Travel reports,
ARTES mobility
Travel Grant report
by
Radoslaw Szymanek
(June-2003)
Design Automation Conference is the biggest and the most prestigious
event in Electronic Design Automation Community. This year more then
5000 participants came to Anaheim and I was one of them. I came to
Anaheim few days earlier so it was easier for me to adapt to a
different time zone. The conference consisted of many parallel
events. I myself spend all three days from the morning until late
evening at the conference site. Conference attracts probably more
people from industry than academia.
The conference was opened Tuesday morning by program and executive
committee of DAC. It was shortly followed by very interesting keynote
speech by Sir Robin Saxby, who is CEO of ARM. His speech was titled
"100nm ... a giant leap for mankind?". Very interesting talk about
possible future which will be heavily influenced by current
developments in Electronic Design Automation industry.
Tuesday morning I partially followed panel discussion about reshaping
EDA industry for power. Later, I went to see one article presentation
in session about Embedded Hardware Design Studies. Right after this
session during lunch break I went for the first time to the exhibit
floor. There were more then 200 companies presenting themselves and
their work. Just a walk through the whole exhibition forum plus few
short conversations ate away my two hours lunch break. After lunch
break I went back to the presentation floor to the next embedded
system session. This session was about Lower-Power Embedded System
Design.
I left a bit earlier this time since next session will be my session
and I wanted to do last minute preparation and relax a bit.
At 16.30 I was introduced by Nikil Dutt, who was the chairman of my
session. My talk was scheduled to take maximum 25 minutes followed by
5 minutes question time. I talked about constraints driven synthesis
and partial assignment technique to simply partitioning and scheduling
problem. My work touches on problems which are not easy to solve and
even 25 minutes is not enough to give deep overview of my
method. Hopefully people working with this type of problems were able
to grasp some intuition behind my approach. After my presentation I
have received one question from the chairman and four
questions from the audience. The questions I have received indicated
that some people did manage to get a good overview of my method and if
necessary could deepen it by reading my article. In my session
there was a presentation of the paper from Swiss Federal Institute of
Technology by Kubilay Atasu and et. This paper was later given the
best paper award. After my session I could relax, I went to a PhD
forum where I could see what other students are doing.
Wednesday morning I have spent discussing with people from companies
such as Celoxica, Magma, Cadence, Synopsis, Mentor Graphics. I have
followed more then 5 different presentations of the recent
developments in those companies. After lunch I have
followed very interesting session about Novel Techniques in High Level
Synthesis. After this session I decided to go to another panel
discussion which had a following title "Nanometer Design: Place Your
Bets". Each of the panelists had a very
interesting view about the future developments and what he thinks will
have highest impact on industry. I particularly liked the insights of
John Cohn from IBM. All panelist have agreed that the biggest
challenges ahead lies at technology level, when it comes to producing
chips that more and more variability of all processes comes to the
picture which make it more and more difficult to produce correct
chip. After this panel I spend evening hours talking to
few other companies like Intel, Accel Chip, and Synplicity. Very
interesting long discussion I had with people from Accel Chip, who are
starting their embedded system design flow in MATLAB, which raises the
modeling level quite high.
Last day of the conference I spent in two panels and two
sessions. First panel was about the importance of the libraries and
modular design flow, where the second was about significance of formal
verification methods and simulation in debugging and verification
process of the electronic circuits. Both panels had a vivid and hot
discussion with high degree of interaction with the audience. The main
message of the first panel was that libraries help to decrease time to
market. The important message of the second panel was that
verification takes nowadays at least 50% of project time and gets longer and
longer and there are immediate needs to short that. Formal
verification can help in achieving faster verification, but will never
replace simulation.
In the middle of the day there was a keynote speech by professor Alberto
L. Sangiovanni-Vicentelli from university of California, Berkeley,
CA. This speech compared the phases of the EDA industry to a Greek
mythology where there was an age of gods, heros, and people. Great
speech with amazing overview of the most significant developments in
EDA and interesting insights to the future. One of most interesting
statements from my point of view was that the process designing of
chips has to go to higher levels of abstractions in order to cope with
chip complexities.
I had a great time at the conference. I only wish there were not so
many parallel sessions, panels which forced me to miss some other
interesting discussions or presentations. I had a chance to meet many
important people in our field, see for myself what is the state of the
art in my field, and participate in discussion about current and
future work in our field.
|