High-performance and Low-power Clock Network Synthesis in the Presence of Variation.

摘要

Semiconductor technology scaling requires continuous evolution of all aspects of physical design of integrated circuits. Among the major design steps, clock-network synthesis has been greatly affected by technology scaling, rendering existing methodologies inadequate. Clock routing was previously sufficient for smaller ICs, but design difficulty and structural complexity have greatly increased as interconnect delay and clock frequency increased in the 1990s. Since a clock network directly influences IC performance and often consumes a substantial portion of total power, both academia and industry developed synthesis methodologies to achieve low skew, low power and robustness from PVT variations. Nevertheless, clock network synthesis under tight constraints is currently the least automated step in physical design and requires significant manual intervention, undermining turn-around-time. The need for multi-objective optimization over a large parameter space and the increasing impact of process variation make clock network synthesis particularly challenging.;Our work identifies new objectives, constraints and concerns in the clock-network synthesis for systems-on-chips and microprocessors. To address them, we generate novel clock-network structures and propose changes in traditional physical-design flows. We develop new modeling techniques and algorithms for clock power optimization subject to tight skew constraints in the presence of process variations. In particular, we offer SPICE-accurate optimizations of clock networks, coordinated to reduce nominal skew below 5 ps, satisfy slew constraints and trade-off skew, insertion delay and power, while tolerating variations. To broaden the scope of clock-network-synthesis optimizations, we propose new techniques and a methodology to reduce dynamic power consumption by 6.8%-11.6% for large IC designs with macro blocks by integrating clock network synthesis within global placement. We also present a novel non-tree topology that is 2.3 x more power-efficient than mesh structures. We fuse several clock trees to create large-scale redundancy in a clock network to bridge the gap between tree-like and mesh-like topologies. Integrated optimization techniques for high-quality clock networks described in this dissertation strong empirical results in experiments with recent industry-released benchmarks in the presence of process variation. Our software implementations were recognized with the first-place awards at the ISPD 2009 and ISPD 2010 Clock-Network Synthesis Contests organized by IBM Research and Intel Research.