![](papers/GSEDAnote_ASP_DAC08.gif)
DOWNLOAD PDF
"ASP-DAC 2008" |
ASP-DAC 2008
- ASP-DAC is one of the major design conferences of
the year, the Asian equivalent of DATE in Europe and DAC in
the US. Attending ASP-DAC is by far the fastest way to keep
up with the Asian design viewpoint. This year ASP-DAC 2008
was held in Seoul, Korea in one of its periodic wanderings
into the major design centers of Asia from its roots in Japan.
The CONDEX venue proved to be as well chosen as the Yokohama
convention center.
Jan Rabaey of the University of California at Berkeley
kicked it off with a very thought provoking keynote. He
was discussing System-Level Design (SLD) but
he made it clear the he wasn’t talking about Electronic
System Level design (ESL) or even System Design
Automation (SDA), both terms which concentrate on
SoC design or at best a system in a box. Although it’s
true that in SDA’s case that box could be a car
or a plane, what he was referring to was a distributed
system where the system could be worldwide. Interestingly
a lot of the research that was carried out by Bell Labs
in the ‘50s, ‘60s and ‘70s had to do
with what they called long lines (the nation
wide distributed telephone infrastructure). This was the
predecessor to what Jan was talking about. Even more interestingly,
some of that research in simulation and routing ended
up being applied to IC design. Maybe the virtualization
technology Jan was calling for isn’t as far out
of the box as we think it is.
One of the reoccurring themes at ASP-DAC this year was Networks
on Chips (NoC). One of the complaints about the OSCI
2.0 release is that it is based on bus protocol and therefore
doesn’t work in NoC designs. The standards community
is having a hard time keeping up with the rapidly changing
world of ESL.
Another topic of interest was the battle between model-based
ESL and language-based ESL methods. These two approaches seem
to be the major cause of confusion over the definition of
ESL. Of course ESL includes both methods; however, the difficulty
in using a model-based design is hard enough for any engineering
team, even before they complicate the design by the lack of
a complete language-based flow to complete their SoC.
F. C. Tseng, Vice Chairman of TSMC, gave a very intriguing
keynote address. He had a little different slant on the emerging
Semiconductor Research Consortia because he included IMEC.
He called them Intel (stand alone), IBM and IMEC, acknowledging
that TSMC was part of the IMEC group. As a response to the
question on TSMC’s position on Restrictive Design
Rules (RDR) he said that they had started a program four
years ago but had recently cancelled the effort after their
customers had rejected the idea as too restrictive.
Yes, ASP-DAC 2008 was a very interesting conference.
Mary A. Olsson Gary Smith Laurie Balch
|