Newsgroups: sci.crypt
Path: msuinfo!caen!sdd.hp.com!usc!rpi!ghost.dsi.unimi.it!vince
From: vince@ghost.dsi.unimi.it (David Vincenzetti)
Subject: Re: Fast MD program
References: <1992Feb5.095352.25776@ghost.dsi.unimi.it> <4900001@hpnmdla.sr.hp.com>
Organization: Computer Science Dep. - Milan University
Date: Thu, 6 Feb 1992 09:15:27 GMT
Message-ID: <1992Feb6.091527.14626@ghost.dsi.unimi.it>
Lines: 16

darrylo@hpnmdla.sr.hp.com (Darryl Okahata) writes:

>     How fast do you need it?  On a 68030-type machine, MD4 crunches at
>over 350,000 bytes/second (counting user+sys time only).  My 68040
>machine crunches at over 750,000 bytes/second.  Modern RISC machines
>will scream at an even faster amount.

I use a hp-s700 workstation with achieves 80000 dhrystone with light
system load. The program's bottleneck is, without any doubts, the checksum
calculating; I believe that CRC32 is totally insecure, so I would use
RSA's MD4 (MD[2, 5], Snefru-[2, 4, 8] are much slower).

What I would need is a fast MD4 implementation, faster than the one available
via ftp at rsa.com. Does such an implementation exist? Hints, anyone?

Ciao,	David
-- 
David Vincenzetti
Dept of CS, University of Milan: vince@ghost.dsi.unimi.it
Dept of CS, Polytechnic of Milan: vince@cdc835.cdc.polimi.it
Real $home: 8th street 39, 20090 Segrate, Milan, Italy
