Abstract:Sequential logic synthesis can provide better Power-Performance-Area (PPA) than combinational logic synthesis since it explores a larger solution space. As the gate cost in advanced technologies keeps rising, sequential logic synthesis provides a powerful alternative that is gaining momentum in the EDA community. In this work, we present a new scalable algorithm for don't-care-based sequential logic synthesis. Our new approach is based on sequential k-step induction and can apply both redundancy removal and resubstitution transformations under Sequential Observability Don't Cares (SODCs). Using SODC-based optimizations with induction is a challenging problem due to dependencies and alignment of don't cares among the base case and the inductive case. We propose a new approach utilizing the full power of SODCs without limiting the solution space. Our algorithm is implemented as part of an industrial tool and achieves 6.9% average area improvement after technology mapping when compared to state-of-the-art sequential synthesis methods. Moreover, all the new sequential optimizations can be verified using state-of-the-art sequential verification tools.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to optimize sequential logic synthesis by developing a new scalable algorithm in the modern electronic design automation (EDA) process, so as to improve the area, power consumption and performance (PPA) of the circuit. Specifically, the paper proposes an optimization method based on Sequential Observability Don't Cares (SODCs). This method takes advantage of the characteristics of sequential logic and can perform optimization without moving registers, thus avoiding the logic replication problem in traditional retiming methods and being able to find optimization opportunities that previous methods could not identify. ### Main Problems 1. **Challenges in Sequential Logic Synthesis**: - Sequential logic synthesis is more complex than combinational logic synthesis because it needs to consider state reachability and observability. - Traditional optimization methods (such as retiming) perform poorly when dealing with circuits with sequential feedback, especially in scenarios such as finite - state machines (FSM). 2. **Optimization Using SODCs**: - Observability Don't Cares (ODCs) have been widely used in combinational logic synthesis, but in sequential logic synthesis, it is very difficult to directly apply these techniques due to dependency and alignment problems. - The paper proposes a new method to fully utilize SODCs through sequential induction, so as to optimize without restricting the solution space. 3. **Scalability and Verification**: - The proposed algorithm needs to maintain scalability in large - scale circuits and at the same time ensure that the optimized circuit can be verified by existing sequential verification tools. ### Solutions 1. **Optimization Based on Sequential Induction**: - Use k - step sequential induction to construct the Base Case Network and the Inductive Case Network. - Determine whether the optimization can be applied to the original network by checking whether the optimizations in the Base Case Network and the Inductive Case Network are valid. 2. **Redundancy Removal and Resubstitution**: - The algorithm not only supports redundancy removal but also supports resubstitution based on ODCs. - Use the SAT solver to verify the validity of each optimization to ensure that the optimization does not change the function of the circuit. 3. **Enhancement Techniques**: - Expand to multi - step sequential induction (k - step induction) to capture more complex optimization opportunities. - Consider redundancy assumptions and gradually apply verified optimizations in the Base Case Network and the Inductive Case Network to further improve the optimization effect. ### Experimental Results - **AIG Experiment**: In the AIG benchmark test, the proposed algorithm reduces the number of NAND2 gates by an average of 4.1%. - **Industrial Process Experiment**: In the industrial process, compared with the existing Sequential SAT Sweep (SSW) method, the proposed algorithm achieves a 6.9% area reduction while the running time increases by 20%. ### Conclusion The paper proposes a new sequential logic synthesis algorithm based on SODCs. This algorithm can perform effective optimization without moving registers through sequential induction and redundancy assumption techniques. It is suitable for large - scale circuits and can find new optimization opportunities. The experimental results show that this method is superior to existing methods in area optimization and can be successfully deployed in industrial tools.

Scalable Sequential Optimization Under Observability Don't Cares

DAG-aware Synthesis Orchestration

PowerSyn: A Logic Synthesis Framework with Early Power Optimization

On Accelerating Domain-Specific MC-TS with Knowledge Retention and Efficient Parallelization for Logic Optimization

Exploiting Challenges of Sub-20 nm CMOS for Affordable Technology Scaling

INVICTUS: Optimizing Boolean Logic Circuit Synthesis via Synergistic Learning and Search

Combinational and sequential logic optimization by redundancy addition and removal

A Combinational Logic Optimization Method for Large-Scale SFQ Circuits

Logic Synthesis Optimization with Predictive Self-Supervision via Causal Transformers

Scheduling to timing optimization for a novel high-level synthesis approach

FADO: Floorplan-Aware Directive Optimization Based on Synthesis and Analytical Models for High-Level Synthesis Designs on Multi-Die FPGAs

A Module-Level Three-Stage Approach to the Evolutionary Design of Sequential Logic Circuits

QGo: Scalable Quantum Circuit Optimization Using Automated Synthesis

LEAP: Scaling Numerical Optimization Based Synthesis Using an Incremental Approach

Timing-driven Approximate Logic Synthesis Based on Double-chase Grey Wolf Optimizer

Sequential Optimization in Locally Important Dimensions

A Comprehensive High-Level Synthesis System for Control-Flow Intensive Behaviors

Performance Modeling and Directives Optimization for High-Level Synthesis on FPGA.

Compilation of Dataflow Applications for Multi-Cores using Adaptive Multi-Objective Optimization

Deep-PowerX: A Deep Learning-Based Framework for Low-Power Approximate Logic Synthesis

Logic Synthesis Optimization Sequence tuning using RL-based LSTM and Graph Isomorphism Network