HpQC: a New Efficient Quantum Computing Simulator

EasyChair Preprint 4050

Abstract

With the continuous popularization of quantum computing, high-efficiency quantum computing simulators have attracted the attention of researchers in many related fields, because the running time and memory overhead of quantum computing is increased exponentially, which means that it is challenging to be simulated on a traditional computer. The current mainstream work solves this problem by using multi-node clusters, and we find that its single-node performance has not been effectively exerted. This paper proposes HpQC (High-performance Quantum Computing), which is a simulator that can efficiently parallel quantum computing on a single-node multi-core processor. First, HpQC used AVX2 and FMA instruction sets to maximize the advantages of SIMD (Single Instruction Multiple Data) vectorizations; second, reduced the CPU calculation cycle by using faster and more efficient bit operations; and finally, we designed innovation data structure to utilize spatial locality of cache effectively. Besides, this article selects the state-of-the-art quantum computing simulator, QuEST (the Quantum exact simulation toolkit), as the benchmark for performance evaluation. Experimental results show that HpQC has a remarkable performance improvement compared with QuEST. For the quantum fourier transform, HpQC can achieve an average acceleration of 2.20x (GNU compiler) and 1.91x (Intel compiler), respectively, compared to QuEST. As for the random quantum circuit program, HpQC can achieve an average speedup of 1.74x (GNU compiler) and 1.51x (Intel compiler), respectively, compared to QuEST. Not only that, but HpQC also has excellent load balancing and universality.

Keyphrases: HpQC, Memory Access Optimization, Quantum Fourier Transform, Quest, SIMD, random quantum circuits

@booklet{EasyChair:4050,
  author    = {Haodong Bian and Jianqiang Huang and Runting Dong and Yuluo Guo and Xiaoying Wang},
  title     = {HpQC: a New Efficient Quantum Computing Simulator},
  howpublished = {EasyChair Preprint 4050},
  year      = {EasyChair, 2020}}