

Efficient Software
Realization of Sequential Machines
Yunjian
Jiang
UC Berkeley
ESD &
Dissertation Seminar talk
Monday, November 24th,
2003, 4  5pm
540A/B Cory Hall (D.O.P. Center Classroom)

Abstract
In this talk we
present a design methodology for embedded control systems, which starts
from highlevel synchronous specifications and produces implementations
in hardware and software for embedded platforms. It includes design
modeling using Extended Finite State Machines (EFSM), technology
independent logic optimization using control and datapath don't cares,
performance driven hardware/software partitioning, and code generation.
A main component of the tool flow is the software realization of EFSMs.
In this regard, we introduce the notion of generalized cofactoring for
multivalued relations, which is a generalization of traditional
functional simulation methods with logic equations, Binary Decision
Diagrams (BDD), sumofproducts and table lookups. This leads to the
discovery of an information theoretic approach for logic simulation,
where source coding techniques are applied to functional simulation by
using entropy as a complexity measure for multivalued relations. We
will also briefly describe the MVSIS project, which is a software
system currently being developed for logic optimization of multivalued
nondeterministic sequential networks.
Speaker
Yunjian Jiang is a
Ph.D. candidate in the department of Electrical Engineering and Computer
Science, UC Berkeley. He received his B. Eng. degree from Tsinghua
University, Beijing, in 1998, and his M. S. degree from UC Berkeley
in 2000. He has interned for companies such as Magma Design
Automation, Cadence Design Systems, Teja Technologies, and IBM T. J.
Watson Research Center.
