- Opening speaker
- Jose Meseguer (UIUC)
- Invited speakers
- Arvind (MIT)
- Manfred Broy (TUM)
- Giovanni De Micheli (Stanford)
- Ken McMillan (Cadence)
- Aims and scope
- Call for papers
- High-level models and languages
- Analysis and verification
- Methodological issues
- Reuse and component-based design
- Distribution: fault-tolerance, scheduling
- Non-functional requirements
-
Industrial case studies
- Conformance checking
- Hierarchical verification
- Abstraction vs. optimization
- Incremental verification
- Post-production patchability
- Final program
The MEMOCODE'2003 proceedings are now available online from the IEEE.
Selected papers from the conference will be published in a
special issue of the ACM TECS (see call for papers) and in a special volume by
Kluwer Academic Publishers on a later date.
-
Sponsorship
IEEE Circuits and System Society
ACM Special Interest Group on Digital Automation
IEEE Computer Society, Digital Automation Technical Committee
Organized
at the Relais du
Mont Saint-Michel
in cooperation with INRIA
and IRISA.
High-level design and modeling of hardware and embedded
hardware-software systems has gained prominence in the face of rising
technological complexities and performances, as well as shortened time
to market demands for complex electronic equipments. Numerous
programming languages, tools and frameworks have been proposed in the
past to design, simulate and validate heterogeneous systems within an
abstract and rigorously defined mathematical model.
Recently, attention has shifted to modeling frameworks based on
variants of general purpose programming languages, in response to the
growing industry demand for use of higher levels of abstraction in the
system design process. Meanwhile, the installed base of existing IP
adds further requirements for the adaptation of existing IPs with new
services within complex integrated architectures, calling for
appropriate methodological approaches.
Whereas abstract frameworks are ways to unambiguously model the
essense of hardware software systems, help understand the design,
implement formal correctness proofs, predict performances and other
metrics; general-purpose languages facilitate programming, reuse and
gain from the popularity of C, C++ like languages. Still, important
gaps need to be filled and bridges to be built between the theory of
modelizing and the practice of programming. Languages shall benefit
the rigorousness of models and models the experience of programming
practice.
This calls for finding a convergence between both approaches. A focus
on formal methods (programming and concurrency models, analysis and
verification techniques) for hardware/software codesign is necessary,
because language in which system designers work are general-purpose
ones, because the only way provably correct systems can be constructed
are by technology transfer of research in formal methods.
Donators : INRIA, University of Rennes 1, CNRS.
| |