Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Tablebase cache size: who has made experiments?

Author: Robert Hyatt

Date: 07:19:34 03/08/01

Go up one level in this thread


On March 08, 2001 at 03:08:51, Jouni Uski wrote:

>I have read recommendations to use 4 or 8 megabytes for tablebase cache. But I
>increased it to 16 MB and it seems to work fine. Isn't RAM always faster to
>access than hard drive, so can't You simple allocate all free RAM to cache?
>Additionally is there may be difference between Chessbase engines and winboard
>engines in this respect?
>
>Jouni


It depends on the operating system and its efficiency at buffering file I/O.

The cache in Eugene's code simply gets hit _before_ the O/S is called to do
any file I/O.  Which means that stuff kept in Eugene's cache will be accessed
more efficiently than if we have to take a system call, even though the system
call might not result in any real I/O since the O/S might have the file block
already in its cache as well.

I would rather use Eugene's cache as it is simply a bit more efficient.  But
we are not talking about huge slow-downs if you use a small cache and let the
operating system do the caching itself...



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.