Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: PONDER=ON and TableBases on 1 PC

Author: Robert Hyatt

Date: 07:53:43 07/28/01

Go up one level in this thread


On July 27, 2001 at 16:55:55, Dann Corbit wrote:

>On July 27, 2001 at 16:51:08, Roy Eassa wrote:
>
>>On July 27, 2001 at 04:32:48, Dann Corbit wrote:
>>
>>>On July 27, 2001 at 04:23:37, Matthias Gemuh wrote:
>>>
>>>>Hi Experts,
>>>>are there any possible conflicts when 2 engines use the same TBS with Ponder ON
>>>>?
>>>
>>>Shouldn't be.  Unless someone did something stupid.
>>>
>>>They're just reading, after all.  I am assuming that you are talking about a
>>>multiple CPU machine, so that ponder=on isn't pure nonesense.
>>
>>
>>Be careful, Dann.  Dr. Hyatt has argued strongly that ponder should always be
>>on, even with a single CPU.  (It seemed counter-intuitive to me too, but you
>>should check out his recent postings -- over the past couple days, I think.)
>
>Not when both engines play on the same machine.
>No way.


Think about it before rushing to judgement.  Points to consider:

1.  most engines are designed and tested thoroughly in ponder=on mode, as that
is _obviously_ the strongest mode the engine has.  If you don't believe this,
play a match against an opponent with ponder=on and ponder=off and see how much
better ponder=on does.

2.  if both programs run with ponder=on, they will be 100% compute-bound for
the entire match.  That means that each will be getting 50% of all available
cpu cycles, which is totally fair.

3.  if you run both with ponder=off, they will also be getting 50% of the total
cpu resources, but on occasion one can get 75%.  You only have to look at past
chessmaster programs that would compute like mad when it was to move, then it
would poll for keyboard input like mad when not pondering, effectively remaining
100% compute-bound whether it was searching or not.  This gave it a significant
advantage over its opponent as it would get 100% of the machine when it was
thinking, but the opponent would only get 50% when it was thinking.

I don't see _any_ reasonable explanation of how ponder=off on a single-cpu
machine hurts a thing, in terms of being totally fair and giving the engine its
best chance to perform as it was designed to perform.




This page took 0.01 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.