Abstract:Computer-Aided Design (CAD) tools are extensively used throughout all of the steps in a VLSI design flow. The physical design domain is one of the three domains that are conceptually abstracted in representing a VLSI design flow. This research work focuses on three areas in VLSI physical design domain: clock tree design and buffer insertion, interconnect model order reduction and LVS verification. First, an optimal buffer insertion for clock delay and skew minimization algorithm is developed based on the theory that the minimal clock delay can be obtained by equalizing derivatives of a convex function, and the minimal clock skew can be achieved by equalizing delay functions of different source-to-sink paths. Experimental results show that the presented algorithm achieves both minimal delay and skew in real clock tree design. Further, an automatic clock tree design algorithm is developed for realizing the pre-specified clock arrival time requirements that often occur in an IP block based design environment. In this proposed strategy, planar clock routing and buffer insertion are carried out simultaneously to minimize clock delay and skew, and a full waveform simulation is applied to ensure the signal integrity necessary for high-speed VLSI design. Second, an efficient balanced truncation realization algorithm is presented for interconnect model order reduction. This algorithm shows that there is no need to solve the whole Lyapunov equation for controllability and observability grammians before obtaining approximation to their predominant spaces, and a linear order reduction algorithm can be achieved by extending the O(n) Krylov Subspace Oblique Projection method. Finally, a unique LVS methodology is proposed for the verification of a quad-core microprocessor designed with a triple-well 90-nm CMOS technology. Due to the IP reuse, the standard LVS flow is incapable of handling the additional design complexity in verifying the LVS for the chip designed with multiple power domains. This proposed LVS strategy consists of two phases. The first phase involves verifying LVS at the block level as well as the full-chip level. The second phase aims at verifying the integrity of the multi-power-domain power grid. The proposed LVS methodology was successfully verified by real silicon.

Parallel cross-layer optimization of high-level synthesis and physical design

A novel cross-layer framework for early-stage power delivery and architecture co-exploration.

A Cross-Layer Approach for Early-Stage Power Grid Design and Optimization.

Multilevel Granularity Parallelism Synthesis on FPGAs

Source-code-level Transformation and APT-Driven Parallelism Pre-processes for Embedded System Automated Design

A new methodology of integrating high level synthesis and floorplan for soc design

3D architecture modeling and exploration

Cross-layer Optimization for High Speed Adders: A Pareto Driven Machine Learning Approach

Improving High Level Synthesis Optimization Opportunity Through Polyhedral Transformations.

Automated Communication and Floorplan-Aware Hardware/Software Co-Design for SoC

3DHLS: Incorporating High-Level Synthesis in Physical Planning of Three-Dimensional (3D) ICs

Algorithms and Methodology in Physical Domain Optimization for High-Performance Vlsi Design

CATALYST: Planning Layer Directives for Effective Design Closure

FCUDA-HB: Hierarchical and Scalable Bus Architecture Generation on FPGAs with the FCUDA Flow

Interconnect Power Optimization Based on the Integration of High-level Synthesis and Floorplanning

System-level optimization of Network-on-Chips for heterogeneous 3D System-on-Chips

Optimizing Hardware Design by Composing Utility-Directed Transformations

A CAD-oriented parallel-computing design framework for shape and topology optimization of arbitrary structures using parametric level set

A Practical and Efficient Integrated System for VLSI/ULSI Physical Design

Parallel Design of Intelligent Optimization Algorithm Based on FPGA

A new methodology of integrating high level synthesis and floorplanning for nanometer regime design