Hi Theodore!
Good job! See below for my comments.
On 07 Apr 2017 14:09:13 +0000
"theodorepringle1_at_gmail.com [fc-solve-discuss]"
<fc-solve-discuss_at_yahoogroups.com> wrote:
> I compiled Freecell Solver 4.8.0 with ./Tatzer -l ci7b -l extra_speed --nfc=3
It is not recommended to run "Tatzer" or cmake in general from the same
directory as the source. Next time run it in an empty build directory.
> and ran (in ranges of 100,000 deals to avoid running out of
> memory) ./freecell-solver-fork-solve --num-workers 8 --worker-step 1
> --total-iterations-limit 1000000 -l as (seemingly the fastest preset with
> three freecells). I found 8,531,478,635 winnable deals, leaving 58,455,957
> deals to be resolved:
> https://bitbucket.org/theodorepringle/freecell-pro-3fc-deals/downloads/
> (warning: the list of deals is over 600 MB).
I have two ideas to make the list smaller in disk size:
1. Encoding it in a sequence of binary files as 32-bit / 4 octets values
representing the lower 32-bits of each range of 4G numbers.
2. based on:
[shell]
< Unknown3.txt perl -lnE '$d = $_-$x;$m=$d if $d > $m; $x = $_;END{ say $m}'
2612
[/shell]
it seems the maximal delta between two consecutive deal numbers is 2,612 which
fits in 12 bits so we can encode the deltas.
I can try writing an encoder/decoder for either scheme if you're interested.
> This includes all "Unsolved" and
> "Intractable" deals. Later I'll try to include other results I have,
> including the first ten million deals (the first "real" intractable and only
> one less than one million is 138026), as well as various results of later
> deals (including all impossibles with four freecells), but for now this is a
> good start.
yes, quite - thanks for sharing.
Regards,
Shlomi
Received on Fri Apr 07 2017 - 21:35:46 IDT