1996-04-17 - Re: rc4 speeds

Header Data

From: Bill Stewart <stewarts@ix.netcom.com>
To: jr@Samba.cnb.uam.es
Message Hash: cdb1ee8b08910ae89dfbb63c91aeff67779ffffd7229382750c73d076b87bb53
Message ID: <199604170744.AAA28455@toad.com>
Reply To: N/A
UTC Datetime: 1996-04-17 11:13:43 UTC
Raw Date: Wed, 17 Apr 1996 19:13:43 +0800

Raw message

From: Bill Stewart <stewarts@ix.netcom.com>
Date: Wed, 17 Apr 1996 19:13:43 +0800
To: jr@Samba.cnb.uam.es
Subject: Re: rc4 speeds
Message-ID: <199604170744.AAA28455@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


>>That 164 mb/sec figure for RC4's speed better be a typo, because I
>>can't even come close to that.  I wrote an Assembler subroutine that 
>>encrypts a 32,768 byte block, and called it 65536 times on a 
>>486/DX2-66, without doing any disk reads or writes.

First of all, is that megabytes, or megabits?  I've forgotten how
many instructions it takes to do RC4, but RC5 takes 8-10 per round per
pair of words encrypted, so it should do about 1/2 bit per instruction for
16-round.
Some processors can do more than one instruction per clock cycle,
though loads and stores are usually a bit slower.  But you only need to
load two words every 128-160 clocks, which is easy on a pipelined machine.
So maybe it's a typo, but it should be far faster than 164 kB/s.

#					Thanks;  Bill
# Bill Stewart, stewarts@ix.netcom.com, +1-415-442-2215






Thread