Reply
 
LinkBack Thread Tools Display Modes
  #1   Report Post  
Old February 17th 04, 11:02 PM
The Darkness
 
Posts: n/a
Default Shredder 8 question

I have the main CB 5 (Hiarcs, CT15, Shredder 7, Fritz, Junior) and have just
ordered Shredder 8 after reading Steve's column at ChessBase.com.... just
wondering if anyone has gotten Shredder 8 yet and what are your impressions
so far, especially in comparison to Shredder 7 and the other four.

Steve stated Shredder 8 just LOVES closed positions, which is disturbing
since computers used to always destroy their opponents with tactics--- seems
insane for a computer opponent to love a closed position!


  #2   Report Post  
Old February 18th 04, 12:21 AM
Euclid
 
Posts: n/a
Default Shredder 8 question

Shredder 8 didn't install correctly for me, the first time.

Based on my experience, to get a good install of Shredder 8, I suggest that
your first uninstall your prior Shredder 7 GUI (or whatever you're using
now). I also went to the trouble to clean up the chssbase.ini file to remove
references to prior GUI installations. After taking those steps, it
installed OK for me.

It's different from Shredder 7/7.04. It goes to deeper ply levels several
times faster using its new "intelligent" playing style. But don't let that
fool you. If you want to compare the analysis with your prior analysis with
Shredder 7/7.04, use the total number of kNodes as the basis (not ply level
achieved).

Does it love closed positions? I don't think so. I've run across some closed
positions where it hasn't the slightest idea what to do, and they weren't
"locked pawn" situations either. It just didn't know what to do. Despite
having a significant advantage it just wiggled its rook(s) back & forth to
make a draw, rather than pushing a pawn to open the position. It saw the
pawn push on the prior move, but couldn't see it on the current move, no
doubt a bug in its algorithm. This is in late middlegame/early endgame
situations. It's just not very smart in that case. This could be a bug
related to tablebases, I'm not sure. It could also be a bug related to
handling of the Rooks in complicated positions, which I've noticed for a
long time with Shredder 7/7.04 too.

Overall, it is very intelligent with its searching. I've never encountered
anything as good.

Its analysis is undependable, worthless in many situations, playing
"give-away" chess instead!? You can't trust anything in the analysis beyond
the 1st move. Shredder 7/7.04 had the same problem, not fixed in version 8.
The author admits to this situation, which he calls a "feature"?!

Overall, it's strong, perhaps the strongest. I'm still waiting for the SSDF
evaluation.
--
Euclid


"The Darkness" wrote in message
. ..
I have the main CB 5 (Hiarcs, CT15, Shredder 7, Fritz, Junior) and have just
ordered Shredder 8 after reading Steve's column at ChessBase.com.... just
wondering if anyone has gotten Shredder 8 yet and what are your impressions
so far, especially in comparison to Shredder 7 and the other four.

Steve stated Shredder 8 just LOVES closed positions, which is disturbing
since computers used to always destroy their opponents with tactics--- seems
insane for a computer opponent to love a closed position!



  #3   Report Post  
Old February 18th 04, 11:55 AM
David Richerby
 
Posts: n/a
Default Shredder 8 question

Euclid wrote:
It's different from Shredder 7/7.04. It goes to deeper ply levels
several times faster using its new "intelligent" playing style. But
don't let that fool you. If you want to compare the analysis with your
prior analysis with Shredder 7/7.04, use the total number of kNodes as
the basis (not ply level achieved).


Neither total number of nodes considered nor ply depth reached is a
reliable method of comparison between different engines, as is often
pointed out here.


Dave.

--
David Richerby Salted Tool (TM): it's like a handy
www.chiark.greenend.org.uk/~davidr/ household tool but it's covered
in salt!
  #4   Report Post  
Old February 18th 04, 01:39 PM
Euclid
 
Posts: n/a
Default Shredder 8 question


"David Richerby" wrote in message
...
Euclid wrote:
It's different from Shredder 7/7.04. It goes to deeper ply levels
several times faster using its new "intelligent" playing style. But
don't let that fool you. If you want to compare the analysis with your
prior analysis with Shredder 7/7.04, use the total number of kNodes as
the basis (not ply level achieved).


