High-Performance Computing Frameworks for Quantum Circuit Simulation

Abstract

Quantum circuit simulation at scale demands high-performance computing (HPC) frameworks capable of distributing exponentially growing state vectors and tensor networks across heterogeneous compute clusters. As quantum algorithm development accelerates and circuits approach 50-100 qubits in practical workloads, simulation bottlenecks shift from single-node memory constraints to inter-node communication overhead, GPU memory bandwidth, and framework-level scheduling inefficiencies. This paper proposes the HPC Quantum Simulation Framework Benchmark (HQSFB), a systematic evaluation of five leading quantum simulation frameworks -- Qiskit Aer, Cirq on GCP, cuQuantum (NVIDIA), QuEST, and Intel Quantum Simulator (IQS) -- across six HPC deployment configurations spanning single-GPU workstations to 512-GPU supercomputer partitions. HQSFB introduces the Framework Efficiency Score (FES) combining simulation throughput, GPU utilisation, inter-node communication overhead, and memory scaling efficiency. Key results: cuQuantum achieves the highest FES (0.924) with 94.2% GPU utilisation and 42x speedup over CPU baseline at 512 GPUs; QuEST achieves the best memory scaling efficiency (0.962) for distributed statevector simulation; Qiskit Aer provides the broadest circuit compatibility with FES = 0.842. The framework provides HPC configuration guidance and an open benchmark suite for quantum simulation infrastructure planning.