Re: Hyper-Threading Considered Harmful

From: Stephen Sprunk (stephen_at_sprunk.org)
Date: 05/13/05


Date: Fri, 13 May 2005 09:38:36 -0500


"Francois Grieu" <fgrieu@francenet.fr> wrote in message
news:fgrieu-6FECBD.12045613052005@news5-e.proxad.net...
> In article <z5Tge.16306$J12.1436@newssvr14.news.prodigy.com>,
> Bryan Olson <fakeaddress@nowhere.org> wrote:
>
> > The world is going to hyper-threading and multi-core, and
> > probably then to hyper-threaded-unified-poly-core. Sure, I just
> > made up that term, but many people foresee the hardware and O.S.
> > cooperating to figure out how many and which threads to run at
> > once, to take best advantage of the various execution units.
> > We'll have a brave new world of side channel attacks.
>
> A nice alternative is to move cryptography to dedicated hardware;
> such as tamper-resistant modules or Smart Cards for Public Key
> crypto and low-bandwidth en/decryption, on-the-die crypto support
> for fast block ciphers.

Or AMD and Intel could just implement Via's AES instructions; they take
minimal die space, especially considering the behemoth processors coming
out these days. Constant-time execution means timing attacks aren't
possible, right?

S

-- 
Stephen Sprunk      "Those people who think they know everything
CCIE #3723         are a great annoyance to those of us who do."
K5SSS                                             --Isaac Asimov