Main
Date: 10 Apr 2006 02:28:02
From:
Subject: multiple engines - two questions - analysis strength
1.

I like to look over GM games with 2-3 engines running. that way, I get
a sense of different engine "personalities." it's cool and SEEMS
illuminating.

I wonder though if I'm getting really dumbed down analysis. How much
weaker would Hiarcs, Fritz, and Junior all running together really be?

In other words, am I better off running just one at a time?

fyi I have a Pentium M 1.86 ghz with 1.00 gb ram.

2.

related:

how much does hashtable size effect analysis strength?

thanks.





 
Date: 10 Apr 2006 13:30:30
From: David Richerby
Subject: Re: multiple engines - two questions - analysis strength
<[email protected] > wrote:
> 1.
>
> I like to look over GM games with 2-3 engines running. that way, I get
> a sense of different engine "personalities." it's cool and SEEMS
> illuminating.
>
> I wonder though if I'm getting really dumbed down analysis. How much
> weaker would Hiarcs, Fritz, and Junior all running together really be?

I'm assuming you only have one CPU.

Suppose you let the engines run for fifteen minutes. That's
equivalent, roughly, to each engine running for five minutes on its
own, so the quality of the analysis you get is weaker. In fact, it
will be ginally weaker than running each engine for five minutes in
turn because the computer will spend some time switching from one to
another (this is known as the context switch overhead).

Clearly, each program would make a better recommendation if it was
able to think for the whole fifteen minutes instead of having to
share.


> In other words, am I better off running just one at a time?

Depends what you want. If you want to have the best move recommended
in a given time, you should choose the best engine for the position
and run it exclusively. But then you lose the interest of watching
how the multiple engines think compared to one another.


> how much does hashtable size effect analysis strength?

It doesn't. Hash table size affects analysis *speed*. Now, in a
game, speed is everything because there's a clock ticking. In
analysis, you'll get to a given ply depth faster with a bigger hash
table but you could achieve the same result with a smaller hash table
by waiting a little longer. In practice, you should set the hash
table as large as you can without causing it to swap out to disc.
Note that, if you want to do other things while the analysis is
running, you need to make sure they can fit in physical memory, too.


Dave.

--
David Richerby Beefy Puzzle (TM): it's like an
www.chiark.greenend.org.uk/~davidr/ intriguing conundrum that's made from
a cow!


  
Date: 10 Apr 2006 13:32:30
From: David Richerby
Subject: Re: multiple engines - two questions - analysis strength
David Richerby <[email protected] > wrote:
> <[email protected]> wrote:
>> In other words, am I better off running just one at a time?
>
> Depends what you want. If you want to have the best move recommended
> in a given time, you should choose the best engine for the position
> and run it exclusively. But then you lose the interest of watching
> how the multiple engines think compared to one another.

Sorry, an afterthought...

You could get the best of both worlds by running your three engines
simultaneously and then running them one at a time only when they
disagree. In many positions, I imagine they'll be recommending the
same moves but the interest comes in those positions where they have
different ideas.


Dave.

--
David Richerby Evil Laptop Atom Bomb (TM): it's
www.chiark.greenend.org.uk/~davidr/ like a weapon of mass destruction
that you can put on your lap but it's
genuinely evil!


 
Date: 10 Apr 2006 02:57:31
From: [email protected]
Subject: Re: multiple engines - two questions - analysis strength
The main thing is whether you enjoy it that way. You don't get the same
ply depth, but you get the contrast between the different engines. So
its a tradeoff, the main thing is how it affects your own human chess
perception, whether you get more out of it one way or the other. You
learn different chess lessons doing it the two different ways, if it
helps your chess perception and you enjoy it then you should keep doing
it.