Abstracts of invited talks SPEED 2007
- Daniel J. Bernstein
(University of Illinois at Chicago, USA)
How fast is cryptography?
Abstract:
Users of public-key cryptography
have a choice of public-key cryptosystems,
including RSA, DSA, ECDSA, and many more.
Exactly how fast are these systems?
How do the speeds vary among Pentium, PowerPC, etc.?
How much network bandwidth do the systems consume?
The eBATS (ECRYPT Benchmarking of Asymmetric Systems) project,
joint work with Tanja Lange, aims to answer these questions.
I'll present the answers obtained so far,
describe the eBATS benchmarking toolkit,
and point out future directions in public-key benchmarking.
I'll also discuss extensions of the toolkit beyond public-key cryptography,
for example measuring the speed of secret-key authenticated encryption
and measuring the speed of new hash functions.
- Torbjörn Granlund (SWOX,
Sweden)
GMP small operands optimization
or
Is arithmetic
assembler automatically optimal?
Abstract:
GMP as a general purpose low-level arithmetic library has not been
specifically tailored for cryptographic applications. Yet
it performs well compared to special-purpose crypto libraries. The
techniques and algorithms used for crypto-relevant low-precision
arithmetic will be explained, and possible future improvments will be
examined.
- Dag Arne Osvik
(EPFL, Switzerland)
Cell SPEED
Abstract:
The Cell processor used in the PlayStation3 provides the potential for
computations at high speed and low cost. In this talk I will discuss
how to design your programs to benefit from this new architecture.
- Daniel Page
(University of Bristol, UK)
Computer Aided Cryptographic Engineering
Abstract:
In developing cryptographic software, most of us are implicitly dependent on
tools that translate our programs into an executable form; experience shows
that the translation process can heavily influence features such as efficiency
and security. Maximising these features while simultaneously minimising
programmer effort is clearly desirable. As a result, an interesting research
question is how best to design domain-specific languages and development tools
that support the description and implementation of cryptographic software. The
aim of this talk will be to highlight existing development tools (which are
perhaps unknown but useful) and to overview new research at the University of
Bristol into a cryptographic-aware language and compiler.
- Matt Robshaw (France Telecom R+D, France)
The eSTREAM Project
Abstract:
In this presentation we survey the origins, ongoing results, and the goals of
the ECRYPT eSTREAM project.