1994-09-15 - Re: thoughts on RC4

Header Data

From: m5@vail.tivoli.com (Mike McNally)
To: Carl Ellison <cme@tis.com>
Message Hash: da43dc696bcd13a473a341bbb36a4afd420175b13e20edb489b2d87b2ab9cbff
Message ID: <9409151921.AA28584@vail.tivoli.com>
Reply To: <199409151735.KAA14334@comsec.com>
UTC Datetime: 1994-09-15 19:21:33 UTC
Raw Date: Thu, 15 Sep 94 12:21:33 PDT

Raw message

From: m5@vail.tivoli.com (Mike McNally)
Date: Thu, 15 Sep 94 12:21:33 PDT
To: Carl Ellison <cme@tis.com>
Subject: Re: thoughts on RC4
In-Reply-To: <199409151735.KAA14334@comsec.com>
Message-ID: <9409151921.AA28584@vail.tivoli.com>
MIME-Version: 1.0
Content-Type: text/plain



Carl Ellison writes:
 > >Can anyone see any reason why one could not change RC4 to
 > >being a word oriented stream cipher, call it "ERC4"?
 > 
 > (1) You could conceivably go to an array of 65K short values and retain the
 > permutation but I wouldn't try to go to an array of longs.  I don't have
 > 32GB of RAM on my workstation.

Perhaps some improvement on RISC architectures could be achieved by
using four arrays, 32 bits wide by 256 entries long.  The arrays would
mirror the single array in the original code, except that the
"interesting" byte would be at a different position in each.  Then,
the main encryption loop could be unwound so that you'd do four
operations to each word, one from each array to hit the four bytes.  
(You'd do the increment/swap between each one.)  This *might* be
worthwhile.

(On an Alpha, you'd have 8 arrays...)

| GOOD TIME FOR MOVIE - GOING ||| Mike McNally <m5@tivoli.com>       |
| TAKE TWA TO CAIRO.          ||| Tivoli Systems, Austin, TX:        |
|     (actual fortune cookie) ||| "Like A Little Bit of Semi-Heaven" |





Thread