Hi all,
today's night I played with generating faster scans while using the
--depth-test-order flag. My results can be found
at /fc-solve/presets/variations-on-depth-test-order/ in the repository. Now,
one scan that held promise was this one:
Preset 13step == -to 01234567 -dto 13,0187465 -dto 26,0134587
Unsolved = 12 ; Iters = 448319
So I timed it and generated a new compound scan, and timed it. However,
to my surprise, while the new compound scan had a lower iterations count than
my best scan so far ("enlightened-ostrich"), it ran considerably slower:
<SHELL>
shlomif[fcs]:$trunk/fc-solve/source$ cat DUMPS-eo-1/dump002
Started at 1310065840.640898
Reached Board No. 4000 at 1310065844.169411 (total_num_iters=864468)
Reached Board No. 8000 at 1310065847.788268 (total_num_iters=2644625)
Reached Board No. 12000 at 1310065851.309104 (total_num_iters=3522950)
Unsolved Board No. 11982 at 1310065851.427864
Reached Board No. 16000 at 1310065854.917023 (total_num_iters=4399582)
Reached Board No. 20000 at 1310065858.445514 (total_num_iters=7896874)
Reached Board No. 24000 at 1310065862.003541 (total_num_iters=8781966)
Reached Board No. 28000 at 1310065865.465704 (total_num_iters=11354889)
Reached Board No. 32000 at 1310065869.282100 (total_num_iters=12285701)
Finished at 1310065869.290063 (total_num_iters=14015402)
shlomif[fcs]:$trunk/fc-solve/source$ cat DUMPS-preset1-1/dump002
Started at 1310065778.655146
Reached Board No. 4000 at 1310065782.466788 (total_num_iters=864281)
Reached Board No. 8000 at 1310065786.252990 (total_num_iters=2600223)
Reached Board No. 12000 at 1310065790.141122 (total_num_iters=4331294)
Unsolved Board No. 11982 at 1310065790.404619
Reached Board No. 16000 at 1310065794.082530 (total_num_iters=6080449)
Reached Board No. 20000 at 1310065797.709297 (total_num_iters=7803782)
Reached Board No. 24000 at 1310065801.698846 (total_num_iters=8692714)
Reached Board No. 28000 at 1310065805.487672 (total_num_iters=11270146)
Reached Board No. 32000 at 1310065809.387902 (total_num_iters=12147137)
Finished at 1310065809.391581 (total_num_iters=13879267)
shlomif[fcs]:$trunk/fc-solve/source$ perl scripts/time-fcs.pl
DUMPS-eo-1/dump002 DUMPS-preset1-1/dump002
DUMPS-eo-1/dump002:28.6491649150848
DUMPS-preset1-1/dump002:30.736435174942
</SHELL>
So I guess some move functions execute their iterations slower than other ones
(which may also depend on the starting position.)
Now that I've realised that, I may need to rethink my strategy for creating
compound scans.
Regards,
Shlomi Fish
--
-----------------------------------------------------------------
Shlomi Fish http://www.shlomifish.org/
First stop for Perl beginners - http://perl-begin.org/
XSLT is the worst thing since nonāsliced bread.
Please reply to list if it's a mailing list post - http://shlom.in/reply .
Received on Thu Jul 07 2011 - 17:07:43 IDT