Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fritz 7 (MMX v No MMX)

Author: William Penn

Date: 03:50:29 11/25/01

Go up one level in this thread


On November 25, 2001 at 06:03:13, Keith Kitson wrote:

>I have been running Fritz 7 on a 1.2 Athlon thunderbird (Win98) with 256Mb Hash
>(from 512 Mb ram) and utilising Infinite analysis.  I have found the MMX version
>to become unstable after several hours analysis.  As quoted elsewhere the
>nodecount is also down.
>
>I have found the 'no MMX' version to be more stable and much faster, and no
>unstability problems so far.
>
>Has anyone else had the MMX version of Fritz 7 eventually come to a grinding
>halt and lock the system up requiring switch off?  This has happened several
>times now, so I have switched to the 'no MMX' version permanently now.
>
>I would welcome any comments on what may be happening here with the MMX version.
>
>Is it perhaps creating memory problems on the system (i.e. using up all
>available ram, slowing the machine down until it grinds to a halt?)
>
>Keith

Which interface? I have tried both Fritz 7 engines under the Deep Fritz
interface (W98se 256MB Celeron 500) and don't have that problem. I would tend to
suspect some other software that is running simultaneously (multitasking). For
example I can't run my anti-virus software or firewall at the same time as
running Chessbase engines, while also online and accessing some java sites like
chesslab.com.  That is a certain recipe for the blue screen of death.  But if I
disable the anti-virus software and firewall, then I seldom have a problem.

After various tests I concluded that the engine with MMX is buggy because it
doesn't handle active Queens properly in analysis. It may give away the Queen,
or fail to capture opponent's Queen, so I can't trust it to produce reliable
analysis. That was the original MMX version, and I haven't tried the updated
version. I'm not sure if I will. That takes a fair amount of time and effort,
but I lost confidence in it.

The version without MMX is of course faster. Whether it is stronger, I don't
know. I'm not too interested in ordinary strength, but I do need reliable
analysis.

The worst I can say about the No MMX engine is that I've noticed one instance of
incredibly dumb endgame analysis. It failed to recognize that two connected
passed pawns one rank apart are unassailable by opponent's King. The trailing
pawn cannot be captured otherwise the leading pawn Queens. Very elementary
stuff. That means I am now unable to trust it for endgame analysis.

One final observation, although I'm not sure exactly what it means...
Chessmaster has a bad habit of producing "ghosts" on the taskbar. Ghosts look
like something (a program, or whatever) is running, but it actually isn't.
Sometimes they are empty, just a little rectangle on the taskbar without any
icon or text.  I guess that somehow the taskbar just isn't updated properly.
Usually the ghosts disappear if I click them, but occasionally a system reset is
necessary to get rid of them. I haven't noticed any ghosts with the Chessbase
engines before, but now the Fritz 7 engines produce them. So whatever
Chessmaster is doing in that regard, Fritz 7 is also now doing. I have seen them
mentioned elsewhere, and apparently some people have more problems with them
than others.
WP



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.