home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.crypt
- Path: sparky!uunet!cs.utexas.edu!swrinde!network.ucsd.edu!qualcom.qualcomm.com!servo.qualcomm.com!karn
- From: karn@servo.qualcomm.com (Phil Karn)
- Subject: Re: DES (Was: Re: 800MHz
- Message-ID: <1993Jan10.101304.20364@qualcomm.com>
- Sender: news@qualcomm.com
- Nntp-Posting-Host: servo.qualcomm.com
- Organization: Qualcomm, Inc
- References: <1993Jan9.163011.23230@csi.uottawa.ca> <1993Jan10.050811.3740@ils.nwu.edu> <1993Jan10.060013.4644@csi.uottawa.ca>
- Date: Sun, 10 Jan 1993 10:13:04 GMT
- Lines: 28
-
- In article <1993Jan10.060013.4644@csi.uottawa.ca> cbbrowne@csi.uottawa.ca (Christopher Browne) writes:
- >It doesn't surprise me AT ALL that the NSA would have "lotsa
- >supercomputers." That kind of goes with the territory. It also
- >doesn't surprise me that they can consider breaking DES. With the
- >kind of computing horsepower they've got, brute force can't be THAT
- >hard.
-
- As the old joke goes, a Cray can execute any infinite loop in under 6
- seconds. :-)
-
- Crays aren't quite as fast as you think. And they're probably not
- especially cost-effective for direct, brute-force searches of the DES
- keyspace, unless of course somebody has discovered some significant
- shortcut. An array of special purpose processors would be much faster
- and far more cost-effective.
-
- NSA undoubtedly uses its Crays mainly for signal processing. Remember
- that before the cryptanalysts can get their ciphertext bits, you have
- to extract them somehow from the terabits of raw broadband digital
- tape recordings made by the intercept sites. For example, if
- frequency hopped spread spectrum is used, one way to attack it is to
- demodulate every channel in the entire hopping range, an obviously
- CPU-intensive job. With their fast floating point and pipelined
- vector processing capabilities, Crays are ideally suited for this kind
- of number crunching.
-
- Phil
-
-