MSWord - Computer Science and Engineering

Panel: Web-Based Frameworks to Enable CAD R&D
Olivier Coudert, Monterey Design Systems, Inc., CA
Sharad Malik, Electrical Engineering Department, Princeton University.
Andrew B. Kahng, University of California, Los Angeles, CA
This panel focuses on how new Web-based frameworks can
enable academic research and development in VLSI CAD.
Open-source initiatives, portals for research communities or for
the EDA community at large, algorithm benchmarking support,
etc. are a few of the possibilities. Each of the three panelists will
describe the current status of efforts toward building new Webbased infrastructure for CAD research.
The approaches and
“mission statements” illustrate the literally unbounded potential
for such frameworks to transform how CAD R&D is performed.
Panelist Statements
The MARCO GSRC Bookshelf for Fundamental CAD-IP
Igor L. Markov, University of California at Los Angeles
Our analysis of common practices in the VLSI CAD (in
particular, Physical Design) community suggests that huge
intellectual losses are incurred by “superficial replication” and
“repeated research” when recent results and conclusions are
overlooked. At the same time, academic researchers are at risk
when prevailing technologies outgrow, and can no longer rely on,
textbook formulations.
A related symptom is that the values perpetuated by leading
conferences and journals in the field have considerably biased the
academic research community towards “novelty per se”, and
hence away from /substantiating the need for novelty/via
(empirical) comparisons to existing approaches.
Finally, the often-lamented secrecy over benchmarks and other
intellectual property adds further showstoppers to fundamental
research. But these are matched by academic protectionism,
incompatible and often insufficient software skills, and dismal
investments into reuse (with ensuing “data death” and “code rot”).
The way to eliminate the above obstacles to CAD R&D is not by
assigning “plaintiffs” and “defendants”: all parties are guilty and
need to be sentenced to community service. We propose a series
of measures at the EDA community level to revitalize
fundamental research and improve interactions between academic
researchers and the industry. A prototype implementation of the
proposed infrastructure is available at our Web site.
Christoph Meinel, University of Trier
Web portals are an important basis for scientific cooperation in
the Internet age. At present there are only a few portal sites that
focus on specific research communities. While research portals
have some aspects in common with ordinary portals, there are also
a number of special requirements, often specific for the
community targeted. The research portal discussed here focuses
on the area of Binary Decision Diagrams. The special feature
needed is the possibility to try out research software in a
controlled, standardized environment, in this case usable via a
web interface. Up to now it was extremely difficult to evaluate the
performance of new methods because published benchmark
results are usually very limited, and because the conditions under
which published results were obtained are nearly impossible to
duplicate (since the performance of these methods is very
sensitive to the actual computational environment used). We
believe that our portal greatly facilitates cooperation between
different research groups and allows a significantly higher level of
accuracy in publishing research results for Binary Decision
Diagrams, by publication of actual functionality via this portal
site. Furthermore, we believe it is the first such effort that allows
to get an overview over members of the research community,
relevant events, and other materials available in the World Wide
Web, without the need for long and tedious searches.
The DAC Portal: Mission and Early Status
Ellen Sentovich, Cadence Berkeley Labs