Re: Another FoRK Contribution to Culture

Date view Thread view Subject view Author view

From: Robert Harley (Robert.Harley@inria.fr)
Date: Thu Feb 01 2001 - 07:10:37 PST


Hi Greg,

You wrote:
>This sounds *extremely* interesting. Are you doing
>consulting or are officially setting up shop?

I have created a company called ArgoTech with a colleague, Henri
Gonzalez, who is a top-notch administrative and legal person. That
way I can stick to writing code. :) We are in discussions with some
other talented technical people and a manager to get them to work with
us or for us.

Our main objective is to develop efficient and reliable software with
a strong mathematical and algorithmic content, concentrating on
critical computational components especially those that are
intrinsically complex. I like to think that our niche is in stuff
that requires too much knowledge of math, crypto, signal-processing or
similar for most programmers to do it well, and too much hard-core
programming for mathematicians, cryptographers etc.

We can develop to specification and also have some projects of our own.

For instance if you want a crypto routine optimized for the ARM
processor in an upcoming portable gizmo, we're there. Or if you need
a very precise discrete cosine transform for compressing medical
images and it has to go fast using SSE2, that's us.

Our main "own project" right now is ECPC (Elliptic Curve Point
Counting) for generating secure curves for elliptic-curve crypto. The
latest version is at least twenty times faster than competing
packages. There is some danger with public-key systems that some
choices of parameters (such an RSA modulus or an elliptic curve) might
get smashed next week or next year by a new mathematical advance.
With ECPC, curve-generation is now fast enough that one can envisage
generating a secure curve in-house or one per workgroup, say, and
switching at will instead of everybody using the same curve all the time.

Just for fun: some researchers try to compute huge curves, too big for
crypto. The record a year ago was 1999 bits done in 65 days of CPU
time on a network of 400 MHz PCs by a Belgian guy, Frederik
Vercauteren. ECPC beats that in 34 minutes on a fast Alpha (maybe 3
hours on one of those PCs).

So that's what I'm up to these days.

Bye for now,
  Rob.
     .-. Robert.Harley@polytechnique.org .-.
    / \ .-. .-. / \
   / \ / \ .-. _ .-. / \ / \
  / \ / \ / \ / \ / \ / \ / \
 / \ / \ / `-' `-' \ / \ / \
            \ / `-' `-' \ /
             `-' http://www.xent.com/~harley/Top.html `-'


Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Fri Feb 02 2001 - 04:03:00 PST