Design Automation Lab
Faculty Members
Professor Florin Balasa
and Professor
John Lillis
Research Scope
The Design Automation Lab is focused on developing algorithms
and tools for the design of integrated systems.
Today software tools (CAD tools) touch virtually every aspect
of the VLSI design process. The work within the lab
examines a number of these issues.
Research in this field can
largely be considered applied combinatorial optimization
involving both theoretical and experimental aspects. Some
of the topics currently being researched in the group are
given below.
-
Timing optimization: In modern integrated circuit fabrication
technologies, interconnect is increasingly the dominant design
constraint, particularly with respect to system performance. As
a result, many fundamental issues in CAD have required re-thinking
in the past few years. Topics of active research in the group
general pertain to interconnect synthysis tasks such as
Buffer tree construction and timing-driven routing.
-
Standard cell placement: We are studying new combinatorial optimization techniques
for cell placement problems. Of particular interest are
search mechanisms which cover a large solution space. Such
methods being studied are based on constraint relaxation and
constraint imposition.
-
Topological representations of non-slicing floorplans: We are studying different
encoding systems of feasible placement configurations like, for instance,
the sequence-pair representation, the ordered and binary trees, the bounded
sliceline grid.
Of particular interest is the applicability of topological representations
to device-level analog placement problems.
-
Memory management for real-time multidimensional signal processing:
For many real-time multidimensional signal processing applications
(like, for instance, video an image processing, medical imaging, artificial vision,
real-time 3D rendering, advanced audio and speech coding)
a very large part of the
power consumption is due to the data storage and data transfer.
Moreover, also the area cost is for a large part dominated by the memories. Hence, the
optimization of the memory architecture is a crucial step in the design methodology
for this type of applications.
(see more detailed research goals)