Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

QueenV2: Future of Quantum Circuit Simulation

Chuan-Chi Wang

TL;DR

QueenV2 is proposed, which builds upon the design principles of Queen and elevates performance to a new level, and is positioned to significantly accelerate quantum circuit simulation, thus promoting the development of innovative accelerators and quantum algorithms.

Abstract

A state vector-based quantum circuit simulation can provide accurate results for the development and validation of quantum computing algorithms, without being affected by noise interference. However, existing quantum circuit simulators have consistently underperformed due to inadequate integration with quantum circuits and high-performance computing architectures. To tackle the challenges in quantum computing, we propose QueenV2, which builds upon the design principles of Queen and elevates performance to a new level. Experimental results on the NVIDIA RTX-4090 demonstrate that QueenV2 achieves up to a 40x improvement in gate performance and a 5x improvement in circuit performance compared to hyQuas. Furthermore, QueenV2 realizes a 137x speedup in gate benchmarks and a 14x speedup in circuit performance relative to NVIDIA cuQuantum, enabled by gate fusion via the IBM Qiskit toolkit. By eliminating reliance on third-party libraries, QueenV2 is positioned to significantly accelerate quantum circuit simulation, thus promoting the development of innovative accelerators and quantum algorithms.

QueenV2: Future of Quantum Circuit Simulation

TL;DR

QueenV2 is proposed, which builds upon the design principles of Queen and elevates performance to a new level, and is positioned to significantly accelerate quantum circuit simulation, thus promoting the development of innovative accelerators and quantum algorithms.

Abstract

A state vector-based quantum circuit simulation can provide accurate results for the development and validation of quantum computing algorithms, without being affected by noise interference. However, existing quantum circuit simulators have consistently underperformed due to inadequate integration with quantum circuits and high-performance computing architectures. To tackle the challenges in quantum computing, we propose QueenV2, which builds upon the design principles of Queen and elevates performance to a new level. Experimental results on the NVIDIA RTX-4090 demonstrate that QueenV2 achieves up to a 40x improvement in gate performance and a 5x improvement in circuit performance compared to hyQuas. Furthermore, QueenV2 realizes a 137x speedup in gate benchmarks and a 14x speedup in circuit performance relative to NVIDIA cuQuantum, enabled by gate fusion via the IBM Qiskit toolkit. By eliminating reliance on third-party libraries, QueenV2 is positioned to significantly accelerate quantum circuit simulation, thus promoting the development of innovative accelerators and quantum algorithms.
Paper Structure (13 sections, 3 equations, 5 figures, 4 tables)

This paper contains 13 sections, 3 equations, 5 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Qubit Representation and Quantum Gates
  4. Optimized Techniques for High-Performance Computing
  5. QueenV2
  6. All-in-one circuit module
  7. All-in-cache simulation module
  8. Evaluation
  9. Experimental Setup
  10. Gate Benchmark
  11. Circuit Benchmark
  12. Conclusion
  13. Postscript

Figures (5)

  • Figure 1: QueenV2 workflow for quantum simulation.
  • Figure 2: The circuit diagram for the simulator of QueenV1 and QueenV2.
  • Figure 3: The H-gate benchmarks on RTX-4090 (The time of 0 seconds for hyQuas indicates out-of-memory).
  • Figure 4: The QFT benchmarks with gate fusion on RTX-4090 (The time of 0 seconds for hyQuas indicates out-of-memory).
  • Figure 5: The QAOA benchmarks with gate fusion on RTX-4090 (The time of 0 seconds for hyQuas indicates out-of-memory).