Information about Shredder
Stefan Meyer-Kahlen
As many people very asking me about more details about Shredder,
I will try to give some information about it.
Shredder was written by myself, Stefan Meyer-Kahlen, and had its
first appearence in the last year WMCCC in Paderborn. It was brand new
at this point as I started with the frist line of code 3 months
before the tournament. It played a very good tournament and scored 6/11.
It could have been more, but Shredder lost some important games in the end of
the tournament. Nevertheless I was very content with its result.
I made several improvements during the last year, but I certainly didn't
expect to win the title in Jakarta. Shredder played some very nice games
this tournament and in some games I was lucky, but I think you can't score 9/11
without any luck in such a strong tournament. In my opinion Shredder and Ferret
played the strongest chess in Jakarta, and it was a very close fight at the top.
Both programs would have deserved the title, but certainly I am happy that it was
Shredder, although a win for Ferret was ok to me, because Bruce and I spent a great
time in Jakarta and Bruce is a very nice guy.
There is nothing magic in Shredder. It has a rather complex evaluation function, it's
value can exceed the value of a pawn very easyly. This makes Shredder play some odd
lines sometimes, but somestimes it could play very strong. Shredder ran on a
PentiumPro 200 with 64megs of RAM in Jakarta. I used 2^20 entries per side in the hashtable.
It searched about 40-60 knps in the middlegame and up to 90 knps in
the endgame. Shredder is not very good in blitz chess. It scores pretty low against
Ferret or Fritz with 5min a game. It is written in ANSI-C and there are versions
for Solaris, LINUX, DOS and Windows etc. It ran on Windows 95 in Jakarta.
I studied computer science at the University of Passau in Germany and right now
I am writing my PhD thesis in Dortmund, Germany. I am 28 years old and play 'human'
chess at a level of about 2000 ELO points.
Maybe Shredder will be available soon, I will let you know on the internet, although
some people will blame me for that because of advertisment, but I got many questions
regarding this.
I hope this article was useful for somebody. If there are more questions I will try to answer
them here in future.
Stefan
Meik Weber
On Tue, 22 Oct 1996 11:33:33 +0200, Stefan Meyer-Kahlen
<mey...@fmi.uni-passau.de> wrote:
Congratulations.
> It ran on Windows 95 in Jakarta.
Check out the new c't 11/96. They compared some very new PCs (PPro)
and found that WinNT performs much better than Win95 even on normal
Pentum processors. They think that Win95 messes up the Cache during
process scheduling. You might gain something from running your program
on NT.
Grüße aus Bonn
Meikel Weber
m.w...@public.ndh.com
Make sure you visit my homepage
http://ourworld.compuserve.com/Homepages/meikel/mew.htm
brucemo
Meik Weber wrote:
>
> On Tue, 22 Oct 1996 11:33:33 +0200, Stefan Meyer-Kahlen
> <mey...@fmi.uni-passau.de> wrote:
>
> Congratulations.
>
> > It ran on Windows 95 in Jakarta.
>
> Check out the new c't 11/96. They compared some very new PCs (PPro)
> and found that WinNT performs much better than Win95 even on normal
> Pentum processors. They think that Win95 messes up the Cache during
> process scheduling. You might gain something from running your program
> on NT.
I suspect this is correct. My program seems to run faster on NT than it
does on Windows '95.
bruce
Vincent Diepeveen
I have the same experience.
Problem of NT and 95 is however that your hashtables decrease considerably
If you know that only already the evaluation hashtable
(not counting the transpositiontable, bishoptable, pawntable, kingsafety
table etc) speeds up the
program 50%, then you know enough... ...DOS the more memory the better!
Vincent Diepeveen
>
>bruce
--
+--------------------------------------+
|| email : vdie...@cs.ruu.nl ||
|| Vincent Diepeveen ||
+======================================+