Accelerating Shor's Factorization Algorithm on GPUs

TL;DR

This paper presents a GPU-based simulator for Shor's quantum algorithm, demonstrating significant speedups over classical CPU implementations and Microsoft Liquid, thus enhancing quantum algorithm simulation efficiency.

Contribution

The study introduces a GPU implementation of Shor's algorithm that outperforms existing classical and quantum simulation platforms in speed.

Findings

GPU implementation achieves 52.5x speedup over single-core CPU.

GPU implementation achieves 20.5x speedup over Liquid quantum simulator.

GPU vector operations are well-suited for simulating Shor's algorithm.

Abstract

Shor's quantum algorithm is very important for cryptography, since it can factor large numbers much faster than classical algorithms. In this study, we implement a simulator for Shor's quantum algorithm on graphic processor units (GPU) and compare our results with Liquid -which is Microsoft quantum simulation platform- and two classical CPU-implementations. We evaluate 10 benchmarks for comparing our GPU implementation with Liquid and single-core implementation. The analysis shows that GPU vector operations is more suitable for Shor's quantum algorithm. Our GPU kernel function is compute-bound, due to all threads in a block reach to the same element of the state vector. Our implementation has $52.5\times$ speedup over single-core algorithm and $20.5\times$ speedup over Liquid.