Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: CMJ12S100 V CM6666 G/60 Game 5 score 1.5 - 3.5

Author: Michael Cummings

Date: 04:28:11 01/08/00

Go up one level in this thread


On January 08, 2000 at 05:57:55, Chessfun wrote:


>I neither work for Mindscape or Michael Cummings. I do not test this for
>Mindscape or Michael Cummings I test it for my own interests and others
>if they are interested in seeing the games.

Since this is the case for your testing, no point in talking about it any
further. I find using the CM6666 personality of little use in obtaining any
results which would lead me to think that it is a good setting or not. But that
is my point of view

>Before I joined CCC there was no CM6K only CM5500.

If this is the case, you were either using another name or were not looking or
posting anything in this forum, or were not interested in anything regarding
chessmaster. There was months of Chessmaster post regarding settings and other
things, to the point which other members were getting fed up. I was a major
poster at that time, and do not remember you name even here at that time.

>I have been running CM personality tests for just over a year now and in
>that time have determined to my own satisfaction that the CM6666 settings are
>better than the default at 40/2hr time controls. I do not have the thousands of
>games you do, as I see sometimes you also play very fast time controls.

When did you see me playing faster time controls, I have not posted anything on
CM6K or a chessmaster games for a very long time, (nearly over a year now)

I have had CM6K since towards the end of 1998, like I said, you either changed
your name or never posted, I know most names I see post on here.

>I usually play is 30 min side and then only to determine if a personality is
>worth taking to the next step (40/2).

This testing method has certain flaws in my view.

>THE CM TEAM* tested at 5/3 against the default, which is of little to no value.

Which adds to my point about testing, and the way to do it. Since you clearly
state you are doing it for your own interests, then that is fine, as for the
interest of the results obtained in my view, then there is not much interest for
me, but that again means very little.

>Only reason I even bothered with this was John asked the question, personally I
>am certain these settings will be of little value and I do not try every set of
>settings I see. John made no statement about testing it
>only against the default.

>Your test itself was conducted without the use of the Josh book and though you
>can state they used the same opening this in itself means little unless you are
>going to play a few hundred games, otherwise we would all only be using Nunn or
>some similar test in the openings. One of the features of the Josh personality
>was it's opening book as well as the other changes.

This is a very important part of the testing. This gives the best indication of
the settings for the personality. After the opening which is usually taken from
the book anyway, this is where the testing of the engine with its settings comes
into its own.

CM6K opening books have never been great, and are of little importance when
testing how strong a personlity is, yes games can be won and lost in its
opening. But in testing CM settings, the opening book is of little importance
when finding what is the best.

Simply put, using the Chessmaster personality and the Josh settings with best
same opening book for both, will give better results as to which settings are
the stronger.


>
>And it is can be said that CM6666 might not perform well at longer time
>controls however I have not read much of that, and if that were said what
>difference if that is not how my computer scores it.

Its has been said, and my tests have gone to strengthen that statement.

>
>I have seen you make claims the last few days about Chess Tiger's strength
>versus the CM6K on games played at http://www.konts.lv/usr/Didzis/index.html
>games that were played at (16Mb hash when possible, 60min/game) this seems
>a similar claim to the one on the CM6666 not performing as well as the CM6K.
>What games or claims is it you refer to exactly?. Quote form Shep's
>"CM 6666  Shep  40/120 on 550 MHz and above; extraordinarily strong".
>Thanks.

Shep can quote what he likes, does not mean that it is correct or right. I have
2 CM settings which are currently ahead of all other settings that have been
floating around. But since you must have been here in CCC when I and the rest
posted settings of various personalities and how they ranked playing each other,
you can see clearly that this means very little. And that my 2 settings could be
strong or weak.

I take program strength with CM as using 16meg hash tables (Johann said to use
that, with selective search on 10) Playing games at 40/2hr time controls. If
CM6K or any other chessmaster program is going to play another program, then two
computers have to be used, CM is a CPU hog, and changing anything else like deep
thinking affect this program in my view.

But since you have now stated that you are doing this for fun, which I thought
you were doing some serious testing to add to tests done over time to give an
indication as to what is better or weaker, then this whole discussion is pretty
pointless.

you claim CM6666 in your tests are stronger than the default, then that is fine.
I have done my tests using a PII 450 with 256meg ram, 16meg hash tables, and
also on my current PIII 550 with 256 ram and 16meg hash tables (which is enough
power to give some good testing) to tell me otherwise.

IMHO, your tests are just as you said, for interest. It is not the way I would
test. You are testing an unknown setting with another unknown settinbg, at least
the default is the standard we are given and is a known setting and benchmark.

Well anyway, this josh setting has already bored me, I claim it as another
hopefull to bite the dust.

I have just finished testing it over 20 games at 5mins a game against other
settings, and it is not ahead of the default. So this setting bites the dust.



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.