Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!princeton!allegra!ulysses!faline!karn
From: karn@faline.UUCP
Newsgroups: sci.math,sci.physics,sci.crypt,sci.math.stat
Subject: Re: Do you use RANDOM NUMBERS?
Message-ID: <525@faline.bellcore.com>
Date: Thu, 9-Apr-87 00:33:44 EST
Article-I.D.: faline.525
Posted: Thu Apr  9 00:33:44 1987
Date-Received: Sat, 11-Apr-87 21:16:27 EST
References: <5712@reed.UUCP> <464@batcomputer.tn.cornell.edu> <444@faline.UUCP> <560@entropy.ms.washington.edu>
Organization: Bell Communications Research, Inc
Lines: 17
Keywords: Random numbers, leprechauns
Xref: utgpu sci.math:880 sci.physics:1079 sci.crypt:300 sci.math.stat:97
Summary: Yes, DES *does* produce good random numbers

> This is an understatement. DES algorithms are VERY slow. The Num.
> Recipies is a prime example. I expect that Phil has speeded it up
> to the point where it might be marginally usable but that
> does not mean that it is a good choice. I don't think that DES
> has been tested very well as a random number generator....

The Numerical Recipes DES is very slow mainly because it was written in
Fortran, a language with virtually no bit-twiddling facilities. My DES
(in the pipeline for mod.sources) can do 5400 encryptions/minute on the
PC/AT, and about 70,000/minute on the VAX-8650.  Not blazingly fast, but
not bad either considering the results ARE statistically random. As they
say in the book, the fact that DES has withstood all known attacks
except perhaps those of the NSA implies that DES is a very good random
number generator.  Statistical testing of ciphertext is one of the most
fundamental cryptoanalytic tools.

Phil