Computer Chess Club Archives


Search

Terms

Messages

Subject: Incremental PCSQ Updates Re: SEE results

Author: Brian Richardson

Date: 08:30:56 08/10/04

Go up one level in this thread


On August 10, 2004 at 11:16:56, Tord Romstad wrote:

>On August 10, 2004 at 11:05:03, martin fierz wrote:
>
>>On August 10, 2004 at 10:59:29, Tord Romstad wrote:
>>
>>>On August 10, 2004 at 10:35:29, Anthony Cozzie wrote:
>>>
>>>>Plus, if you have only PST eval you should be getting 3-4M nps, so SEE probably
>>>>slows you down a _lot_.
>>>
>>>Yet another proof of how bad my programming skills are.  With PST eval and
>>>nothing
>>>else, I get about 800,000 nps (on a PIV 2.4 GHz).  Adding SEE slowed me down to
>>>around 750,000 nps.
>>>
>>>Tord
>>
>>and nothing else means... what?
>>
>>for example, are you computing hashkeys, and doing hashstores & hashlookups? are
>>you checking for repetitions?
>
>Yes to all of the above.
>
>>are you computing attack information on the way?
>
>No.
>
>>are you making some kind of complicated decision on extensions and reductions?
>
>No.
>
>This was a very simple recursive null-move searcher with no fancy pruning or
>reductions, and no extensions except the most elementary (checks and
>single replies to checks, IIRC).  The eval consisted of material and
>piece square tables, which were updated incrementally when moves were
>made and unmade.

Incremental PCSQ updating were a net loser for my program, I suspect
because there are more make/unmakes than evaluations, although pawn
hashing may have factored in, so I was actually only incrementally
doing pieces, not pawns.

Brian

>
>After adding some additional eval terms, a whole-board swapoff function, and
>some simple forward pruning, I'm now down at about 450,000 nps.  It's still
>more than twice as fast as my old program, though.  :-)
>
>>and so on - it doesn't mean your programming skills are bad.
>>
>>on the other hand, fast engines probably run at around 1Mnps on your machine,
>>and they do all kinds of other stuff too...
>
>Yes, they do.  Sometimes they are even faster than 1M n/s.
>
>Tord



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.