Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Ruffian 2.0.0 vs. Shredder 7.04 +0 =0 -4

Author: Russell Reagan

Date: 14:54:55 12/22/03

Go up one level in this thread


On December 22, 2003 at 03:12:03, Christian Koch wrote:

>- hashtables = 256 MB for both engines

How much total physical RAM on the system?


>- GUI = Hiarcs9

Check and make sure that each engine is actually getting its full hash table
allocation. Gian Carlo Pascutto reported here that UCI engines would sometimes
only use a 1MB hash table in the Fritz GUI even when a larger size was
specified.


>- book = REMIS.CTG for both engines

I agree with others that each engine using its own book would be much better.
Otherwise your results really don't show anything meaningful to potential users
of Ruffian. Maybe the remis.ctg book and Ruffian don't go well together.

I recall Omid talking about how his program scores better than 80% against
Junior when Junior isn't using its tournament book (I think he said he used the
Fritz book instead), but when Junior uses its tournament book (and thus gets
into positions that are better suited for Junior), Junior wins the majority of
the time.

In any case, your games don't tell us much about what we might expect from
Ruffian 2.0 if we purchased it. If one of the engines cannot use its own book,
that is one thing, and the lesser of two evils would be to give use another book
for the engine that can't use its own. But if they can use their own books, why
isn't that being done? That seems to be the most logical and straightforward way
to test the engines. I think most users are interested in how the engines
compare to one another out of the box.



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.