|
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
On Sat, 15 Apr 2000, Benjamin Folk Jr. wrote:
> Sorry for the off topic post...
>
> Has anyone thought about starting a Seti@Home team for PLUG. I have been
> running it for about a year now and it is very interesting. They have several
> Linux clients and it gives "idle" machine something to do.
>
> http://setiathome.ssl.berkeley.edu/
Incase you're going to skip large chunks of my post, the point of it is:
* do not run seti@home (because they are evil)
* preferably run GIMPS, & if not GIMPS, run distributed.net
(GIMPS = the Great Internet Mersenne Primes Search /
http://www.mersenne.org/prime.htm)
Now for my reasoning....
BTW, I'm sure I posted something to this list similar to this before, but
I can't find it.
I'm all for distributed processing. It's awesome. And seti@home's better
than nothing. But as far as distributed processing projects go, seti is
evil.
* I believe they are completely closed source
* when distributed.net offered to help, they declined in fear that their
sponsors would realize there was something wrong if someone was willing
to help them for free
* when they ran out of info for clients to process, they started looping
it, making clients process redundant info, jut to not loose people's
interest, and sponsorships
In other words, it's all about the money.
Back in the day, there were, I believe, 3 groups competing to crack
rc5-56. The Cyberian Project, the Infinite Monkeys project (IM) (I love
that name), and distributed.net. Distributed.net blew everybody else
away, (merely because they happened to attract more people to run their
client), and the others faded out of existance. Somewhere I have the
source code to both the IM & Cyberian clients, and got permission to do
what I wished with them.
Distributed.net has always been full of unfulfilled promises. Cracking
strong encryption was an easy core to impliment, and fun. But we did
that. A couple times. It got old, and there were other more interesting
things to do. Like OGR (Optimul Golomb Rulers, used in actual science
stuff, like x-ray christolography (no, I don't actually know what it is,
but it sounds neat)). They promised an OGR core forever. They finally
released it in the past couple months, and it was quite broken, and they
had to discard all work done & start over. But that's a forgivable error.
And I believe it's actually working now.
Mor irratating was the V3 vaporware. It was supposed to be this great
distributed processing protocol, completely opensourced, and easy to
impliment & distributed processing cores for any processing intensive task
you wished.
This was immensly frustrating to me. I wanted it so badly. Partially
because before I found out about it, I was contemplating creating
something like it myself. And they wouldn't let me help.
The guy behind V3 split off of distributed.net not long ago, due to
differences in direction. Distributed.net realized they really didn't
care. The other guy did, but still hadn't a clue about open development.
The only assistance he would accept was with documenting his idea
rediculously extensivly (which only he actually comprehended), and only
from people he could communicate with face to face (people geographically
nearby).
Distributed.net was never particularly into the opensource concept.
Terrified of people using the source to "cheat". The whole cheating
problem arrised out of a bad source of motivation. The way they record &
display stats causes people to compete for the fastest block crunching
rate. The more blocks a person has reported as checked, the more highly
they are reguarded in the community. This is bad. This has nothing to do
with the desired end result. The desired end result being finding the
single key that will actually unlock the encryption.
I have, at times, been so irratated with distributed.net that I was
considering creating a completely opensourced competing program, without
need for a central keyserver, and no concern for cheating. & based off of
the cyberian & IM cores that I'd collected.
The idea was for every client to process blocks in totally random
sequence, with no concern for redundant checking. I got a number of
mathimatically inclined people to convince me that the amount of time to
find the right key with this method would be, on average, exactly twice
what it would take if there was no redundancy.
The part that made this even more neat was the money. There's a $10k
prize for cracking rc5 stuff. If you win with distributed.net, $8k of
that is devided up between distributed.net &, I believe, eff.org. I
believe eff.org gets the great majority of it. If you're on a team, the
remaining $2k is split between you & your team (you get $1k). If you're
not on a team, you get $2k.
If you were to run this opensourced client, with half the chance of
wining, you'd get at least 5 times the amount of money. All the math guys
thought that was pretty good odds.
If you still want to run a distributed.net client, I suggest one of the
non crypto cores (like OGR, if I'm correct in believing it's currently
usable).
I believe the actual processing cores are available for both
distributed.net & GIMPS clients, but that the networking bits are not
available, due to paranoia about cheating.
Now for GIMPS (http://www.mersenne.org/prime.htm). The Great Internet
Mersenne Primes Search has been around for a long time. The goal of the
project is to find mersenne primes -- primes of the form 2^(n-1), where n
is a positive integer. The form doesn't matter, the fact that
historically, almost allways, the largest known prime has been a mersenne
prime. And since 1996, the largest known prime has been one found by
GIMPS. GIMPS has found the 4 largest known primes (1 per year). In the
beginning, they transferred blocks (chunks of work to do) manually over
email. This has since been fixed, and the client is fully automated, as
is distributed.net's.
The current largest known (mersenne) prime is 2^6972593-1 (from memory, &
confirmed).
Primes are useful. There are 38 known mersenne primes. If you happen to
find one, your name will actually go in history books.
Now that you're a bit more informed, go do whatever you want.
And if you feel like working on that opensourced rc5 client with me, let
me know :)
__________________________________________________________________
PGP fingerprint = 03 5B 9B A0 16 33 91 2F A5 77 BC EE 43 71 98 D4
darxus@op.net / http://www.op.net/~darxus
There is no fine line between genius and insanity.
______________________________________________________________________
Philadelphia Linux Users Group - http://plug.nothinbut.net
Announcements - http://lists.nothinbut.net/mail/listinfo/plug-announce
General Discussion - http://lists.nothinbut.net/mail/listinfo/plug
|
|