Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: ponder_on ponder_off comparision

Author: Mogens Larsen

Date: 15:14:40 07/19/00

Go up one level in this thread


On July 19, 2000 at 16:41:11, Volker Pittlik wrote:

>On July 19, 2000 at 13:19:07, Mogens Larsen wrote:
>
>[...]
>
>If I'm allowed to give a statement.
>
>Firstly what I wanted to test is: are the the results of ponder _on and
>ponder_off tournaments are comparable. And it's very likely to me they are.
>
>If you you look in the results in my games archive you will easily find that
>Crafty's outcoming in the ponder_off round is probably randomnly. There isn't
>any evidence that Crafty is playing stronger with ponder_off. I assume _every_
>program is playing stronger with ponder_on.
>
>Secondly Crafty wasn't the only program envolved in this test. The outcoming of
>all other program are nearly the same in both tournaments . Maybe _all_ other
>authors of the other programs have implemented ponder_off time management in a
>similar rudimentary way as it has done in Crafty. But this seems very unlikely
>to me.
>
>So what to do next to test the effect of ponder_on? I don't know yet. To do
>another test with other engines, different time controls, more/less games seems
>to be the wrong way because I don't know anything about the time management in
>the other programs.
>
>To use special versions of the programs which are allways or never ponder are
>possibly the way to do such a test.
>
>Volker

I agree with most of what you say, because my experience is very similar when
comparing my own results with ponder=on tournaments. The random remark I protest
against, is the one claiming that the results of ponder=off is random.
Obviously, Crafty isn't stronger with ponder=off, but it might very well be
stronger compared with other programs using ponder=off than with ponder enabled.
As far as I'm concerned ponder=off results are good indicators of relative
strength between programs. Your results don't prove that, but they indicate that
my hunch isn't too far off, which is good enough for me. More data would be nice
though.

Best wishes...
Mogens



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.