Re: 64 bit DES (forwarded message from Matthew Kwan)

Justin Dolske (dolske@cis.ohio-state.edu)
Thu, 24 Apr 1997 15:08:46 -0400 (EDT)


On Thu, 24 Apr 1997, Nelson Minar wrote:

> ------- start of forwarded message (RFC 934 encapsulation) -------
> From: Matthew Kwan <mkwan@cs.mu.oz.au>
> Subject: Bitslice DES - I'm releasing the source
>
> A 32-bit version will be available soon, but, as I keep telling people,
> unless it runs on a RISC machine with lots of registers, it will be slower
> than tradiational DES implementations. That means it is not suitable for
> Intel machines. Sorry.

I've already tried the 32 bit version on the HP's here, and the DESCHALL
clients are 48% faster. I had to leave before a 64 bit version was
compiled on some other systems by Guy Albertelli... Guy, have any results
yet?

Out of curiousity, what kind of machines are those of you near the top in
the rankings running? Specifically, is anyone running a signifigant number
of 64 bit (SGI, Alpha, whatever) machines? I guessing not, so Kwan's
Bitslice code probably isn't of much help to us.

Justin Dolske <URL:http://www.cis.ohio-state.edu/~dolske/>
(dolske@cis.ohio-state.edu)
Graduate Fellow / Research Associate at The Ohio State University, CIS Dept.
-=-=-=-=-=-=-=-=-=-=-=-=-=- Random Sig-o-Matic (tm) -=-=-=-=-=-=-=-=-=-=-=-=-
Linux -- the choice of the GNU generation.