Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Important: Never use WB-adapter of May 2001 with Shredder5 under CB-GUI

Author: José Carlos

Date: 00:58:50 09/17/01

Go up one level in this thread


On September 17, 2001 at 02:51:46, Christophe Theron wrote:

>On September 17, 2001 at 02:42:53, Uri Blass wrote:
>
>>On September 17, 2001 at 02:31:44, Christophe Theron wrote:
>>
>>>On September 17, 2001 at 01:22:41, Uri Blass wrote:
>>>
>>>>On September 17, 2001 at 00:17:12, Christophe Theron wrote:
>>>>
>>>>>On September 16, 2001 at 17:35:42, Peter Berger wrote:
>>>>>
>>>>>>On September 16, 2001 at 17:06:14, Christophe Theron wrote:
>>>>>>
>>>>>>>
>>>>>>>A naive question: why don't you just assume that The King is stronger than
>>>>>>>Shredder instead of calling for some obscure reason related to time controls?
>>>>>>>
>>>>>>>
>>>>>>>    Christophe
>>>>>>>
>>>>>>
>>>>>>He used another ( probably somehow broken) adapter before and got a 17-3 result
>>>>>>for TheKing. Now with a working adapter he gets more realistic results.
>>>>>>
>>>>>>He didn't suggest that either engine is any stronger I think.
>>>>>>
>>>>>>There is no reference to timecontrol at all - it's simply about some broken
>>>>>>Chessbase WinBoard adapter ( and that one definitely really is quite broken ).
>>>>>>
>>>>>>Regards,
>>>>>>pete
>>>>>
>>>>>
>>>>>
>>>>>"These results look very realistic since Shredder5 does not feel comfortabel on
>>>>>the slow hardware."
>>>>>
>>>>>This is a direct reference to some hypothetic difference in performance
>>>>>depending on the speed of the computer (which is the same as saying that the
>>>>>programs perform differently depending on the time controls).
>>>>
>>>>It is not the same because it is possible that Shredder earns more speed
>>>>relative to the king from fast hardware.
>>>>
>>>>I did not check it but I know that there are programs that earn more speed from
>>>>fast hardware.
>>>>
>>>>I know that gandalf feels better on fast hardware relative to Junior not because
>>>>of the time control but because it earns more speed from the fast hardware.
>>>>
>>>>Uri
>>>
>>>
>>>
>>>Using time controls twice as long is the same as doubling the computer speed.
>>>
>>>Faster hardware or longer time controls, that's the same.
>>>
>>>
>>>    Christophe
>>
>>The problem is that when Junior become twice faster  thanks to better hardware
>>Gandalf becomes 3 times faster.
>>
>>These are not the exact numbers but this is the idea.
>>
>>Uri
>
>
>It's not going to happen on any hardware.
>
>It can only happen if the hardware architecture has a major change (generally a
>different generation of processor core).
>
>And it is not guaranteed that the next update in hardware architecture would
>give a similar result. In the next update, Junior could be 3 times faster and
>Gandalf only twice faster (maybe due to simpler code that puts less stress on
>the branch prediction system of the P4 for example, who knows).
>
>In any case, on a given processor, doubling the speed is the same as having time
>controls twice longer.
>
>
>    Christophe

  No it isn't at all.
  If I use EGTB, I'll benefit from increasing the disk speed. Otherwise, I
won't.
  If my program is very big, I'll benefit from increasing the size of the cache.
If it is small enough, I won't notice any difference.
  If I have a bunch of arrays with precomputed data, I'll benefit from fast
memory more than if I calculate things on the fly, in which case the processor
speed will be more important.
  I can give you more examples, but I think it's enough. Different hardware
cannot be simulated by modifying the time controls. This is so obvious that I
can't belive you deny it.

  José C.



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.