| |
Main
Date: 21 Jan 2007 20:40:49
From: george
Subject: how to test mp engines within chessbase gui on dual core machines
|
I have a dual core 4200 and several MP engines, running without the Fritz10
GUI, I was wondering if I wanted to test a MP program such as deep fritz or
loop64:32 would I use ponder on or ponder off? Also if I run two MP engines
against themselves should I turn off ponder or not?
thanks for the help
|
|
| |
Date: 22 Jan 2007 11:02:18
From: Dave (from the UK)
Subject: Re: how to test mp engines within chessbase gui on dual core machines
|
george wrote:
> I have a dual core 4200 and several MP engines, running without the Fritz10
> GUI, I was wondering if I wanted to test a MP program such as deep fritz or
> loop64:32 would I use ponder on or ponder off?
Depends what you are trying to test.
> Also if I run two MP engines
> against themselves should I turn off ponder or not?
> thanks for the help
From a 'fairness' point of view, it should make no difference - either
ponder on or off (assuming same on each) will give each an even chance.
But you would I suspect get a better game (engines able to find a better
move), by having ponder off. When it is off, the engine on move will
have 100% of the CPU time to think about the move to be played. If
ponder is on, the engine on move has only 50% of the CPU time - the
engine not on move has the other 50%. But it can't put that 50% CPU time
to such effective use as the engine on move, as it will have to consider
all moves the other engine can play. That would limit it to one ply
deeper I suspect.
If I'm playing a game, I get the same time as my opponent. I can think
in his time. But if instead of thinking in my opponents time, I had
twice as long to think on my own move, it would be preferable.
Hence I feel ponder should be off in engine vs engine games.
I'm no expert though - this just seems intuitive to me.
--
Dave (from the UK)
Please note my email address changes periodically to avoid spam.
It is always of the form: [email protected]
Hitting reply will work for a few months only - later set it manually.
http://chessdb.sourceforge.net/ - a Free open-source Chess Database
|
| | |
Date: 22 Jan 2007 17:01:25
From: Gian-Carlo Pascutto
Subject: Re: how to test mp engines within chessbase gui on dual core machines
|
Dave (from the UK) wrote:
> Hence I feel ponder should be off in engine vs engine games.
Not if you're testing 2 single CPU engines on a dual CPU system, then
ponder is fine, since each will get the full CPU power they can use.
But for 2 dual engines on a dual system, it should be off again.
--
GCP
|
| | | |
Date: 22 Jan 2007 17:31:37
From: Dave (from the UK)
Subject: Re: how to test mp engines within chessbase gui on dual core machines
|
Gian-Carlo Pascutto wrote:
> Dave (from the UK) wrote:
>
>
>>Hence I feel ponder should be off in engine vs engine games.
>
>
> Not if you're testing 2 single CPU engines on a dual CPU system, then
> ponder is fine, since each will get the full CPU power they can use.
>
> But for 2 dual engines on a dual system, it should be off again.
>
Yes, I forgot to say that. When I play engine vs engine on this machine,
assuming the engines are single threaded, I still leave ponder on as it
has 4 CPUs (rather slow old ones it must be said, but still 4).
--
Dave (from the UK)
Please note my email address changes periodically to avoid spam.
It is always of the form: [email protected]
Hitting reply will work for a few months only - later set it manually.
http://chessdb.sourceforge.net/ - a Free open-source Chess Database
|
|
