I feel like I ought to respond, given that I've been trying for almost
3 years to beat the performance of the solver I released 3 years ago.
I've tried a lot of things similar to what you are suggesting, and have
abandoned them. I believe I'm working on the most likely-to-succeed
approach yet. Very roughly it involves gathering 14-bits of data on
each viable alternative move at any given "board layout." This data
is used to index various lookup tables (256 to 1024 bytes in length).
The lookups are extremely fast, and the initial data can be computed
at a rate several times as fast as my old version processes moves.
I'm in its neighborhood on solution length and processing speed, and
expect to beat it on all measures with this approach. -Gary Campbell
----- Original Message -----
From: Shlomi Fish
To: fc-solve-discuss_at_yahoogroups.com
Sent: Monday, December 28, 2009 7:30 AM
Subject: Idea: Adapt the scans based on the parameters of the initial board.
Hi all!
Shortly after I worked on the meta-scan optimisation (see the previous
messages), I came up with this idea to further improve the Freecell Solver
performance:
<<<<<<<<<<<<<<<<<<<
* Adapt the scans based on the parameters of the initial board.
** Try to find a correlation between various parameters of the initial board
(such as those calculated in the A* scan or the number of steps required to
sort the cards in each column by rank), and the performance of various scans
and then:
+
1. Calculate the initial parameters on startup.
+
2. See what would be a good meta-scan based on them.
+
3. Use it.
>>>>>>>>>>>>>>>>>>
Any comments?
Regards,
Shlomi Fish
--
----------------------------------------------------------
Shlomi Fish http://www.shlomifish.org/
What Makes Software Apps High Quality - http://shlom.in/sw-quality
Bzr is slower than Subversion in combination with Sourceforge.
( By: http://dazjorz.com/ )
Received on Mon Dec 28 2009 - 08:37:16 IST