Approaches to multi-level sequential logic synthesis
DAC '89 Proceedings of the 26th ACM/IEEE Design Automation Conference
NOVA: state assignment of finite state machines for optimal two-level logic implementations
DAC '89 Proceedings of the 26th ACM/IEEE Design Automation Conference
Algebraic structure theory of sequential machines (Prentice-Hall international series in applied mathematics)
FSM decomposition revisited: Algebraic structure theory applied to MCNC benchmark FSMs
DAC '91 Proceedings of the 28th ACM/IEEE Design Automation Conference
An efficient procedure for the synthesis of fast self-testable controller structures
ICCAD '94 Proceedings of the 1994 IEEE/ACM international conference on Computer-aided design
EURO-DAC '91 Proceedings of the conference on European design automation
A parallel state assignment algorithm for finite state machines
HiPC'04 Proceedings of the 11th international conference on High Performance Computing
| Hi-index | 0.01 |
We present a unified framework and associated algorithms for the optimal decomposition and re-decomposition of sequential machines. This framework allows for a uniform treatment of arbitrary decomposition topologies operating at the State Transition Graph (STG) level, while targeting a cost function that is close to the eventual logic implementation. Previous work has targeted specific decomposition topologies via the formulation of decomposition as implicant covering with associated constraints. It is shown that this formulation can be used to target arbitrary desired topologies merely by customizing the constraints during implicant covering. It is shown how this work relates to preserved partitions and covers traditionally used in parallel and cascade decomposition, and how this formulation establishes the relationship between state assignment and FSM decomposition.In many cases, an initial decomposition is specified as a starting point. Attempting to flatten a set of interacting circuits into a single lumped STG in order to modify the decomposition structure could require astronomical amounts of CPU time and memory. Memory and CPU time efficient re-decomposition algorithms that operate on distributed-style specifications and which are more global than those presented in the past have been developed. These algorithms have been implemented in the sequential logic synthesis system, FLAMES, that is being developed at UCB/MIT.