Bernstein's NFS machine

[David Wagner]

Very interesting.  Thanks for the analysis.

Bernstein's analysis is based on space*time as your cost metric.
What happens if we assume that space comes for free, and we use simply
time as our cost metric?  Do his techniques lead to an improvement in
this case?

It looks to me like there is no improvement in the first phase, but
there is some improvement in the second phase (reduction in time from
B^2 to B^1.5).  Then, of course, we need to re-consider the parameters
to balance the work, and it seems we would want to choose E = B^1.25,
subject to the constraint that E is large enough to produce at least
B relations.  What does this solve for, in terms of L?  How much do his
matrix solving techniques speed up the total computation, in the case
where we count only the running time of the adversary?

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at wasabisystems.com