Hi all!
As you may recall from:
http://tech.groups.yahoo.com/group/fc-solve-discuss/message/331
I implemented a prelude for the scans which improved the performance of multi-
tasking scans in Freecell Solver considerably. Recently, I got the idea to run
the greedy prelude-generating algorithm to have it use quotas of 1 iterations
each to make it super-greedy.
Today I implemented the result. This involved tweaking the Perl source code of
the prelude-generating program to accept a subroutine reference as well as an
array (because I figured out that a giant array of 1's will consume a lot of
memory in Perl). After I did that, I was able to run it.
The results were abysmal. The program choose only a single scan, and ended up
at 40 million iterations in total, while the 350-per-iteration scan was about
18 million iterations in total. I suspected such a result would happen and now
my suspicions were confirmed.
In any case, I checked in the script that generates it into the repository.
Some other trunk news:
1. I passed the repository revisions 1,248 and 1,250.
2. I did some extensive refactoring of scans.c: I removed many accessors for
the_soft_dfs_info and some unused variables. Many no-longer-existent
declarations of functions in fcs.h were removed.
3. Hard-DFS was removed. It just occupied space and was completely underused.
Specifying it now uses Soft-DFS instead which should yield the same results,
but has some advantages.
4. I updated the links page with several new links and updated some old links.
Regards,
Shlomi Fish
--
-----------------------------------------------------------------
Shlomi Fish http://www.shlomifish.org/
Funny Anti-Terrorism Story - http://xrl.us/bjn7t
God gave us two eyes and ten fingers so we will type five times as much as we
read.
Received on Mon Mar 16 2009 - 10:39:12 IST