Synopsys, Inc.

Aiqun Cao

Ssu-Min Chang

Sanjay Dhar

Xiaojun Ma

Cheng-Liang Ding

Zuo Dai

Min Pan

Anand Kumar Rajaram

Lin Yuan

Dah-Cherng Yuan

Systems and techniques for clock tree optimization are described. An electronic design automation (EDA) tool can receive a graph that represents a circuit design, wherein a set of trees in the graph can correspond to a set of clock trees in the circuit design. For each tree in the set of trees, a set of leaf node pairs can be determined. Next, for each leaf node pair, a flow can be created in the graph between the two leaf nodes in the leaf node pair. Aggregate flows can be determined for edges in the graph based on the flows. A set of edges based on the aggregate flows can be identified, and then circuitry corresponding to the set of edges can be identified. Next, the identified circuitry in the circuit design can be optimized.

Network flow based framework for clock tree optimization

Systems and techniques are described for context aware clock tree synthesis (CTS). A probability value can be computed for each clock sink in the set of clock sinks, wherein each probability value represents a probability that the corresponding clock sink has a critical clock latency. Next, the set of clock sinks can be clustered into a set of clock sink clusters based on the probability values. An optimization goal for each clock sink cluster can be selected, and an optimized subtree can be constructed for each clock sink cluster based on the selected optimization goal. The synthesized clock tree can be obtained by combining the optimized subtrees.

Clock tree synthesis based on computing critical clock latency probabilities

Techniques and systems for guiding circuit optimization are described. Some embodiments compute a set of aggregate slacks for a set of chains of logic paths in a circuit design. Each chain of logic paths starts from a primary input or a sequential circuit element that only launches a signal but does not capture a signal and ends at a primary output or a sequential circuit element that only captures a signal but does not launch a signal. Next, the embodiments guide circuit optimization of the circuit design based on the set of aggregate slacks.

Timing bottleneck analysis across pipelines to guide optimization with useful skew

Dynamic power driven clock tree synthesis is described. Some embodiments can select one or more cells from a cell library based on power ratios of cells in the cell library. The embodiments can then construct a clock tree based on the one or more cells.

Dynamic power driven clock tree synthesis (CTS)

Techniques and systems for performing clock tree synthesis (CTS) across multiple modes are described. Some embodiments traverse one or more clock trees from the root of each clock tree to a set of sinks of the clock tree. During the traversal, each clock gate can be marked with a traversal level, and each sink can be marked with one or more clocks and one or more modes that are associated with the sink. A task queue can then be created based on the information collected during the clock tree traversal and populated with different types of tasks based on skew balancing requirements across different modes, and the task queue can be provided to a CTS engine to achieve high-quality skew-balanced clock trees across all modes.

Multi-mode scheduler for clock tree synthesis

Dual-structure clock tree synthesis (CTS) is described. Some embodiments can construct a set of upper-level clock trees, wherein each leaf of each upper-level clock tree is a root of a lower-level clock tree. Each upper-level clock tree can be optimized to reduce an impact of on-chip-variation and/or cross-corner variation on clock skew. Next, for each leaf of each upper-level clock tree, the embodiments can construct a lower-level clock tree to distribute a clock signal from the leaf of the upper-level clock tree to a set of clock sinks. The lower-level clock tree can be optimized to reduce latency, power consumption, and/or area.

Dual-structure clock tree synthesis (CTS)

Systems and techniques are described for automatically generating a set of non-default routing rules for routing a net in a clock tree based on one or more metrics. The metrics can include a congestion metric, a latency metric, a crosstalk metric, an electromigration metric, and a clock tree level. Next, the embodiments can generate the set of non-default routing rules for routing the net based on one or more metrics. A routing rule can specify how wide the wires are supposed to be and how far apart adjacent wires are to be placed. A non-default routing rule can specify a wire width that is different from the default width and/or specify a spacing (i.e., the distance between two wires) that is different from the default spacing.

Automatic clock tree routing rule generation

Methods and apparatuses are described for optimizing local clock skew, and/or for synthesizing clock trees in an incremental fashion. For optimizing local clock skew, the circuit design can be partitioned into clock skew groups. Next, for each clock skew group, an initial clock tree can be constructed that substantially minimizes worst case clock skew in the clock skew group, and then the initial clock tree can be further optimized by substantially minimizing worst case local clock skew in the clock skew group. For performing incremental clock tree synthesis, a portion of a clock tree in the circuit design can be selected based on a set of modifications to the circuit design. Next, a new clock tree can be determined to replace the selected portion of the clock tree. The circuit design can then be modified by replacing the selected portion of the clock tree with the new clock tree.

Incremental clock tree synthesis

Systems and techniques are described for automatically generating clock tree synthesis (CTS) exceptions. The process can use on one or more criteria to identify sequential circuit elements that can be ignored during clock skew minimization. For example, the process can identify sequential circuit elements whose clock skew cannot be balanced with other sequential circuit elements due to structural reasons, identify sequential timing elements that do not have a timing relationship with other sequential timing elements in the clock tree, and/or identify sequential circuit elements whose data pins have a sufficiently large slack so that clock skew is not expected to cause timing violations at any of the data pins. Next, the process can generate clock tree exceptions based on the identified sequential circuit elements.

Growing community of inventors

Sunnyvale, CA, United States of America

Aiqun Cao