Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Questions for Bob ...

Author: Robert Hyatt

Date: 07:09:22 02/02/04

Go up one level in this thread


On February 01, 2004 at 21:36:13, Mike Byrne wrote:

>On February 01, 2004 at 21:31:35, Robert Hyatt wrote:
>
>>On February 01, 2004 at 21:05:41, Mike Byrne wrote:
>>
>>>On February 01, 2004 at 20:56:28, Robert Hyatt wrote:
>>>
>>>>On February 01, 2004 at 20:54:14, Mike Byrne wrote:
>>>>
>>>>>On February 01, 2004 at 20:50:56, Robert Hyatt wrote:
>>>>>
>>>>>>On February 01, 2004 at 20:47:11, Mike Byrne wrote:
>>>>>>
>>>>>>>for CCT 6 , did you use the /DDETECTDRAW option ?  /DFUTILITY?
>>>>>>
>>>>>>I did not do /DDDETECTDRAW.  It is cute but it also slows it significantly.  I
>>>>>>did use -DFUTILITY as recent testing has shown that it seems (to me) to be
>>>>>>working better with than without.  I've used it off and on for months...
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>I suspect not in the  /DDETECTDRAW - just want to double check.
>>>>>>>
>>>>>>>I would guess yes for /DFUTILITY.  I want to be sure Crafty 19.10 SE is exactly
>>>>>>>how you use it CCT 6.
>>>>>>>
>>>>>>>Do you have  a node count for single CPU running the "bench" command?
>>>>>>
>>>>>>On the opteron?:
>>>>>>
>>>>>>Running benchmark. . .
>>>>>>......
>>>>>>Total nodes: 71122647
>>>>>>Raw nodes per second: 1975629
>>>>>>Total elapsed time: 36
>>>>>>SMP time-to-ply measurement: 17.777778
>>>>>>White(1): quit
>>>>>>
>>>>>
>>>>>I'm not getting that node count - is that with default hash sizes?
>>>>>
>>>>>Do you get the same on your laptop?
>>>>
>>>>Ugh.  No.  I ran it in the directory that had the .craftyrc for the cct event.
>>>>I'll try again when crafty isn't playing...
>>>>
>>>>
>>>>>
>>>>>>
>>>>>>>
>>>>>>>TIA
>>>>>>>
>>>>>>>Mike
>>>
>>>I am getting:
>>>
>>>Total nodes: 89942714 with default hash
>>
>>. with -DFUTILITY?
>>
>>Running benchmark. . .
>>......
>>Total nodes: 89942714
>>Raw nodes per second: 2091691
>>Total elapsed time: 43
>>SMP time-to-ply measurement: 14.883721
>>White(1):
>
>
>>
>>max threads set to 4
>>White(1): bench
>>Running benchmark. . .
>>......
>>Total nodes: 88657759
>>Raw nodes per second: 7388146
>>Total elapsed time: 12
>
>whoaaa ...that's hummimg  ...
>
>>SMP time-to-ply measurement: 53.333333
>>White(1): quit


What I'd like to do next is incorporate the personality stuff into Crafty.

However, what I want to do most is to change it to a big table-driven affair
where each personality represents one row of a large matrix, maybe the first
column is the personality name, the next N colums contain the evaluation scaling
factors, the search extension amounts, etc.  IE that would make the code in
option.c very small and compact, and all the "parameters" move to data.c where
the array is initialized.

This is probably a good point to do such stuff, as version 20 is going to start
a new binary book format and so forth for total portability, as well as cleaning
up the Makefile options to make things a bit simpler and saner.  Best to get all
the other changes in before making that leap...

Bob



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.