Re: [Cfrg] Request For Opinions

On Tue, 13 May 2003, Alex Alten wrote:

> The facts about PK are:
> 1. Key generation is dog slow
> 2. Key setup is damn slow.
> 3. Encryption is like watching grass grow or paint to dry.

4. It's often the only tool that gets the job done.

If we cannot solve a problem satisfactorily using our usual symmetric
primitives like PRPs, PRFs, hash functions, MACs etc. we will have to
resort to more expensive primitives.  They're the cheapest ones that get
the job done.

Some operations are actually quite efficient, for example RSA encryption
using a low exponent (3 or 65537).

I wrote a signature verification routine in portable C about 4-5 years
back, to verify Rabin-based signatures. The large integer routines were
home grown and far from optimized. Verification took about 1 ms on a 200
Mhz pentium clone processor (1024 bit keys). Our hardware is at least 10
times faster now, so we're probably talking about 100us verification
times.

Signature generation (2 512-bit exponentiations and an application of
Garner's algorithm) took 500 ms using a good bignum library, so on current
hardware we're probably getting close to 50 ms. No paint dries this fast.

Some new processors have instruction sets optimized to perform big number
arithmetic efficiently, so this will soon even be less of an issue. And if
you're really in a hurry you can buy accellerators that perform modular
exponentiation in hardware.

Ge'

-- 
Gé Weijers                mailto:ge.weijers@sun.com
_______________________________________________
Cfrg mailing list
Cfrg@ietf.org
https://www1.ietf.org/mailman/listinfo/cfrg