Neither total number of nodes considered nor ply depth reached is a
reliable method of comparison between different engines, as is often
pointed out here.

Dave.
_______________________________

People want a fast & easy way to compare engines, and either
kiloNodes/second or total number of nodes are the most commonly used
methods. What better method is there? The average person doesn't want to
spend a lifetime running tests, such as solving chess problems, or engine vs
engine matches. It takes thousands of games to get reliable results with
that method, which is only possible for large groups such as SSDF. What else
is there that the average person can use for engine comparisons? Surely not
Fritzmarks... )
-E


  #5   Report Post  
Old February 18th 04, 11:54 PM
Derek Wildstar
 
Posts: n/a
Default Shredder 8 question


"Euclid" wrote in message newsWJYb.8976

People want a fast & easy way to compare engines, and either
kiloNodes/second or total number of nodes are the most commonly used
methods. What better method is there?


The better method, in fact the only method to evaluate engine performance,
is to see the engine play against itself, a human, or other engine.

The mere fact that people find it convienient to measure an engine
performance by how many nodes a sec it (may or may not) evaluates, or the
indicated depth-of-search (selectively, or exhaustively) does not make it a
meaningful indicator of anything.

The only time those individual specifications mean anything is when specific
changes are made to a specific engine which directly affect the evaluation
function (increasing or decreasing nodes evaluated by the primary evaluation
routine, some engines have more than one) and/or search extensions which may
increase depth of ply for selective seraching. Even then, the technical
output is secondary to how the engine plays.

The average person doesn't want to
spend a lifetime running tests, such as solving chess problems, or engine

vs
engine matches. It takes thousands of games to get reliable results with
that method, which is only possible for large groups such as SSDF. What

else
is there that the average person can use for engine comparisons? Surely

not
Fritzmarks... )


I personally have run engine-engine matches in the thousands of games,
results and move times and evaluations stored, and I am not a 'large group'
like the SSDF. All you need is the programs and the hardware, and any
'average person' who has the motivation can duplicate the same results. In
fact, with my superior hardware to that posted by the SSDF, I can argue that
my results are more meaningful than theirs! At the very least, I can rank
engines against each other, if not give them a wholly accurate Rating.

In none of these engine matches is either depth of ply, nor kN/S a
consideration. You really need to wrap your arms around this concept...and I
know how stubborn you can be. But I also know your enthusiasm, so here it
goes: nodes per second is not an indicator of engine strength, nor of
improvement within an engine, mere depth of search is not an indicator of
engine strength. the only meaningful indicator of engine strength is how it
performs against known rated players (including engines).







  #6   Report Post  
Old February 19th 04, 05:59 AM
Euclid
 
Posts: n/a
Default Shredder 8 question

My original post was in response to a question about how Shredder 8 compared
to Shredder 7. I have analyzed many games and positions with Shredder
7/7.04, so have a basis for such comparisons.

When I started using Shredder 8, I immediately noticed that it reached a
deeper ply level quicker, much quicker. It goes to about 2 ply levels deeper
in the same amount of time as Shredder 7. So the first impression is that
it's calculating better because of the deeper ply levels achieved, but
that's deceptive. My comment was then that it is more reliable to use the
total kilonodes calculated when comparing Shredder 8 vs Shredder 7, rather
than the ply levels. I believe that is generally true, based on my
subsequent use of Shredder 8. Now, I only use the ply level as a bookmark to
compare different runs of analysis of a particular position, and look mainly
at the total nodes calculated. That works pretty well. Thus if I calculated
a position to 5 billion nodes with Shredder 7 in infinite analysis mode, I
get approximately the same result if I calculate it to 5 billion nodes with
Shredder 8. So that's the best way to compare their respective analyses.

I don't use these engines to play chess in the usual way. I only use them to
calculate for very long times in infinite analysis mode. Typically I run
each position for at least 1 hour, sometimes 8 hours or longer, searching
for the absolute best move. Human vs engine or engine vs engine matches run
at ordinary time controls are of little interest to me. They're like
shooting craps, with a large element of luck involved in which side wins the
game.

