QCEDA: Using Quantum Computers for EDA

TL;DR

The paper addresses the challenge of solving NP-hard EDA optimization problems by recasting a representative Min-$k$-Union problem, derived from DRAM scrambler optimization, into a Quadratic Unconstrained Binary Optimization (QUBO) and solving it on real quantum hardware. It presents a concrete QUBO formulation with binary variables $x_M$ and $y_v$ and a three-term objective $H = H_A + H_B + H_C$ using penalties $A,B$ and count term $C$, ultimately setting $A=B=|V|+1$ and $C=1$ to ensure feasibility and optimality. The work compares two quantum-optimization paradigms—QAOA and quantum annealing—and validates the approach through a toy example solved on IBM Quantum and D-Wave Leap, showing feasible outcomes but highlighting current hardware limitations. It provides a scalability discussion, noting linear growth in required qubits with problem size and outlining future work to tackle larger real-world instances and other EDA problems such as scheduling, placement, and routing. Overall, the paper demonstrates a foundational pathway for leveraging quantum devices in EDA optimization, quantifies hardware needs, and sets the stage for potential quantum advantage as qubit resources expand.

Abstract

The field of Electronic Design Automation (EDA) is crucial for microelectronics, but the increasing complexity of Integrated Circuits (ICs) poses challenges for conventional EDA: Corresponding problems are often NP-hard and are therefore in general solved by heuristics, not guaranteeing optimal solutions. Quantum computers may offer better solutions due to their potential for optimization through entanglement, superposition, and interference. Most of the works in the area of EDA and quantum computers focus on how to use EDA for building quantum circuits. However, almost no research focuses on exploiting quantum computers for solving EDA problems. Therefore, this paper investigates the feasibility and potential of quantum computing for a typical EDA optimization problem broken down to the Min-$k$-Union problem. The problem is mathematically transformed into a Quadratic Unconstrained Binary Optimization (QUBO) problem, which was successfully solved on an IBM quantum computer and a D-Wave quantum annealer.

Figures (2)

• Figure 1: Hypergraphs and Solutions for Example Problem Corresponding to Table \ref{['tab:seq_example']}
• Figure 2: Histogram of solutions from D-Wave run