Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site gargoyle.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!ihnp4!gargoyle!loyola
From: loyola@gargoyle.UUCP (Loyola Math Dept)
Newsgroups: net.crypt
Subject: Re: Is RSA useful?
Message-ID: <532@gargoyle.UUCP>
Date: Sat, 27-Jul-85 13:18:41 EDT
Article-I.D.: gargoyle.532
Posted: Sat Jul 27 13:18:41 1985
Date-Received: Mon, 29-Jul-85 06:22:23 EDT
References: <>
Reply-To: loyola@gargoyle.UUCP ( Math Dept)
Organization: U. Chicago, Astronomy & Astrophysics
Lines: 30
Summary: RSA benchmark on IBM PC

In response to the recently posted article which questioned the usefulness
of the RSA Public-Key Cryptosystem because of its inherent slowness, I offer
the following benchmarks:

				             modulus (decimal digits)
		function	        116              77             39

key generation (average time		6 mins.		4 mins.		1 min.
   required to find a pair of primes)
encryption (per block)		      2.3 secs.       .88 secs.       .28 secs
decryption (per block)		      22  secs.         9 secs.       4.5 secs

Theses timings were obtained on an IBM PC (4.77 MHz 8088) with all arithmetic
routines written in assembly language.  Each function runs approximately 3-4
times faster on an IBM AT.  (Encryption uses 211 as the exponent except in 
the unlikely event that 211 divides the Euler totient, in which case
successively larger primes are considered.)

While the smallest version (39 digits) is unacceptable for security reasons,
the middle version could only be broken by the folks at Sandia (or the NSA,
I presume).  The largest version seems quite secure at the present time.

While admittedly not spectacular, these times make it quite feasible to use
RSA to apply digital signatures to short messages, such as electronic funds
transfer orders, or to encrypt DES keys in a secure key distribution scheme.

Comments?  (I'm carefully avoiding the patent issue!)

					Michael J. Markowitz
					ihnp4!gargoyle!cantor!abel!mjm