Ply levels and total kilonodes calculated are excellent indicators of what
an engine is doing, and are reasonable ways to compare different engines.
They're as reliable as anything else, once you understand what they're
saying. David Richerby's and your comments to the contrary are nonsense.
--
Euclid


  #7   Report Post  
Old February 20th 04, 12:27 AM
Derek Wildstar
 
Posts: n/a
Default Shredder 8 question


"Euclid" wrote in message
ink.net...

Ply levels and total kilonodes calculated are excellent indicators of what
an engine is doing, and are reasonable ways to compare different engines.
They're as reliable as anything else, once you understand what they're
saying. David Richerby's and your comments to the contrary are nonsense.
--
Euclid


I have come to the conclusion you have little understanding of how a
chess-playing program works...you might watch the numbers increase and see
the move order change in the output, but you do not understand why.

I have tried to tell you why, so have others, but you refuse to consider
that you are incorrect, or your understanding, flawed. There is little I can
do to change your narrow-minded obstinance, except provide you with a link
to the thesis for the chess playing program, Rookie 2.0.

http://brick.bitpit.net/~blik/

It is an excellent paper, not quite written for the layman, but perfectly
understandable given a modicum of effort. If you find it hard to follow at
first, please try "How Stuff Works: Chess Computers"

http://entertainment.howstuffworks.com/chess.htm

Knowledge is Power Euclid.








  #8   Report Post  
Old February 20th 04, 05:25 AM
Euclid
 
Posts: n/a
Default Shredder 8 question

Your posts are meaningless to me. I have no interest in your advice or
opinions. Logic is my first and last love, not your nonsensical verbiage.
Please be advised that I don't intend to respond to you in the future.
--
Euclid


  #9   Report Post  
Old February 20th 04, 06:05 AM
Derek Wildstar
 
Posts: n/a
Default Shredder 8 question


"Euclid" wrote in message
news
Your posts are meaningless to me. I have no interest in your advice or
opinions. Logic is my first and last love, not your nonsensical verbiage.
Please be advised that I don't intend to respond to you in the future.
--
Euclid


If you do not respect my dissemination, surely you must be deafened by the
silence of the accomplished, world-renowned programmers supporting your
assessments. Alas, you refuse to consider anything but your position, how is
that logical?







  #10   Report Post  
Old February 20th 04, 10:48 AM
David Richerby
 
Posts: n/a
Default Shredder 8 question

Euclid wrote:
Your posts are meaningless to me. I have no interest in your advice
or opinions. Logic is my first and last love


If your love is for logic, try this. Write a chess program that does
a little bit of move ordering (captures first) and whose evaluation
function looks just at material, with whatever point scores for the
pieces you like. It will play truly awful chess but will do a fanta-
stic number of nodes per second and should get a pretty decent ply
depth, too. It is trivial to analyze faster than any commercial
engine on the same hardware; the difficulty is in analyzing *better*.


Dave.

--
David Richerby Solar-Powered Hi-Fi (TM): it's like
www.chiark.greenend.org.uk/~davidr/ a music system but it doesn't work in
the dark!
Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Shredder 8-Junior 8, at game 30, Shredder squares up, 50% Live games! Anson rec.games.chess.analysis (Chess Analysis) 0 March 11th 04 11:17 PM
Shredder 8 is much faster Euc1id rec.games.chess.computer (Computer Chess) 23 January 27th 04 04:27 PM
deep fritz 8 vs shredder 704 Blackbeard's Ghost rec.games.chess.computer (Computer Chess) 30 January 24th 04 07:25 PM
Shredder 8 Question MiltonZuc rec.games.chess.computer (Computer Chess) 2 January 22nd 04 02:01 PM
problem installing Shredder 7.04 UCI engine (XP) Euc1id rec.games.chess.computer (Computer Chess) 1 August 12th 03 07:36 PM


All times are GMT +1. The time now is 12:54 PM.

Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
Copyright 2004-2019 ChessBanter.
The comments are property of their posters.
 

About Us

"It's about Chess"

 

Copyright © 2017