Part 2: Colocation, High Performance Computers, and Co-Design
by Amara Graps
The hybrid classical-quantum computing environments present a bewildering variety of functionality. Here, we continue to show where and how you might see this activity.
Colocation and High Performance Computers
You may hear the term colocation, with Quantum Computing as a Service (QCaaS), as these qubits are joined with High Performance Computers (HPC), i.e., classical systems (bits), to provide an environment well-suited for hybrid classical-quantum computing.
The joining of HPCs with quantum computers is a good fit, as seen in these two white papers by IQM and Atos:
- Atos-IQM State of quantum research in HPC centers (2021)
- Atos-IQM Bringing quantum acceleration to supercomputers (2022)
The Atos-IQM 2021 paper concludes that:
- staying ahead of the competition is a top priority for 52% of HPC centers worldwide,
- by 2023, 76% of HPC centers worldwide will be using quantum technology- the majority with an on-premises infrastructure, and that
- 49% of HPC centers plan to adopt quantum computing technologies for the first time by 2023.
If you are an HPC center, and have not yet adopted quantum computing, the question you should be asking yourself: When will you adopt quantum?
The Atos-IQM 2022 paper expands on the colocation theme, how the allocation of system resources benefits from on-premises location, and why now is a good time for the quantum and HPC stakeholders to align towards standardization, which will accelerate the technological development and its adoption.
The joining of HPCs with quantum computers is additionally a good fit from the A.I. perspective. We suggest one reason why quantum technology is not more pervasive in HPC centers is because A.I. entered front and center, after 2022. But the adoption of HPC, AI, big data analytics, and quantum computing working together will become inevitable. Although the exact implementation may differ based upon the type of organization (national research laboratory, corporate business, or academic institution) that will be using it.
We recommend the 17 min talk by Shahin Khan at the Q2B23 SV titled: Integrated Heterogeneous Supercomputing, where he places HPCs in its rightful place to drive innovation in the IT Industry, now updated with A.I. and quantum computing. Khan’s talk provides a business framing, which is especially suitable:
There is an integration opportunity that can drive business. Consider A.I. as an HPC application and Quantum Computing as an HPC Library Call.
The following are examples of Multiple real QPU cloud vendors with quantum simulators. The QPU vendors are listed in (). Some of these cloud entities offer quantum-inspired vendors, as well.
- Amazon AWS Braket (IonQ, IQM, QuEra, Rigetti)
- QM Ware (QuEra, QuiX)
- QuTech Quantum Inspire (Spin-4, Starmon-5)
- qBraid (Amazon AWS Braket offerings, IonQ, NVIDIA cuQuantum, Oxford OQC)
- Strangeworks (D-Wave, NEC quantum annealer, QuEra, others)
- Microsoft Azure Quantum (Quantinuum, QCI, IonQ, Rigetti, Pasqal)
- Beijing Academy of Quantum Information Sciences BAQIS Quafu (BAQIS’ own superconducting quantum computer, USTC/Jiuzhang, USTC/ Zuchongzhi 2)
Examples of in-progress integrations of real QPUs with HPCs:
- Riverlane, Rigetti, and ORNL research on an HPC-Quantum Integration Using ORNL’s Summit Supercomputer, which should be reporting results in September.
- The acquisition and integration of quantum computers into existing European HPC data centers continues, this time in Italy.
- RIKEN has selected Quantinuum’s H1 ion-trap quantum computer for its hybrid quantum-supercomputing platform. The H1 processor is expected to be installed and operational by early 2025.
Hybrid Classical-Quantum Computing Interactions
In the research literature, you’ll likely see any one of these definitions for hybrid classical-computing. Many thanks to QuEra.
- Standard. Classical computers transfer data to the quantum processor, manage the control system, interpret the measurements, provide cloud accessibility, and perform additional functions. The quantum processor solely focuses on computation and is not very useful without being fully integrated with classical systems as hybrid computing systems.
- HPC_integration_Classical_Tasks. The incorporation of high performance computing (HPC) by leveraging HPC to accelerate some or all of the classical tasks.
- HPC_integration_Classical_VQAs. The incorporation of high performance computing (HPC) by leveraging HPC to accelerate the variational quantum algorithms (VQA).
- Parameterized Quantum Circuits. This is a detailed description of a hybrid system that is usually referenced in deep tech circles to illustrate the use of A.I. The quantum algorithms consist of parameterized quantum circuits, which run on real quantum computers or quantum computer simulators, and classical neural networks, which train and update the parameters. These systems are optimal if classical computer and quantum computer are colocated.
- Dynamic Quantum Circuit. Like the Standard, except some classical logic is executed at the quantum processor level as one run of a quantum circuit, based on the results of quantum measurements. Those decisions are used to condition and control future quantum operations within the coherence times of quantum registers. In this way, some quantum information can be preserved on qubits on which mid-circuit measurements are not applied, whereas quantum information is lost in between runs of multiple quantum circuits as in the Standard method.
The Concept of Co-Design
Similarly, it should be no surprise that some QPU vendors fine-tune their quantum processors for best outcome for specific problems. It’s a concept they call: co-design. Hybrid classical-quantum computing serves a primary function of Useability or Practical Utility.
Wintersperger, et al., 2022, QPU-System Co-Design for Quantum HPC Accelerators, explains:
Problem formulation and algorithm, but also the physical QPU properties are tailored to the specific application. Since the QPUs will likely be used as accelerators for classical computers, details of systemic integration into existing architectures are another lever to influence and improve the practical utility of QPUs.
In Part 3 of The Many Faces of Hybrid Classical-Quantum Computing, we’ll continue with the primary algorithms VQA, VQE and QAOA.
August 26, 2024