Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: What's the latest Athlon for Chess?

Author: Aaron Gordon

Date: 11:08:30 12/21/02

Go up one level in this thread


On December 21, 2002 at 08:54:58, Bob Durrett wrote:

>On December 20, 2002 at 23:33:35, Aaron Gordon wrote:
>
>>On December 20, 2002 at 21:55:31, Bob Durrett wrote:
>>
>>>On December 20, 2002 at 21:32:18, Sean Mintz wrote:
>>>
>>>>It's a hobby for some people to overclock their computers and for some it's
>>>>interesting to see what they can get done. Also, it's just one way to compare
>>>>AMD/Intel in a way a lot of people only understand (mhz to mhz). I think
>>>>everyone used the same hash sizes for that benchmark.
>>>
>>>What hash sizes do you suppose they use?  Anybody?
>>>
>>>Bob D.
>>
>>http://www.talkchess.com/forums/1/message.html?271912
>
>I must be getting senile.  The question was already answered.  : (
>
>But the main point is that the hash table sizes used for the tests was the
>absolute minimums.  That means that the Kn/s figures given in the Crafty v18.11
>Benchmark test were INFLATED.  It would have been much more informative to see
>the Kn/s figures for *realistic* hash table sizes.  Why?  Because if someone
>[like me] is considering the purchase of a newer [supposedly faster] computer,
>then they will want wish to know what to expect.
>
>Bob D.

I used the default hash size becase as I said before crafty scales linearly and
it's easier for people to test, they don't need to use a crafty.rc file or mess
with anything. Most of the computers on that list are from friends that haven't
the slightest clue about computer chess.

Also one thing I've noticed here when you crank up the hash table on a P4 and
Athlon is the Athlon actually pulls ahead even more. Why this is I'm not sure
(because it's done it on DDR & RDRAM P4's). Like a P4-2.8 to P4-3.0 would hold
basically even with my AthlonXP @ 1.8-1.86GHz at lower hash sizes, when we used
256mb (for like Hiarcs8, Fritz 7, Crafty, etc) my Athlon would start pulling
away and take the lead. If anyone could explain why this is I'd appreciate it.
You'd figure it would be the other way around (since P4's have more memory
bandwidth).



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.