Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Quantum Mini-Apps: A Framework for Developing and Benchmarking Quantum-HPC Applications

Nishant Saurabh, Pradeep Mantha, Florian J. Kiwit, Shantenu Jha, Andre Luckow

TL;DR

A mini-app framework is introduced that offers the necessary abstractions for creating and executing mini-apps across heterogeneous quantum-HPC infrastructure, making it a valuable tool for performance characterizations and middleware development.

Abstract

With the increasing maturity and scale of quantum hardware and its integration into HPC systems, there is a need to develop robust techniques for developing, characterizing, and benchmarking quantum-HPC applications and middleware systems. This requires a better understanding of interaction, coupling, and common execution patterns between quantum and classical workload tasks and components. This paper identifies six quantum-HPC execution motifs - recurring execution patterns characterized by distinct coupling and interaction modes. These motifs provide the basis for a suite of quantum mini-apps - simplified application prototypes that encapsulate essential characteristics of production systems. To support these developments, we introduce a mini-app framework that offers the necessary abstractions for creating and executing mini-apps across heterogeneous quantum-HPC infrastructure, making it a valuable tool for performance characterizations and middleware development.

Quantum Mini-Apps: A Framework for Developing and Benchmarking Quantum-HPC Applications

TL;DR

A mini-app framework is introduced that offers the necessary abstractions for creating and executing mini-apps across heterogeneous quantum-HPC infrastructure, making it a valuable tool for performance characterizations and middleware development.

Abstract

With the increasing maturity and scale of quantum hardware and its integration into HPC systems, there is a need to develop robust techniques for developing, characterizing, and benchmarking quantum-HPC applications and middleware systems. This requires a better understanding of interaction, coupling, and common execution patterns between quantum and classical workload tasks and components. This paper identifies six quantum-HPC execution motifs - recurring execution patterns characterized by distinct coupling and interaction modes. These motifs provide the basis for a suite of quantum mini-apps - simplified application prototypes that encapsulate essential characteristics of production systems. To support these developments, we introduce a mini-app framework that offers the necessary abstractions for creating and executing mini-apps across heterogeneous quantum-HPC infrastructure, making it a valuable tool for performance characterizations and middleware development.
Paper Structure (16 sections, 4 figures, 1 table)

This paper contains 16 sections, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Quantum-HPC Integration Types and Execution Motifs
  3. Stages of Developing Quantum Mini-Apps
  4. Quantum-HPC Integration Types
  5. Execution Motifs
  6. Discussion
  7. Quantum Mini-Apps
  8. Design Objectives
  9. Mini-App Framework
  10. Mini-App Examples
  11. Quantum Simulation
  12. Variational Quantum Algorithms
  13. QML Pipeline
  14. Experimental Results
  15. Related Work
  16. ...and 1 more sections

Figures (4)

  • Figure 1: Quantum-HPC integration types, execution motifs, and mini-app conceptual stages
  • Figure 2: Quantum-HPC core motifs: circuit execution, distributed state vector, variational algorithms (standard and optimized), and the pipeline motif: quantum-HPC applications can exhibit complex task structures and increasingly integrate optimized HPC capabilities, e. g., GPU acceleration and other forms of parallelization.
  • Figure 3: Quantum Mini-App Framework Architecture: The framework facilitates the creation, execution, and management of quantum mini-apps. It comprises a mini-app executor and cluster manager abstracting different computing environments, including HPC, cloud, and quantum resources.
  • Figure 4: Circuit execution motif characterization on Perlmutter: compute time for 1024 randomized circuit tasks for 25 qubits. Executed 1024 circuit execution tasks on up to 16 Perlmutter GPU & CPU nodes, where each GPU node has 4 A100 GPUs and CPU node has 128 cores. We scale from 1 node (4/128 GPUs/Cores) to 16 nodes (64/2048 GPUs/Cores).