Abstract:Combinatorial optimization (CO) is essential in various real-world decision-making and planning problems. However, most CO problems (COPs) are NP-hard, demanding substantial computational resources with conventional computers. The Ising machine is promising in addressing them efficiently thanks to its natural convergence behavior. Recently, CMOS-based Ising machines have been developed as low-cost COP solvers, which can be categorized as discrete-time (DT) [1–4] and continuous-time (CT) ones [5–7]. Benefiting from scalable coefficient levels and available annealing strategies, DT Ising machines can solve complicated COPs and reach a lower Hamiltonian. However, DT solvers usually suffer from limited parallelism to prevent potential spin-state oscillations. Together with the requirement of synchronized clocks, the time-to-solution (TTS) is long. Moreover, the random number generators (RNGs) required for the annealing lead to significant area and power consumption. In contrast, CT Ising machines remove the clock constraint and enable fully parallel interaction among all spins, leading to an over 1000$\times$ TTS reduction. Yet, current CT designs suffer from limited coefficient precision and lack effective annealing mechanisms, impeding their practicality. Additionally, analog/mixed-signal operations in prior CT designs are susceptible to variations, demanding costly calibration circuits [5]. In light of this, this work presents a variation tolerant CT Ising machine featuring embedded annealing. Prototyped in 65nm CMOS, this design demonstrates a 1056-spin directed King’s Graph with 15-level coefficients. It realizes a short TTS of 20.7ns, showing great potential in solving real-life time-critical COPs.

Double random sources: low-cost method to enhance local optima escaping ability in CMOS-type Ising chips

Design of probabilistic-based Markov Random Field logic gates in 65nm CMOS technology

Design of Nano-Scale Noise Tolerant CMOS Logic Circuits Based on Probabilistic Markov Random Field Approach

A statistical approach for full-chip gate-oxide reliability analysis

Exploiting Uniform Spatial Distribution to Design Efficient Random Number Source for Stochastic Computing

A Variation-Tolerant Continuous-Time Ising Machine with Edram-Based Spin Interaction and Leaked Negative Feedback Annealing

A CMOS-compatible Ising Machine with Bistable Nodes

30.5 A Variation-Tolerant In-eDRAM Continuous-Time Ising Machine Featuring 15-Level Coefficients and Leaked Negative-Feedback Annealing

Characterization of Locality in Spin States and Forced Moves for Optimizations

Joint Optimization of Randomizer and Computing Core for Low-Cost Stochastic Circuits.

An Output-Controllable Stochastic MTJ Spin Based Ising Machine for Solving Combinatorial Optimization Problems

A Study on the Methodology of IC Yield Optimization

A 1024-Spin Scalable Ising Machine with Capacitive Coupling and Progressive Annealing Method for Combination Optimization Problems

FlexSpin: A CMOS Ising Machine With 256 Flexible Spin Processing Elements With 8-b Coefficients for Solving Combinatorial Optimization Problems

A Compute-in-Memory Annealing Processor With Interaction Coefficient Reuse and Sparse Energy Computation for Solving Combinatorial Optimization Problems

Energy-efficient superparamagnetic Ising machine and its application to traveling salesman problems

Numerical study on effect of random dopant fluctuation on double gate MOSFET based 6-T SRAM performance

Annealing Processing Architecture of 28-nm CMOS Chip for Ising Model With 512 Fully Connected Spins

Design of high-performance entangling logic in silicon quantum dot systems with Bayesian optimization

The Orientational Dependence of Single Event Upsets and Multiple-Cell Upsets in 65 Nm Dual DICE SRAM