Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: software against hardware question

Author: Peter McKenzie

Date: 18:22:35 03/25/00

Go up one level in this thread


On March 25, 2000 at 17:35:27, Christophe Theron wrote:

>On March 25, 2000 at 13:16:02, blass uri wrote:
>
>>On March 25, 2000 at 11:49:38, Ed Schröder wrote:
>>
>>>On March 25, 2000 at 08:22:14, blass uri wrote:
>>>
>>>>suppose you give Fritz6a pentium90(the hardware that Fritz3 used when it won the
>>>>world championship)and give it to play against Fritz3 with the best hardware
>>>>that it can use today.
>>>>
>>>>Which program is going win at tournament time control?
>>>>The same question is for other programs.
>>>>
>>>>Do most programmers did bigger software improvement or the main reason that
>>>>their program has better rating is the hardware improvement?(I do not include
>>>>beginners who did mainly software improvements and I do not include people who
>>>>did not work or almost did not work on improving their programs in the last
>>>>years).
>>>>
>>>>I think that Genius3 had a clear advantage against other programs some years ago
>>>>and the fact that the programmer did almost nothing to improve his program is
>>>>the reason that hardware was dominant in 1994-1997 if you look at the best
>>>>program.
>>>>(I mean only to programs that people can buy and I ignore cray blitz or deep
>>>>thought).
>>>>
>>>>The reason that people in the israeli league decided that programmers has no
>>>>right to do upgrade is that they were afraid that programs will get 200 elo
>>>>improvement by an upgrade.
>>>>
>>>>I explained rami tal that I think that most of the improvement in the last years
>>>>if you look only at the best program are by hardware.
>>>>
>>>>I explained him that theoretically it is possible to do 200 elo improvement but
>>>>practically this thing does not happen and an upgrade can probably give
>>>>something like 0.5 point in 10 games.
>>>>
>>>>Uri
>>>
>>>I think software and hardware go hand in hand but that all the hardware
>>>speed-up of the past years is by far superior to software changes. This
>>>has been the case since day one and I wonder if it will ever change.
>>>
>>>Ed
>>
>>I remmeber that Amir Ban claimed in the past that half of the improvement is
>>software and half is hardware.
>>
>>I see from the ssdf list that there is a significant difference between Junior5
>>and Junior6 and it seems that in this case the hardware improvement is really
>>close to be equal to the software improvement(I do not include using more than
>>one processor in my comparison because the ability to use more than one
>>processor is also part of the software)
>>
>>Maybe the fact that you work with christophe is also going to help  Rebel-tiger
>>also to get similiar results.
>>
>>Uri
>
>
>Certainly. For example I estimate that Chess Tiger has improved by roughly 70
>elo points between mid '98 and mid '99. Part of this improvement comes from my
>cooperation with Ed, and another part comes from my personal work.
>
>Trying and testing new ideas is a very long process. I think that so far I have

Yes, I too have a huge number of ideas that I haven't had time to implement and
test...

>been able to use only 10% of the ideas that came from my discussions with Ed.
>That leaves room for further improvements, believe me!

Out of interest, what language to you and Ed use to discuss chess programming
ideas?  Dutch, French, or English?

>
>
>    Christophe



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.