically about updates!
Go to
News-Ticker: Main Page,
Page 1,
Page 2,
Page 3,
Page 4,
Page 5,
Page 6,
Page 7, Page 8,
Page 9 |