Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Ruffian versions - I am really confused now

Author: Peter Skinner

Date: 10:05:51 02/27/04

Go up one level in this thread


On February 27, 2004 at 12:27:07, Peter Skinner wrote:

>On February 27, 2004 at 04:29:02, Robert Allgeuer wrote:
>
>>On February 27, 2004 at 01:31:26, Peter Skinner wrote:
>>
>>>On February 27, 2004 at 01:22:06, Jouni Uski wrote:
>>>
>>>>After more testing: may be 1.0.5 is the strongest version????
>>>>
>>>>Jouni
>>>
>>>Personally testing 1.05 and 2.1.0 there is almost no difference in strength.
>>>
>>>They almost score identically vs most other programs. 2.1.0 does do slightly
>>>better vs Shredder 8, and Hiarcs 8. Both versions seem to have problems with
>>>Hiarcs 9, and Junior 8.
>>>
>>>What is not surprising is that all Ruffian versions seem to have serious
>>>problems against Chess Tiger 15.0 - Gambit Tiger personality.
>>>
>>>Ruffian likes to dictate play, not have it dictated to..
>>>
>>>Peter.
>>
>>What is the time control you are testing with and get these results?
>>
>>Thanks
>>Robert
>
>The main time controls that I test at are:
>
>5 5 blitz
>30 30 (blitz or standard.. however you see the time control)
>90 10
>120 0
>
>Peter.

I suppose I should explain my reasoning for what I have posted.

Ruffian (all versions) seems to use an excessive amount of the time control when
the score drops due to a missed tactical shot. I have seen Ruffian use as much
as 45 minutes for one move when the score drops by more than 1 pawn (-1.00).
Then for the rest of the game, it is short on time, and just plain loses due to
not being able to search as deep as the opponent during the remaining time
control.

Ruffian is a great attacker and likes to dictate the play. Whenever an opponent
starts a very strong attack Ruffian tries to counter-attack when defense is the
best course of action.

My results can also be looked upon as YMMV (Your Milage May Vary), as I have run
tests between Crafty 19.10 and Ruffian 2.0.0 and Crafty won _easily_. Others
have posted similiar results, and others have posted quite the opposite. I
personally like to have enough data to support my claims and thus do not post
every little 20 game match as really 20 games is not enough of a measurement to
justify statements like "Tiger is stronger than Ruffian!". I prefer to gather
very large amounts of games vs many opponents (like the SSDF does) and look at
the evidence from that.

Here are some of my "findings", and you can take them with a grain of salt if
you wish:

1. The Crafty that I compile, and the books I created (offered for download at
http://crafty.vze.com) does incredibly well vs all programs I have. It certainly
is better than _any_ ruffian release to date. I don't know the reason for this,
but I attribute it to a great book.

2. Gambit Tiger is still a fantastic program. Ever since I was a beta tester for
the Rebel 11 project I have used the same settings and they never fail. I use
the following:

Hash = 96 megs
Anti-Human = On (Yes even against computers)

I know Christophe has said the hash size should be "the bigger the better", but
I have found the optimal hash setting for Tiger is 96megs. This could only be my
systems, but with talking with several operators with Tiger online they have all
found the same thing.

The anti-human setting is important as it will break up blocked positions and
will gain nps from it thus resulting in faster deeper searches. This is where
Tiger excells.

3. For the most part Ruffian 2.1.0 does well enough against Shredder to say it
is an improvement over 2.0.0 VS SHREDDER. Overall my results are very consistant
with the new SSDF rating list in that the 2.x series are not really stronger
than the 1.x free series release. There are others that believe this as well, so
I am not that far off. There was some good? or dishonest? marketting ploys done
here in the CCC to make people believe the program 2.x was stronger than 1.x. It
is coming to show that indeed this is not the case, thus I will never trust
results from certain individuals no matter what evidence they have. I believe
their only goal was to make a quick buck.

4. The best two program (atleast on my computers) are Shredder 8, and Junior 8.
I prefer longer time controls, and both are incredibly tough at timers greater
than >90 mins per side. This is of course verified by the SSDF list.

So as you can see you can take my results or opinions and toss them to the side,
but I do have evidence to support it, and others certainly agree with my
findings.

Peter.



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.