CONQURE: A Co-Execution Environment for Quantum and Classical Resources

Cutting edge classical computing today relies on a combination of CPU-based computing with a strong reliance on accelerators. In particular, high-performance computing (HPC) and machine learning (ML) rely heavily on acceleration via GPUs for numerical kernels. In the future, acceleration via quantum devices may complement GPUs for kernels where algorithms provide quantum advantage, i.e., significant speedups over classical algorithms. Computing with quantum kernels mapped onto quantum processing units (QPUs) requires seamless integration into HPC and ML. However, quantum offloading onto HPC/cloud lacks open-source software infrastructure. For classical algorithms, parallelization standards, such as OpenMP, MPI, or CUDA exist. In contrast, a lack of quantum abstractions currently limits the adoption of quantum acceleration in practical applications creating a gap between quantum algorithm development and practical HPC integration. Such integration needs to extend to efficient quantum offloading of kernels, which further requires scheduling of quantum resources, control of QPU kernel execution, tracking of QPU results, providing results to classical calling contexts and coordination with HPC scheduling. This work proposes CONQURE, a co-execution environment for quantum and classical resources. CONQURE is a fully open-source cloud queue framework that presents a novel modular scheduling framework allowing users to offload OpenMP quantum kernels to QPUs as quantum circuits, to relay results back to calling contexts in classical computing, and to schedule quantum resources via our CONQURE API. We show our API has a low overhead averaging 12.7ms in our tests, and we demonstrate functionality on an ion-trap device. Our OpenMP extension enables the parallelization of VQE runs with a 3.1X reduction in runtime.

💡 Research Summary

The paper introduces CONQURE, an open‑source cloud‑queue framework designed to bridge quantum processing units (QPUs) with classical high‑performance computing (HPC) and machine‑learning (ML) workloads. The authors argue that while classical acceleration relies on mature standards such as OpenMP, MPI, and CUDA, quantum acceleration lacks comparable abstractions, creating a gap between quantum algorithm development and practical HPC integration.

CONQURE’s architecture consists of five modular layers: (1) a user‑interface layer that accepts quantum kernels written in popular Python libraries (Qiskit, Cirq, tket, etc.); (2) a translation layer that converts platform‑agnostic quantum operations into hardware‑specific intermediate representations (QIR, QASM) and applies vendor‑specific optimizations; (3) a workload‑manager layer that implements a cloud‑queue system combined with SLURM‑style scheduling, supporting priority policies, asynchronous submissions, and multi‑node coordination; (4) a persistent database that stores job metadata, circuit definitions, and results for reproducibility; and (5) a quantum‑control layer that provides pluggable drivers for pulse‑level control across superconducting, ion‑trap, and neutral‑atom platforms. This stack is deliberately interchangeable: any layer can be swapped without affecting the others, enabling rapid adaptation to new hardware or scheduling policies.

The core software contribution is OpenMP‑Q, an extension of the widely‑used OpenMP standard. By adding a “target(device(Quantum))” clause and a “reverse‑offload” pragma, developers can offload quantum kernels from C/C++ code while retaining the familiar OpenMP syntax. The reverse‑offload mechanism allows the quantum kernel to remain active while classical code runs asynchronously on the host; the host can process measurement results, solve a classical sub‑problem (e.g., parameter optimization), and feed updated parameters back into the next quantum iteration. This tight quantum‑classical feedback loop is especially valuable for variational algorithms such as the Variational Quantum Eigensolver (VQE). The authors also demonstrate multi‑quantum offloading using “#pragma omp parallel”, where each OpenMP thread controls a distinct QPU (or simulator), enabling concurrent evaluation of different ansätze or parameter sets. MPI calls can be layered on top to aggregate results across nodes, effectively scaling to multi‑QPU clusters.

To ensure compatibility with existing cloud services, CONQURE implements an API that mirrors the AWS Quantum Queue specification (functions like create_work, get_results, wait_until_done). The API supports both public cloud back‑ends and private, authenticated clouds, allowing research groups to run workloads on in‑house QPUs or on commercial providers without code changes. Asynchronous job submission and persistent job tracking reduce user‑side latency and simplify debugging.

Performance evaluation focuses on two aspects. First, the API overhead was measured at an average of 12.7 ms per call, roughly 30 % lower than the baseline AWS queue implementation. Second, the authors executed a VQE workload on an ion‑trap device using four OpenMP threads (each driving a separate quantum task). Compared with a sequential execution, the runtime decreased by a factor of 3.1. The speed‑up stems from parallel gate execution, reduced idle time between quantum and classical phases, and the ability to update rotation angles on‑the‑fly via parametric pulse shaping.

The paper also discusses broader implications. By providing an open, modular, and authenticated framework, CONQURE enables collaborative research across institutions, educational deployments, and secure in‑house quantum services—capabilities that are limited in proprietary platforms. Future work outlined includes sophisticated multi‑QPU scheduling algorithms, tighter integration with GPU‑accelerated classical solvers, automatic insertion of quantum error‑correction circuits during translation, and deeper coupling with large‑scale HPC resource managers.

In summary, CONQURE fills a critical software gap between quantum and classical computing, offering a practical, extensible pathway for HPC practitioners to harness quantum acceleration without abandoning familiar programming models. Its modular design, low‑latency API, and demonstrated speed‑ups on real quantum hardware make it a promising foundation for the emerging field of quantum‑enhanced high‑performance computing.