Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Does SSDF list show diminishing returns? Some calculations

Author: Uri Blass

Date: 04:15:15 10/08/01

Go up one level in this thread


On October 08, 2001 at 06:19:37, Jouni Uski wrote:

>On October 08, 2001 at 06:07:01, Uri Blass wrote:
>
>>On October 08, 2001 at 03:13:22, Jouni Uski wrote:
>>
>>>In latest SSDF list benefit from Athlon 1200 Mhz to K6-2 450 Mhz is:
>>>
>>>Deep Fritz      72
>>>Gambit Tiger	80
>>>Tiger           71
>>>Shredder        99
>>>Junior          55
>>>Gandalf        142
>>>
>>>In average this is +84 points (weighted by number of games). In CSS 1/01 there
>>>is some benchmarks for Atlon with all these program. I compared them to my own
>>>K6-450 and got speedup:
>>>
>>>Deep Fritz      3,40
>>>Junior          3,43
>>>Shredder        4,73
>>>Tiger(both)     2,67
>>>Gandalf         5,30
>>>
>>>In average 3,70. This gives (log2/log3,7)*84 = 45 points for speed doubling.
>>>Previously it has been spoken about 60-70 points for doubling, so there is
>>>diminishing returns now!
>>>
>>>Jouni
>>
>>I think that the reason may be the fact that opening books
>>start to dominate when the hardware gets better.
>>
>>It may be interesting to get a nunn2 rating list when programs play without
>>opening books.
>>
>>I believe that the nunn2 positions are complicated enough so the result is
>>practically not decided in them by the initial position.
>>
>>I already do the nunn2 match between Tiger14 and Deep Fritz at depth 13 on p850.
>>
>>I think that I wasted 160 hours of computer time about these game and Tiger is
>>leading 23-18.
>>
>>This result proves nothing
>
>True. May be Tiger uses always 2 times more time to reach same depth. Why not
>simple test with same time level for both!?
>
>Jouni

There are some problems with it.

1)I use only one computer for my testing so the games are with ponder off
2)The target is not to find if tiger s better or worse than Fritz(ssdf does this
job)

The target is to see if there is deminishing returns from depth.
I need to compare depth 13-depth 13 results with depth 13-depth 12 in order to
see the returns from 1 ply.

I already did it for small depths and did not find significant deminishing
returns from depthes.

Uri



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.