[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] Re: Bug report gnugo-3.3.8
From: |
Gunnar Farneback |
Subject: |
Re: [gnugo-devel] Re: Bug report gnugo-3.3.8 |
Date: |
Wed, 11 Sep 2002 17:41:32 +0200 |
User-agent: |
EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode) |
Markus wrote:
> I already set level to 0.
I should have guessed you did.
> If you say, that GnuGo was already using its persistent reading
> cache, then there is probably not much room for gaining speed.
There are a few more things you can do.
1. Drastically reduce the value of max_nodes_connect in
engine/readconnect.c
2. Reduce the value of max_connect_depth2 in engine/readconnect.c
3. Make the persistent caches more aggressive in reusing read results
by modifying the active areas computed in
store_persistent_owl_cache() in engine/owl.c and in
store_persistent_reading_cache() in engine/reading.c. The smaller
the active area, the more likely the result is to be reused.
The connection limits are not yet affected by the level, but they
should be.
> Right now, GnuGo does less than 3-5 examine_positions per second
> at level 0 on my 1GHz Duron. To make a NeuroGo training possible
> I would need at least 10-20 evaluations per second.
So you need a speedup of about a factor 4. But I suspect that speed
isn't everything. How much worse can you allow GNU Go to play before
it becomes useless for your purposes?
/Gunnar