Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Glaurung Does Better!

Author: Tord Romstad

Date: 04:09:34 06/09/05

Go up one level in this thread


On June 08, 2005 at 15:42:55, Vladimir Xern wrote:

>On June 08, 2005 at 11:21:30, Steve Maughan wrote:
>
>>Vladimir,
>>
>>>You should try Ruffian 1.0.5 in this position.  It finds the mate after about
>>>5 and a half minutes, second to The King's 2:30
>>
>>Glaurung does better, finding Bh7 in 3 secs, announces mate in 17 after 1:44 and
>>resolves to mate in 14 after 2:15 - (running on a 2 GHz Centrino)

Which is no big surprise.  Glaurung and Gothmog love bishop sacs on h7, and
often play them even when they are unsound.

>That surely is an impressive laptop.  My 1.4 GHz Athlon cannot compare. Here's
>Glaurung on my machine, albeit with 64 MB hash, though I doubt that matters too
>much.

In fact, it is very likely that it does.  When thinking for several minutes,
big hash tables help a lot for Glaurung.  Here are the times needed to find
mate in 15 and 14 from this position on my iMac G5 1.8 GHz with various hash
table sizes:

-----------+---------------+---------------+---------------+---------------+
Hash       |     32 MB     |     64 MB     |    128 MB     |    256 MB     |
-----------+---------------+---------------+---------------+---------------+
Mate in 15 | 8:20,15 plies | 6:01,15 plies | 3:05,15 plies |       *       |
Mate in 14 |10:18,16 plies | 8:17,16 plies | 4:02,16 plies | 2:15,15 plies |
-----------+---------------+---------------+---------------+---------------+

As you can see, from 64 MB to 256 MB each doubling of hash table size reduces
the time by a factor of about 2.

Tord



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.