Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shreddermark linearity

Author: Aaron Gordon

Date: 15:05:52 02/11/04

Go up one level in this thread


On February 11, 2004 at 16:34:05, Roy Eassa wrote:

>My PC is a 1.2 GHz Athlon.  I get a ShredderMark of 556 / 128k nps when Shredder
>8 is set to 190k hash.
>
>I just bought a cheap very Dell system that usess a 2.8 GHz Pentium IV.  I
>figured it would be less than twice as fast.  Instead, I get a ShredderMark of
>2227 / 297k nps (also using 190k hash)!
>
>First, why is the ShredderMark 4x as high when the integer performance of the
>CPU should be less than 2x as high?
>
>Second, why isn't the ShredderMark's improvement proportional to that of the
>nps?

First of all.. don't bother with Shreddermark. It is messed up. The test is too
short and provides false results. To benchmark Shredder try this, and I will get
some results for you to compare to here later...

Follow these instructions carefully.

Go into your chessbase\engines directory and rename (or delete) shredder.plr.
This is the learning data. EVERY time you run this test you will have to delete
it.

Now, load shredder 8.. make sure it is set to the start position. Set the hash
to 128mb. Now, hit ALT-F2 to initiate infinite analysis. Wait until ply 18 is
finished. Once it is finished at 18 look on the far right.. it should say
"29791kN". Take this number and divide it by the time taken to that ply in
seconds. For example.. if your cpu takes 2 minutes (2:00), thats 120 seconds.
29791 / 120 = 248kn/s.  Test it and let me know. I'll get some results on
various Duron, Athlon Tbird and Athlon XP systems tonight...



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.