Hi Steve,
game and can't build a computer that can give us the winning
strategy.
simplified versions of them to prove their correctness.
Here's an example:
name normalsize compressedsize rate% moves
spec304-65500.lpb 65566 94 0.14336 !!!
411.lpb.bz2 45547 5730 12.5804
Burgthea.lpb.bz2 4775 789 16.5236 ~2800
FearNoEv.lpb.bz2 4004 848 21.1788 ~1800
EDGECNT3.LPB.bz2 3628 864 23.8148
BURGTJIM.LPB.bz2 3022 570 18.8617
JERIKHO.LPB.bz2 1660 292 17.5904
Level "Fera No Evil" has less moves in it's solution
than level "Burgtheater", but a greater complexity!
I also made some tests with files having n times the same letter.
2004.02.01 19:07 47 9999.txt.bz2
2004.02.01 19:07 47 99999.txt.bz2
2004.02.01 19:15 48 999999.txt.bz2
2004.02.02 01:03 117 110m.bz2
2004.02.02 01:04 273 330m.bz2
2004.02.02 01:04 1 424 1980m.bz2
2004.02.02 01:10 141 9999.zip
2004.02.02 01:10 230 99999.zip
2004.02.02 01:10 1 103 999999.zip
2004.02.02 01:08 107 037 110m.zip
2004.02.02 01:09 320 865 330m.zip
You can see why haven't I suggested pkzip ;)
You can also see that the ~20000 move lpb for level 411
has compressed size 5730, which is much greater than
the compressed size 1424 of an almost 2 Gigabyte large file
with the a single repeated character (1 980 000 000 moves,
without header).
I think we should consider to cut off the header before
compressing for comparision. Little difference can
be caused in the compressed size depending on the level name :(
And finally:
you have good chances to save a bzip2-ed lpb for the
trinary levels: about 6,652,800,000,000 bytes
can be sent through bzip2 in a week -- on a 2Ghz P4.
That is, you'll have it done for 10^15 moves in less
than 3 years ;-)
and 6 secs with small size + window sent to lower left corner
so that changing bitmap is not shown. Maybe this trick is
already used by Donald.
m.
>Not surprising. We have the same with chess: we have a 3000 year old
> that's pretty funny - we don't have computers fast enough to solve
> the VERY long levels in our lifetimes, but we also don't have hard
> drives big enough to store the playbacks.
game and can't build a computer that can give us the winning
strategy.
> have you zipped the lpb files for theI don't have the lpb for my longest ones. I solved only
> levels in your world record list yet to determine how much they
> compress? i'd like to see your results.
simplified versions of them to prove their correctness.
Here's an example:
name normalsize compressedsize rate% moves
spec304-65500.lpb 65566 94 0.14336 !!!
411.lpb.bz2 45547 5730 12.5804
Burgthea.lpb.bz2 4775 789 16.5236 ~2800
FearNoEv.lpb.bz2 4004 848 21.1788 ~1800
EDGECNT3.LPB.bz2 3628 864 23.8148
BURGTJIM.LPB.bz2 3022 570 18.8617
JERIKHO.LPB.bz2 1660 292 17.5904
Level "Fera No Evil" has less moves in it's solution
than level "Burgtheater", but a greater complexity!
I also made some tests with files having n times the same letter.
2004.02.01 19:07 47 9999.txt.bz2
2004.02.01 19:07 47 99999.txt.bz2
2004.02.01 19:15 48 999999.txt.bz2
2004.02.02 01:03 117 110m.bz2
2004.02.02 01:04 273 330m.bz2
2004.02.02 01:04 1 424 1980m.bz2
2004.02.02 01:10 141 9999.zip
2004.02.02 01:10 230 99999.zip
2004.02.02 01:10 1 103 999999.zip
2004.02.02 01:08 107 037 110m.zip
2004.02.02 01:09 320 865 330m.zip
You can see why haven't I suggested pkzip ;)
You can also see that the ~20000 move lpb for level 411
has compressed size 5730, which is much greater than
the compressed size 1424 of an almost 2 Gigabyte large file
with the a single repeated character (1 980 000 000 moves,
without header).
I think we should consider to cut off the header before
compressing for comparision. Little difference can
be caused in the compressed size depending on the level name :(
And finally:
you have good chances to save a bzip2-ed lpb for the
trinary levels: about 6,652,800,000,000 bytes
can be sent through bzip2 in a week -- on a 2Ghz P4.
That is, you'll have it done for 10^15 moves in less
than 3 years ;-)
> "theoretical max" takes about 40 minutes onDid you mean 40 secs? On my 2Ghz machine its 40secs with large size
> my PC to verify the solution with "f8 - resume recording" (which is
> much faster than a normal playback "f7").
and 6 secs with small size + window sent to lower left corner
so that changing bitmap is not shown. Maybe this trick is
already used by Donald.
m.