[198] in cryptography@c2.net mail archive
RC5 key setup time [was Re: 40-bit rc2/4]
daemon@ATHENA.MIT.EDU (Jim Gillogly)
Fri Feb 7 23:44:22 1997
Date: Fri, 7 Feb 1997 16:43:27 -0800
From: jimg@mentat.com (Jim Gillogly)
To: cryptography@c2.net
Lucky Green <shamrock@netcom.com> said:
> > Interesting. RSA touts their RC-5 cipher as not spending much time in key
> > scheduling. Can somebody confirm this?
"Donald E. Eastlake 3rd" <dee@cybercash.com> said:
> I believe key set up is pretty fast. Description is at
> http://theory.lcs.mit.edu/~rivest/rc5.ps
To quantify that a bit, my RC5 cracking hack spends around 80% of its
time doing key setup; the reference implementation is about 90%. I think
other implementations see similar numbers. By contrast, DES cracking
is much cheaper -- using Gray code order and bit twiddling, the setup
time is almost nil. That's not really a fair comparison, of course,
since the legitimate user would want to change to unrelated keys in
the course of business, which would cause DES to take more time...
but not 80%. Under normal use (e.g. you're not changing keys more than
once every few K) this shouldn't be a show-stopper for RC5.
Jim Gillogly
