The following parallel run:
#!/bin/sh
./mptest 1 32000 20 \
--method random-dfs -seed 1 -to '[0123456789]' --next-instance \
--method random-dfs -seed 2 -to '[01][23456789]' --next-instance \
--method random-dfs -seed 3 -to '[01][23467]' --next-instance \
--method a-star --next-instance \
--method a-star -to 0123467 --next-instance \
--method random-dfs -seed 4 -to '[0123467]'
Managed to solve the Microsoft 32,000 in a record number of 29,614,897
iterations. It took it a total of 95 minutes on a Pentium III 667 MHz,
making it 336 boards per minute or 5 per second.
Furthermore, there was only one intractable state - the notorious board
No. 1941.
In any case, I encountered a memory leak in test_multi_parallel.c in the
process and it was fixed in FCS 1.11.24. I have a very fast NT computer
here in the Com-Net Lab farm, and I can try to run both
test_multi_parallel and FC-Pro with Don Wood's solver and see which one
finishes faster.
A funny thing is that a vanilla soft-dfs scan can solve board 1941 in 5989
iterations short. Maybe I'll add it to the scan and see how such a
benchmark fare.
Regards,
Shlomi Fish
--
----------------------------------------------------------------------
Shlomi Fish shlomif_at_vipe.technion.ac.il
Home Page: http://t2.technion.ac.il/~shlomif/
Home E-mail: shlomif_at_techie.com
If:
1. A is A
2. A is not not-A
does it imply that
1. B is B
2. B is not not-B
Received on Mon Dec 03 2001 - 11:10:41 IST