[RC5] Did the calculations again...

Timothy Marsh Timothy.Marsh at usm.edu
Wed Jan 7 07:47:52 EST 2004


I remember this came up quite a while back and had some good answers.  I'll
copy them at the bottom of this email.

Tim

----- Original Message -----
From: "Richard Farmbrough" <rawkiw at cix.co.uk>
To: "'D.net Discussion'" <rc5 at lists.distributed.net>
Sent: Wednesday, January 07, 2004 4:37 AM
Subject: RE: [RC5] Did the calculations again...


> Some time ago calculations were done to the effect that it was more
> productive to solve hard problems by waiting until they could be completed
> in 18 months before starting.
> This was based on buying hardware of course. However a similar
> consideration may apply to other costs, such as power and soicla costs of
> running a 17 year project.
> The challenge for this type of organisation is finding problems of the
> right level of computational difiiculty, programming simplicity and
general
> interest.
------------------------------------------------
At 05:15 PM 10/1/2002 +0100:
While this may not apply to dnet since we use otherwise unused resources, I
am sure I remember reading a research paper that suggested any
computationally intensive program taking more than time T (18 months (?))
with a given budget X was better left until the technology advanced to the
point that X would buy enough processing power to complete in time T. Better
in this context means completing the calculation earlier.
-------------------------------------------------
Yep your correct this doesn't apply to dnet at all because the equipment
upgrades itself automatically at no cost to the project budget.
Every bit of computation done between now and the time when it would take
less than 18 months to complete makes the project end in 18months - X
percent completed. Since there is no cost other than the web hosting
(afaik) all projects using the dnet model should be started as soon as
technically possible.
Dnet Keyrates increase on a curve because people upgrade their processors.

And hurry up with rc5-72 already my bedroom is getting cold, and winter is
coming! :-)

oh... and removing stats pisses people off, so whoever does that.. don't..
your going to lose the people who get off on the competitive part of  your
projects. The Apollo project was a "who's got the biggest dick contest" and
it got the USA to the moon first. So whip it out boys, lets get our rc5-72
on and see who's the real "man" around here.

Adrian Neill
A proud Canadian
-------------------------------------------------
It depends what you're trying to prove. If you're just trying to get
the project done as soon as possible, the only advantage to waiting is
cost-effectiveness, which doesn't really apply to d.net-style projects. If
what you're trying to do is get the project done as *quickly* as possible -
minimum time from start to end, but not necessarily the earliest possible
end date - then it's to your benefit to wait until faster processors are
available.

OGR falls into the former category, DES into the latter. RC5 is
somewhere in between... we want to prove that it can be easily broken, which
means quickest possible crack, but we want to prove it fairly soon. Adding a
year to the cracking time is worthwhile if it can cut a year off the end
date... maybe. The exact exchange rate there is kind of fuzzy.

---
John Campbell
jcampbel at lynn.ci-n.com




More information about the rc5 mailing list