GPU Benchmark through QPE Emulator with cuQuantum for Practical Quantum Applications

TL;DR

This paper demonstrates the implementation of Quantum Phase Estimation (QPE) on GPUs using cuQuantum and CUDA, evaluating performance metrics to optimize quantum algorithm emulation for practical applications.

Contribution

It introduces a GPU-based QPE emulator leveraging cuQuantum and CUDA, with detailed performance analysis and efficient data handling via HDF5.

Findings

Efficient GPU utilization for QPE emulation.

Detailed performance metrics including time, VRAM, and error.

Potential for practical quantum application benchmarking.

Abstract

The quantum algorithm of Quantum Phase Estimation (QPE) was implemented to make the maximum use of GPU emulation with cuQuantum and CUDA Toolkit by NVIDIA. The input and output were handled by HDF5 to make the process as easy as possible. The computational time, VRAM usage, value error, and overhead was evaluated against the developed application. VRAM usage and the profiler analysis suggested that the developed application could make the maximum use of GPU capability.