Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: about search space, state space, and moore's law

Author: Uri Blass

Date: 13:49:13 11/04/02

Go up one level in this thread


On November 04, 2002 at 16:07:15, martin fierz wrote:

>On November 04, 2002 at 13:08:13, Dan Andersson wrote:
>
>>But there was an option of having some hash there. And even if it was
>>impossible. Near perfect move ordering can be achieved anyway. Half depth search
>>and heuristics does it well. Deep tactics will be missed, but it would probably
>>be missed in any case.
>>
>>MvH Dan Andersson
>
>hash is not an option because if you want to search trillions of nodes/sec, you
>cannot wait for a slow memory read...
>
>aloha
>  martin

I suspect that if you can search trillions of node/second by a single processor
than the memory is always going to be fast.

You also do not need to remember every position in the hash tables and
remembering part of the positions is enough to get a significant improvement.

Speed+Selective search+Hash may solve chess(at least people cannot prove that it
cannot solve chess)

We do not need to know the exact distance to mate in order to be sure that
white is winning after 1.e4 e5 2.Nf3 Qh4 and there are a lot of positions that
it may be possible to save the analysis.

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.