Hi all,
Here's a what's-new on the Freecell Solver trunk as of the Subversion
revision 3174:
http://svn.berlios.de/svnroot/repos/fc-solve/trunk
1. It seems that now the code is finally a little faster with the card
flipped-bit logic compiled-out, whereas previously it was a little slower
for no good reason. As a result, I think I'll disable it by default.
2. Added a --rwd flag to ./Tatzer to build with RelWithDebInfo (release-with
-debug info).
3. I've added the --set-pruning / -sp flag which currently accepts only
the "r:tf" (short for "Run:tofounds") value. This sets the Horne play prune
of moving cards that can no longer be used to build other cards upon
to the foundations, and adding it to the end of all the existing foss-nessy
scans yields a significant boost in performance. The new heuristic was called
"enlightened-ostrich" (or "eo") for short, named by "Amadiro" who helped me
with the Black Hole solitaire solver and will be available in the next
release of Freecell Solver (3.4.0 probably).
4. ptr_state_key and ptr_state_val were merged into a pair again in one
struct.
5. There has been some work on the
fc-solve/presets/soft-threads/meta-moves/auto-gen directory, which generates
the presets. I've started written a program to help construct
solution-shortening heuristics, but didn't finish it yet.
6. Silenced the "skip all" warning on build-process.
7. Add some new scans that generate especially short solutions. See:
http://tech.groups.yahoo.com/group/fc-solve-discuss/message/1067
8. There's now an option to set a different hard-coded freecells' num. This
allows to solve 2-freecell deals (or those of any other number) at a
greater speed.
9. Add an experimental flag of "--trim-max-stored-states" that allows to
trim the states collection after a certain limit has reached and to continue
solving from the reduced states collection. It makes reaching a verdict take
much longer though.
10. Add "-l p4bb" and and "-l x64bb" to "./Tatzer" which don't use the
--fc-only flag.
11. Add support for -fwhole-program and static linking the Freecell Solver
executables. This added another speed boost.
12. New benchmarks - the MS 32K benchmark now runs at 79.5340600013733 seconds
(402.34 deals per second) on my old Pentium 4-2.4GHz machine, and at
31.2516779899597s on my newer Dual Core laptop (1,023.95 deals per second).
This is especially significant because we've passed the
80 seconds / 400 deals-per-second mark on the Pentium 4, and the 32 seconds
/ 1,000-deals-per-second mark on the laptop.
===
Enjoy!
Regards,
-- Shlomi Fish
--
------------------------------------------
Shlomi Fish http://www.shlomifish.org/
Electrical Engineering studies. In the Technion. Been there. Done
that. Forgot a lot. Remember too much.
Received on Sat Sep 11 2010 - 10:56:32 IDT