[RC5] OGRs meet D.Net, was "sexy" projects

Shannon Jones scj at pobox.com
Mon Mar 2 10:45:15 EST 1998


>Is OGR a project that gets helped by knowing the results of
>other answers found by other computers?  I would think that
>would give d.net's clients a problem right now.
>
>Nathan Barber

The short answer is that OGR is helped slightly by knowing results on other
computers, but it won't be a problem.  Implementing an OGR engine will
probably be easier than the DES engine because the C source is available.

Now for the long answer:

The current search is for a 21 mark ruler with a length of 332 or shorter
(currently, the best ruler is 333 in length).  Searching for shorter rulers
takes slightly less time.  If a client finds a ruler 331 in length, then the
other clients would benefit from knowing this, but it won't invalidate their
search.  They could still find one shorter than 331, it will just take them
slightly longer.

When a client finishes its stub, it will return the results and be sent a
new one.  The new stub will search only for rulers 330 or shorter.  (NOTE:
If we find one 332 in length or shorter, we may still search for other
rulers of the same length.  Knowing all rulers shorter than 333 might be
more valuable than saving a few days or weeks off the search time.  If we do
it this way, clients would not benefit AT ALL from the results of other
computers).

I don't know how much faster it is to search a shorter stub.  I would say it
is no more than 5% faster for every unit you trim off the "best
known"length.

So it does matter slightly, but it will be fixed when you grab your
nextstub.

<<End of long answer>>

On another note, OGR-21 will not take very long.  Clients could be released
quickly (fast cores are in C), and the power of d.net could squash it in
less than a week.

We aren't guaranteed a better ruler.  We are confident that a better OGR-21
will be found, but there are no promises.  We might just verify that the
known ruler of length 333 is the best possible.  The effort wouldn't be
wasted though.  OGR-22 waits for us, and will be much harder (no one knows
exactly how hard, but probably 5-20 times harder than OGR-21).  And then,
OGR-23 . ..

I hope d.net and the golomb people get together.  It would be great.  A wide
choice of projects will make d.net stronger in the long run.  The "sexy"
projects will get the most time.  Very democratic.

--
Shannon Jones
scj at pobox.com

P.S. The source code to the GVANT and GARSP clients is available from the
OGR-21 home page at http://members.aol.com/golomb20/index.html
--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list