Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Quantum resources in resource management systems

Utz Bacher, Mark Birmingham, Christopher D. Carothers, Andrew Damin, Carlos D. Gonzalez Calaza, Ashwin Kumar Karnad, Stefano Mensa, Matthieu Moreau, Aurelien Nober, Munetaka Ohtani, Max Rossmannek, Philippa Rubin, M. Emre Sahin, Oscar Wallis, Amir Shehata, Iskandar Sitdikov, Aleksander Wennersteen

TL;DR

This paper tackles the challenge of integrating heterogeneous quantum resources into existing HPC workload managers by introducing a vendor-agnostic abstraction, QRMI. It proposes a Slurm-based reference architecture with a SPANK plugin and a thin middleware layer that standardizes resource acquisition, task execution, and monitoring across on-prem and cloud backends. Key contributions include the QRMI design, its adapter-based implementation, and deployment in production data centers to enable unified, co-scheduled quantum–classical workflows without modifying core scheduler components. The approach reduces integration overhead, enhances resource visibility, and is extensible to other schedulers and container platforms, paving the way for practical, scalable quantum acceleration in scientific computing.

Abstract

Quantum computing resources are increasingly being incorporated into high-performance computing (HPC) environments as co-processors for hybrid workloads. To support this paradigm, quantum devices must be treated as schedulable first-class resources within existing HPC infrastructure. This enables consistent workload management, unified resource visibility, and support for hybrid quantum-classical job execution models. This paper presents a reference architecture and implementation for the integration of quantum computing resources, both on-premises and cloud-hosted into HPC centers via standard workload managers. We introduce a Slurm plugin designed to abstract and control quantum backends, enabling seamless resource scheduling, minimizing queue duplication, and supporting job co-scheduling with classical compute nodes. The architecture supports heterogeneous quantum resources and can be extended to any workload (and container) management systems.

Quantum resources in resource management systems

TL;DR

This paper tackles the challenge of integrating heterogeneous quantum resources into existing HPC workload managers by introducing a vendor-agnostic abstraction, QRMI. It proposes a Slurm-based reference architecture with a SPANK plugin and a thin middleware layer that standardizes resource acquisition, task execution, and monitoring across on-prem and cloud backends. Key contributions include the QRMI design, its adapter-based implementation, and deployment in production data centers to enable unified, co-scheduled quantum–classical workflows without modifying core scheduler components. The approach reduces integration overhead, enhances resource visibility, and is extensible to other schedulers and container platforms, paving the way for practical, scalable quantum acceleration in scientific computing.

Abstract

Quantum computing resources are increasingly being incorporated into high-performance computing (HPC) environments as co-processors for hybrid workloads. To support this paradigm, quantum devices must be treated as schedulable first-class resources within existing HPC infrastructure. This enables consistent workload management, unified resource visibility, and support for hybrid quantum-classical job execution models. This paper presents a reference architecture and implementation for the integration of quantum computing resources, both on-premises and cloud-hosted into HPC centers via standard workload managers. We introduce a Slurm plugin designed to abstract and control quantum backends, enabling seamless resource scheduling, minimizing queue duplication, and supporting job co-scheduling with classical compute nodes. The architecture supports heterogeneous quantum resources and can be extended to any workload (and container) management systems.

Paper Structure

This paper contains 30 sections, 7 figures.

Table of Contents

  1. Introduction
  2. Background
  3. Resource Management Systems and the SPANK Framework
  4. Quantum Computing Providers and Architectures
  5. IBM Quantum
  6. Pasqal
  7. Personas
  8. Design
  9. Quantum Resource Abstraction Model
  10. Quantum Resource Access Model
  11. On-Premises (Direct Access)
  12. Quantum Cloud Bursting
  13. Quantum Resource Management Interface (QRMI) - Design and Motivation
  14. Full System Architecture
  15. Limitations and Advanced Capabilities
  16. ...and 15 more sections

Figures (7)

  • Figure 1: Quantum resource abstraction model. A quantum resource can represent a single QPU or multiple QPUs. QPUs can have different properties such as certain connectivity between QPUs, or virtualization/parallelization capabilities like partitions or execution lanes.
  • Figure 2: Quantum resources access model. Quantum resources can be accessed using direct connections to quantum computer or through cloud APIs. Quantum resources can be node-bound or shared resources accessible by multiple nodes.
  • Figure 3: QRMI overview. QRMI is a thin middleware layer that contains 3 API groups: resource acquisition, task running and metrics.
  • Figure 4: Workload management system-agnostic integration overview. Flow for different backends.
  • Figure 5: QRMI implementation. Implementation of middleware in Rust with C and Python language bindings.
  • ...and 2 more figures