Re: Crypto hardware (was: Using sound cards to accelerate RSA?)

At 1:49 AM 9/29/95, Douglas Barnes wrote:
>[Tim May writes about why crypto h/w acceleration using DSPs,
>and DSPs in general, are likely doomed niche markets.]
>
>I pretty much agree with Tim, except it's important to realize
>that a for a _server_ that is doing a lot of RSA operations, the
>difference between a 3.2 second encryption and a 1.9 second
>encryption is significant. Peak transaction volume for any
>public key-based payment system is going to be a factor of how many
>RSA ops you can do per second.

I don't disagree with Doug about this. But I don't think there are many
"server" systems running a lot of RSA at this point...for most of us, the
amount of RSA (or PGP, IDEA, DES, etc.) computation is a tiny fraction of
the total computons consumed running screen savers.

I'm just not convinced I'll soon invest in a company offering RSA acceleration.

(I recall seeing articles about specialized modular exponentiation hardware
in 1988, and Cylink was offering several such chips. I've yet to see any
commercial boards, for reasonable prices. And I'm willing to be that no
more than 3 members of our list would buy such a board, even if the hooks
were in place to let PGP, RSAREF, etc. use it. Just a hunch.)

--Tim May

---------:---------:---------:---------:---------:---------:---------:----
Timothy C. May              | Crypto Anarchy: encryption, digital money,
tcmay@got.net  408-728-0152 | anonymous networks, digital pseudonyms, zero
Corralitos, CA              | knowledge, reputations, information markets,
Higher Power: 2^756839      | black markets, collapse of governments.
"National borders are just speed bumps on the information superhighway."

Timothy C. May writes:
> At 1:49 AM 9/29/95, Douglas Barnes wrote:
> >[Tim May writes about why crypto h/w acceleration using DSPs,
> >and DSPs in general, are likely doomed niche markets.]
> >
> >I pretty much agree with Tim, except it's important to realize
> >that a for a _server_ that is doing a lot of RSA operations, the
> >difference between a 3.2 second encryption and a 1.9 second
> >encryption is significant.
> 
> I don't disagree with Doug about this. But I don't think there are many
> "server" systems running a lot of RSA at this point.
[...]
> for most of us, the
> amount of RSA (or PGP, IDEA, DES, etc.) computation is a tiny fraction of
> the total computons consumed running screen savers.

And if problems like this don't get solved, how do you expect digital
online banking to be done? Psychic quantum transfers between the
machines? What do you think a bank in the future is, if not a server
that has to do lots and lots of RSA or D-H or what have you?

Sorry for being nasty, Tim. Its just that some of us live in the real
world, have real clients, and actually worry about this as a
problem. This *is* a legitimate problem. Consider what the load on a
web site using D-H key exchange for every connection gets like when
you have millions of people hitting it every day.

> (I recall seeing articles about specialized modular exponentiation hardware
> in 1988, and Cylink was offering several such chips. I've yet to see any
> commercial boards, for reasonable prices. And I'm willing to be that no
> more than 3 members of our list would buy such a board, even if the hooks
> were in place to let PGP, RSAREF, etc. use it. Just a hunch.)

You obviously haven't heard of Fortezza cards. Yup, they are key
escrowed -- but they do in fact do public key operations on
board. There are a lot of them floating around.

The reason the market for this is weird is the same reason Sun took
the DES chips off its motherboards years ago -- you can't conduct
modern business with the fucked up export regime we are dealing with.

.pm

On Thu, 28 Sep 1995, Perry E. Metzger wrote:

> world, have real clients, and actually worry about this as a
> problem. This *is* a legitimate problem. Consider what the load on a
> web site using D-H key exchange for every connection gets like when
> you have millions of people hitting it every day.

This is the problem I was concerned about (actually RSA rather than D-H). 
In HTTP-NG, in addition to supporting PK for key exchanges and 
authentications, there is now support that allows most values used in the 
protocol to be signed. Now that non repudiability is becoming legally 
significant, there are all sort of things that either party might want to 
have signed, for example negotiation options (e.g. wont-log-transactions) 
and meta-information (e.g. kidcode: NC-17,barney-boffing).

More clients and more signings means that conventional chips arent't 
going to be economical for this.

[stuff on hardware]
> 
> The reason the market for this is weird is the same reason Sun took
> the DES chips off its motherboards years ago -- you can't conduct
> modern business with the fucked up export regime we are dealing with.

That's another question. A DSP chip can also be used for crypto - yet 
sound cards and nexts aren't ITARed, and aren't really considered 
dual-use. A Modular exponentiator isn't a crypto device  (hey - 
it's a bignum accelerator for Mathematica). Now, if I had a pipelined 
WSI chip capable of delivering one result per cycle, I could think of 
some useful applications, but ...

Simon