Harnessing the Power of HPC and Quantum Computing

In our article titled “Hybrid Quantum Computing: Bridging Classical and Quantum Worlds,” we identify half a dozen definitions of the term 'hybrid quantum computing.' Two-thirds of these terms are unrelated to high-performance computing (HPC) applications, such as the use of classical computing for accessing and controlling quantum computers, algorithms that leverage both classical and quantum computation, the implementation of classical conditional logic next to the quantum processor, and the marriage of digital and analog quantum computation. However, two definitions are explicitly dependent on the use of HPC:

1. HPC as an accelerator, leveraging HPC to prevent classical computational bottlenecks from compromising potential opportunities to realize a quantum computing advantage.
2. Quantum for HPC, wherein the quantum processor is the accelerator, providing a quantum computing advantage for HPC-based enterprise applications. 

Together, high performance computing (HPC) and quantum computing (QC) promise to accelerate or outright enable solutions to some of the greatest computational challenges as yet imagined. These challenges are too complex, either in computational steps or in memory requirements, for HPC to efficiently solve on its own.

Beyond their integration, HPC offers additional benefits for quantum computing:

• HPC can be used to simulate small, fault-tolerant quantum computers. Memory requirements limit the size of these simulations, which is why real quantum computers still need to be built.
• HPC can emulate Noisy Intermediate-Scale Quantum (NISQ) computers with their single-qubit and multiple-qubit error rates, enabling relevant research into error mitigation and other areas.
• HPC can be leveraged to develop new quantum technologies, which will, in turn, someday accelerate and improve the development of new materials and new technologies.
• HPC can be used, at small scales, to verify that quantum algorithms can actually solve their respective problems with the intended degree of accuracy and precision.
• HPC can benchmark the overall performance of quantum computers and quantum computing algorithms against the theoretical claims that have been made about them. 

It’s worth noting that HPC-QC integration will not be of value to all businesses. After all, most businesses today don’t leverage HPC by itself. However, large businesses, the universities, and governments tackling today’s greatest challenges with HPC stand to benefit tremendously. Entire industries are poised to be disrupted by the combined computational power that will be unleashed.

Understanding High-Performance Computing (HPC)

High-Performance Computing (HPC) is synonymous with the term 'supercomputer' which is probably more familiar to many. A supercomputer is not just a classical computer with high-end components, although you should certainly expect to see the highest-end components in use. Rather, a supercomputer is the interconnection of many powerful processors and many large memory modules into one physically large classical computer. Imagine taking a desktop or laptop computer and adding a memory module to it, but at a much grander scale.

Technically, HPC is not limited to referring to specially-built supercomputers. The term can also be applied to large groups, or “clusters", of hundreds to even thousands of standalone computers. Each server, or “node", in the cluster is connected to all the other nodes in the cluster via a network. The aggregation of processing power and memory is conceptually similar to a supercomputer, with the main difference being that the components are distributed, potentially around the world.

Whether using a supercomputer or a cluster, the goal of HPC is to solve the most complex computational tasks by using all of these powerful processors with all the memory in parallel. Classical computers are natively serial, dividing workloads into tasks that are then executed sequentially. HPC is fundamentally no different; however, it leverages its architecture to execute larger tasks and more tasks simultaneously. It then becomes possible to process extremely complex problems and massive multidimensional datasets at speeds exceeding a million times faster than the most powerful individual servers. Interestingly, the power of quantum computers comes from the fact that they are natively parallel, inherently processing all quantum information simultaneously.

HPC resources, like quantum computers, are accessible via the cloud. However, networks introduce latency issues. In other words, the transmission of data becomes a bottleneck, slowing computation down. And this bottleneck would apply if HPC and quantum computing resources were connected via a network. Therefore, part of the impetus for integrating quantum computers into HPC centers is to eliminate this latency and facilitate the fastest possible data transfer.

The Synergy of HPC and Quantum Computing

High Performance Computing (HPC) and Quantum Computing are not rival technologies, which is why their integration is being explored. The ways these technologies complement each other include:

• As previously mentioned, HPC can be used to benchmark quantum computers and quantum algorithms, as well as to verify, at small scales, the accuracy of quantum computation.
• HPC can be used to optimize the quantum circuits that quantum computers execute, reducing circuit depth and multi-qubit operations, and thus improving their accuracy and precision.
• Although quantum computers are inefficient at solving small problems, they promise to solve problems that are impractical, or outright impossible, for the most powerful HPC resources.
• Even if a computational advantage is not to be realized, offloading computation to quantum processors can considerably reduce energy consumption, thus reducing energy costs.

In our article titled “What does it mean for quantum computers to be HPC ready?,” we note that HPC-QC integration must consider these following challenges:

• The quantum computers must physically fit within the datacenters, which is not guaranteed considering the size of some quantum computer architectures.
• Middleware must still be developed that can effectively integrate existing HPC software with the frameworks that enable quantum computation.
• Hybrid algorithms must still be developed that effectively harness the full advantages of both technologies.
• Software must still be developed that can manage the workflows involving both technologies, as well as balancing loads should multiples of each resource be available.
• Monitoring tools and performance indicators must still be developed to ensure that maximum performance can be achieved.

It’s worth noting that HPC-QC integration is well underway. Fault-tolerant quantum computers are still quite some time away, however an HPCwire article titled “Pawsey and Quantum Brilliance Announce Hybrid Quantum-Classic Computing Milestone with Room-Temperature Quantum System” announced that an algorithm had been successfully executed at the Pawsey Supercomputing Centre in Australia. The quantum computer has only two qubits and much more work needs to be done, but it’s a start.

Key Benefits of Quantum Integration

The integration of HPC and QC resources has several potential benefits, all of which are significant:

• Maximum computational power, leveraging the strengths of both technologies
• Enabling the solution of classically-intractable problems, such as combinatoric problems
• Optimal quantum algorithms, which remain challenging to design efficiently
• Accelerated machine learning, which is used throughout the full quantum stack
• True cybersecurity, by eliminating the transmission of information
• Protection of intellectual property, by keeping all data in-house
• Preservation of national security, especially considering cryptographic applications

A Quantum Computing Report article titled “Integrating Quantum Computing into HPC Centers: A Guide for Managers” identifies a dozen considerations that need to be made before making the serious decision to integrate HPC and QC resources in a datacenter. But this article identifies some additional potential benefits of such an integration:

• HPC resources consume vast amounts of energy, which can be mitigated by using QC resources.
• Some QC architectures, like neutral atoms, can be more space efficient than supercomputers.
• At least in the case of QuEra, a strong partnership of maintenance and support is included.
• The combined computational power can be resold, potentially offsetting the investment a bit.
• Training and workforce development can be accelerated with dedicated in-house access.

Some of these points above, especially the energy savings and the potential to resell, contribute to a quicker return on investment.

‍