Abstract:Scaling up the number of qubits and speeding up the execution of quantum algorithms are important steps towards reaching quantum advantage. This poses heavy demands particularly on the control stack, as pulses need to be distributed to an increasing number of control channels and variational algorithms require rapid interleaving of quantum and classical computation. Assessing the bottlenecks in the control stack is therefore key to making it ready for reaching quantum advantage. However, existing benchmark suites suffer from lack of detail due to indirect access to the control hardware. In this work, we present Q-Profile, a tool to profile quantum control stacks that circumvents these issues by utilizing a direct connection from the host CPU to the control stack, providing fine accuracy in measuring the runtime and allowing to identify performance bottlenecks. We demonstrate the use of our tool by benchmarking the Quantum Approximate Optimization Algorithm (QAOA) on a Qblox Cluster for a virtual 4 to 14-qubit transmon system. Our results identify the major execution bottlenecks in the passive qubit reset and communication overhead. We estimate a 1.40x~speedup with respect to the total runtime by using an active qubit reset, instead of passive reset, and demonstrate a further speedup of 1.37x by parallel initialization of the control modules. The presented method of profiling is applicable to other control-stack providers, as well as to other benchmarks, while still providing detailed information beyond a single metric. By extension, this tool will enable identifying and eliminating bottlenecks for future quantum acceleration. The profiling tool is included in the open-source Quantify quantum control software, which allows support for multiple back-ends.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is the performance bottlenecks in the quantum control stack when executing quantum algorithms, especially during the process of achieving quantum advantage. Specifically, the paper focuses on how to identify and alleviate these bottlenecks through accurate performance - analysis tools, thereby accelerating the execution of quantum algorithms, especially for the execution of approximate optimization algorithms such as QAOA. ### Main Problems 1. **Performance Bottlenecks of the Quantum Control Stack**: - The paper points out that as the number of qubits increases and the execution speed of quantum algorithms improves, the requirements for the quantum control stack (including hardware and software) are getting higher and higher. This has led to multiple performance bottlenecks in the control stack, especially the distribution of pulses to an increasing number of control channels and the need for variational algorithms to rapidly alternate between quantum and classical computations. - Existing benchmarking suites lack details due to indirect access to control hardware and are unable to accurately assess these bottlenecks. 2. **Identification and Alleviation of Specific Bottlenecks**: - **Passive Reset and Active Reset**: The paper finds that passive reset is a major execution bottleneck. By using active reset, the reset time can be significantly reduced, thereby improving the overall execution efficiency. - **Communication Overhead**: The paper also finds that communication overhead is another major bottleneck. By initializing control modules in parallel, the communication time can be further reduced. ### Solutions 1. **Q - Profile Tool**: - The paper introduces Q - Profile, a tool for benchmarking the quantum control stack. This tool provides high - precision runtime measurements by directly connecting the host CPU and the quantum control stack and can identify performance bottlenecks in detail. - The Q - Profile tool can bypass the problem of long - distance and high - latency connections, provide fine - grained runtime measurements, and help developers identify and alleviate performance problems. 2. **Experimental Verification**: - The paper uses the Q - Profile tool to benchmark the QAOA algorithm and shows its performance on virtual superconducting quantum systems of different scales (4 to 14 qubits). - The experimental results show that by using active reset and parallel initialization of control modules, the execution speed of the QAOA algorithm can be significantly improved. ### Specific Improvements - **Active Reset**: Using active reset instead of passive reset can reduce the total runtime by 1.40 times. - **Parallel Initialization**: By initializing control modules in parallel, the total runtime can be further reduced by 1.37 times. ### Conclusion By introducing the Q - Profile tool, the paper successfully identifies and alleviates the main performance bottlenecks in the quantum control stack, providing strong support for achieving quantum advantage. This tool is not only applicable to Qblox's control stack but can also be applied to other control stack providers and different benchmarking, having broad application prospects.

Q-Profile: Profiling Tool for Quantum Control Stacks applied to the Quantum Approximate Optimization Algorithm

QOPS: A Compiler Framework for Quantum Circuit Simulation Acceleration with Profile Guided Optimizations

QOS: A Quantum Operating System

Profiling quantum circuits for their efficient execution on single- and multi-core architectures

Optimization Applications as Quantum Performance Benchmarks

Quantum Algorithm Exploration using Application-Oriented Performance Benchmarks

The QAOA with Few Measurements

Scaling of the quantum approximate optimization algorithm on superconducting qubit based hardware

Superstaq: Deep Optimization of Quantum Programs

Evidence of Scaling Advantage for the Quantum Approximate Optimization Algorithm on a Classically Intractable Problem

Towards a universal QAOA protocol: Evidence of a scaling advantage in solving some combinatorial optimization problems

Prog-QAOA: Framework for resource-efficient quantum optimization through classical programs

Coqa: Blazing Fast Compiler Optimizations for QAOA

Approximating under the Influence of Quantum Noise and Compute Power

A Multilevel Approach For Solving Large-Scale QUBO Problems With Noisy Hybrid Quantum Approximate Optimization

A Synergistic Compilation Workflow for Tackling Crosstalk in Quantum Machines

One-Time Compilation of Device-Level Instructions for Quantum Subroutines

QPack Scores: Quantitative performance metrics for application-oriented quantum computer benchmarking

Fast Simulation of High-Depth QAOA Circuits

Multiscale Quantum Approximate Optimization Algorithm