Re: NSA enhancing Linux security?

From: Bill Unruh (unruh_at_string.physics.ubc.ca)
Date: 02/28/04 

Date: Sat, 28 Feb 2004 18:14:35 +0000 (UTC)

Jem Berkes <jb@users.pc9.org> writes:

]> One of the ancient secrets of codebreaking is to use parallel
]> computing in order to break down the workload into manageable
]> segments, and then to analyze the ciphertext for correlating bit
]> sequences. This technique dates back all the way to Bletchley Park,
]> and is still valid today.

]You may be right about that, but parallel computing typically increases
]your processing power linearly. On the other hand, the discovery of a new

Only in the best cases. Usually it is more like square root. Few
problems are efficiently parallelizeable.

]factorization algorithm can bring your processing time down by an entire
]order. They employ top scientists and could have such algorithms.

In most fields algorithmic speedups have outpaced hardware speedups.

Relevant Pages

  • beta version of Victor Shoups book, "A Computational Introduction to Number Theory and Algebra&
    ... Computing with Large Integers ... The Basic Euclidean Algorithm ... Factoring and Computing Euler's phi-Function are Equivalent ... The Existence of Finite Fields ...
    (sci.crypt)
  • Re: Why does Cantor a target for cranks?
    ... and wildberger doesn't give this the recognition it deserves ... that if we have a computing process that generates a stream of ... then it's meaningful for such a specific process (algorithm) ... or conflating various groups of order 6 into the ...
    (sci.math)
  • Re: Correctness proving (Was: Clear and Unambiguous SOFTWARE requirements/specifications possible?)
    ... We come up with reasonable schemes that find "good" results (by some ... The bulk of OR techniques -- like linear programming, dynamic programming, Out Of Kilter, etc. -- all provide a deterministic solution for problems like Traveling Salesman that will always be the absolute optimum. ... Solving np-Complete problems requires both an algorithm and a representation of the problem that is appropriate for the algorithm. ... The issue here is proving correctness of programs that are well-formed within the constraints of the computing space. ...
    (comp.software.testing)
  • Re: Penroses Computing Pi Description?
    ... > <snip penrose> ... > digit as output. ... > "computing Pi" on its own, ... is the algorithm the computation of one particular digit or is ...
    (sci.logic)
  • Re: qubits
    ... One cycle of the algorithm would collapse ... Most np-Complete problems have been ... Thus any Qubit ... Though Qubits has obvious potential for increasing computing power at ...
    (comp.object)