[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnubg] Measuring performance levels

From: Joern Thyssen
Subject: Re: [Bug-gnubg] Measuring performance levels
Date: Thu, 24 Oct 2002 07:03:26 +0000
User-agent: Mutt/1.4i

On Wed, Oct 23, 2002 at 12:10:12PM -0400, Douglas Zare wrote
> About how much time would a gnu 4-ply  (versus 2-ply for the side
> without decisions) rollout of n games take? Can you perform the
> variance reduction using 2-ply evaluations? 

The variance reduction implemented in gnubg is general, so we can do
variance reduction on n-ply evaluations. However, since it requires a
n+1 ply evaluation it becomes incresingly expensive.

It's also possible to a 2-ply vs 4-ply rollout [*]. A 2-ply chequerplay
takes approximately 5 seconds on my box, and 4-ply approximately 1-2
minutes, so each trial could take from 10 minutes to 1 hour. I think
this rollout should be combined with truncation when the bear-off
database is reached (which is not implemented yet).


[*] This rollout has to be set up using commands until Jim finished his
GTK work on the rollout settings dialog :-)

Joern Thyssen, PhD
Vendsysselgade 3, 3., DK-9000 Aalborg, Denmark
+45 9813 2791 (private) / +45 2818 0183 (mobile) / +45 9633 7036 (work)
Note: new mobile number!

reply via email to

[Prev in Thread] Current Thread [Next in Thread]