|
From: | Michael Petch |
Subject: | Re: [Bug-gnubg] Cache question |
Date: | Thu, 03 Sep 2009 07:39:59 -0600 |
User-agent: | Microsoft-Entourage/12.20.0.090605 |
The cache uses a hash function to map the position onto a cache entry, there is
every chance of several positions mapping to the same position (the cache stores
2 positions per entry to minimise this).
The reason things get better even when the cache is bigger than the total number
of evaluations is there will be less collisions in the cache. There is a
#define you can set and then a command to get the cache statistics (lookups, hits).
It might be possible to improve the hash function (GetHashKey), but it does seem
to do quite a good job.
Jon
[Prev in Thread] | Current Thread | [Next in Thread] |