Re: [gnugo-devel] cosmic patch

[Arend Bayer]

Gunnar wrote:

> Previously the Zobrist numbers never had any effect on the reading
> (except in the extremely unlikely occurence of a board hash
> collision). The new cache together with Evan's killer heuristic
> changes this. The Zobrist numbers affect the placement within the hash
> table implementing the cache. More to the point they affect which
> positions/read routines/targets/komasters compete about the same
> entries in the hash table. Due to the design of the new cache, it

It is worth pointing out that changing the size of the hash table
has exactly the same effect.
(It also changes which "positions" compete with each other about the same
entries: Entries compete for the same entry if
        hashval(position, routine_id, targets, komaster state) % cache_size
are identical.)

I think it is near to impossible to make, say, readconnect results
completely independent of
a) cache size,
b) what has happened before (i.e. how the cache is populated at the
beginning.)

(Since readconnect has a node limit in its reading, such as owl.)

Arend