Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Wcrafty 16.2 hash bug?

Author: Robert Hyatt

Date: 06:43:52 01/05/99

Go up one level in this thread


On January 05, 1999 at 01:18:15, Jouni Uski wrote:

>On January 04, 1999 at 18:01:21, Robert Hyatt wrote:
>
>>On January 04, 1999 at 02:56:46, Jouni Uski wrote:
>>
>>>There is something wrong in hash with Wcrafty 16.2. When I increased hash from
>>>3 MB to 6 MB NPS goes about 20 % lower and also times to certain ply level were
>>>increased. This makes no sense - and please don't speak about hard disk swapping
>>>I have enough RAM - this isn't the reason.
>>
>>
>>try more positions.  This is not uncommon.  First, hashing should not affect
>>NPS in any form.  But it can and often does affect the "quality" of the search.
>>And that "quality" might be in the form where the search (say to 10 plies)
>>takes longer than before, but the search result has as much information as a
>>full 11 ply search without the hashing.
>>
>>Or feel free to post a couple of positions where you thing something is wrong
>>and I'll certainly look at them carefully.  16.2 had a rewrite of the hashing
>>code so something _could_ be wrong of course...
>
>After more testing everything seems to work correct! Sorry Bob.
>Crafty is now up to cemmercial programs in endgames (without TBs), but tactical
>speed is mediocre (at least in my P90).
>Two minor bugs in 16.2:
>1) "d" doesn't display ascii board
what happens?  I get a normal display when I do this...


>2) maximum search depth is always 32


this I can't explain.  I just ran fine.70 and stopped it at depth=37...  are
you running it as a fritz engine or something???




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.