[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Distributed computing


This may be more of a computer science question than a cryptography
one, but it seems relevant to the current discussion:

What is the current technology for running large compute-intensive
jobs accross a whole lotta computers?  Specifically, what's the best
way to run something like a big key-cracking job on a farm of
workstations?  Run separate jobs on each machine that check in to a
main server for new keys to crack?  Or is there a better way?

Also, on the note of producing a cheap parallel setup: a simple 90MHz
pentium motherboard is ~$150.  A cheap network card is under $50, and
I'll bet it's possible to build a custom power-supply that'll juice a
whole bunch of motherboards for under $200.  You don't need monitors
or keyboards for these machines -- just one of each to plug into the
various machines to troubleshoot. Thus, the incremental cost of adding
another pentium to a big cracking job is about $200 -- the startup
cost is the cost of the distributed power supply, monitor, keyboard,
etc. plus one server.  And it's even cheaper to build this sort of
setup if you go to someone who knows what s/he is doing, and can put a
whole lotta CPUs onto one motherboard...

The point is, that while one pentium can't do a whole lot, it's pretty
cheap to build a DIY pentium "farm."  And if a whole lot of people
have these farms, and are willing to run distributed cracking jobs on
them, computing power like that of the systems used in the RC40 crack
is very easy to come by.

Finally, there is a factor that hasn't been touched on much yet.
Credit cards are not the crux of the problem: in the years to come,
there will be more and more ways to manipulate money on the net.
Charles Schwab is already working on a way to trade stocks wholesale
over the net (or have they already started?), and I know of a couple
other companies that have plans in the works to do the same.  If it's
possible for a small group to break one of *these* keys in a
relatively short time (a month or two, I suppose), the money at stake
could be well worth the compute time involved.

- -sq

Version: 2.6