RE: Recruiting More People; .EDU sites (rambles)

Adam Haberlach (HaberlaA@testlab.orst.edu)
Wed, 2 Apr 1997 08:47:31 -0800


Carleton = Carleton Jillson <careleton@wpi.edu>

> Carleton> First off, let's look at the top ten contributors: #4 is
> 207.196, which I
> Carleton>have no idea what it is, except judging by the amount it puts
> forth with
> Carleton>just one computer, must be 1 hell of a computer..., #10 is
> ibm.net, which
>
> We have been trying to figure that out as well--can we tell if they
> have more then one proc or which client they're running? We're
> thinking of trying a port to the school's Maiko (sp) box. 16 Sparc
> processors, 32 fujitsu vector coprocs. It should scream once we get
> parallel code, or even code to use the vector procs.
>
> Carleton>kinda speaks for itself. All the remaining are .EDU's....
>
> And they say that we students just sit around and _waste_ all of their
> time... :)
>
> Carleton>Colleges/universities have immense computing power, and
> unlike businesses,
> Carleton>much of it is untapped except to play games. (Actually,
> that's not entirely
> Carleton>true... from experience, many businesses seem to have a lot
> of computer
> Carleton>power that is only used for things like Quake, but I digress
> :) At WPI, my
> Carleton>school, and current 2nd placer (GO TECH!), 24 personal
> computers are
> Carleton>currently spewing out 748064*2^20 keys per second. This is
> pretty
> Carleton>impressive considering that these are just individual owned
> computers; the
> Carleton>head honchos of our computer labs seem to have decreed that
> deschall shall
> Carleton>not be run on school computers... If some sort of
> inter-school rivalry could
>
> Bummer--is this for all of the computer labs, or just the EE ones?
>
> Here's the story behind our effort (first place for weeks [GO
> BEAVERS!]).
>
> I found a reference to this effort on one of the lists I'm on
> (for a 'competing' effort that hasn't written any code yet)., and
> downloaded the code and ran it on my laptop at home. We've been
> looking into doing this kind of thing for quite a while at the testlab
> that I work for--we usually have about 60 test clients sitting on
> racks doing nothing. I brought the software in, and set it up on my
> computer, mentioning it to a co-worker, who put it on his computer,
> and it kind of leapfrogged from there. Eventually, we had it on
> pretty much all of the computers down here. I think this is kind of
> the effect we should be expecting at .COM sites--word of mouth capped
> either by running out of resources or a mandate from above demanding
> that ununsed CPU cycles remain wasted. [1]
>
> This testlab is run by the Business department (note: I am an
> engineering major, I just work here). We realized that at the other
> end of the hall were 160 machines that were going to go to running
> login screens over spring break, and asked the head honcho here about
> running DESCchall on those machines. [2] He agreed, and within 30
> minutes of the lab being closed, we had those machines running. (Side
> Note: We have enough data by now that I'm thinking of writing a
> script to generate graphs of our search rate with regard to time, to
> see if we can correlate with things like spring break and holidays).
> He is so impressed by our being first place, and the odds that we're
> going to win that he is now prosetylizing around campus to everyone
> else in Wintel platforms, trying to drum up more support.
>
> We are currently waiting for word from some of the Engineering
> profs for a blessing to set the HP-PA client up on around 200 HP 7xx
> machines run by the department. Our local cryptography god is
> interested in this project, and I know there was already a senior
> project that was working on writing code to reliably do process
> control over huge clusters of machines.[3]
>
> -- Non-linear part of writing follows :)
>
> [1] This seems to be the problem for .COM sites--management might not
> like some of the ideas. I think that we need to either update the web
> site or do something to make the point that this client is NOT a
> security problem (although that may require source release, which
> Rocke does not like). We also need to keep driving in the point that
> unless you're doing something with your computer, it could be checking
> millions of keys in the time between keystrokes. Unfortunately, .COM
> sites tend to have quite a bit of inertia, because there's always some
> guy in charge who doesn't quite understand and doesn't want to go out
> on a limb...
>
> [2] We buttered him up with the prestige that being part of the first
> successful project to crack DES would bring, especially if we were the
> top search contributers when it happened. He goes for that sort of
> thing. This seems to be a fairly good way to get adiminstrators
> involved--they want the publicity angle. He would like to see some
> competition out there as well...I don't know if the full impact of "7
> years to completion has sunk in yet." I can see by looking at today's
> stats that a lot more .EDU sites seem to be expanding the number of
> machines. This is a good thing--word of mouth works. Has anyone been
> keeping track of hits to the website on a day-to-day basis? We
> should, to guage how much difference the press release makes.
>
> [3] I keep forgetting to contact this group, I will soon.
>
> Carleton>be setup, I think a lot more people would be willing to take
> part. This
> Carleton>would be especially true at places like MIT, RPI, RIT, WPI,
> and other techie
>
> I don't know how much it matters that it be a techie school. I
> DO consider OSU to be a techie school, but most of our effort has come
> off of the business department's computers. I think that having LOTS
> of computers that are well-administrated helps. It also helps that we
> have such a simple client. It will be incredibly easy to set the
> client up on the ..ENGR.ORST.EDU clusters, because they all share home
> directories, etc... If I were living in one of our dorms, I would
> probably have the client running on the computers of everyone on my
> floor, for example.
>
> Carleton>schools, where there are enough dedicated computer scientists
> who would care
> Carleton>about such things, and coincidentally enough, tend to have
> powerful
> Carleton>computers to crack considerable numbers of codes...
>
> I keep trying to make the point that it's much better to go
> around finding lots of little computers then to try and find the
> biggest. It would be NICE if we could get time on the CM5 over in the
> Oceanography department, but it will be 100x easier to get the client
> going on the hundreds of machines that are sitting on the desktops of
> grad students. Big machines are merely impressive to look at. In
> this case, we need lots of little machines.
>
> ---
> Adam Haberlach haberlaa@ucs.orst.edu
> http://www.engr.orst.edu/~haberlad
> Crack DES now! http://www.frii.com/~rcv/deschall.htm
>