cuQuantum SDK: A High-Performance Library for Accelerating Quantum Science

We present the NVIDIA cuQuantum SDK [1], a state-of-the-art library of composable primitives for GPU-accelerated quantum circuit simulations. As the size of quantum devices continues to increase, making their classical simulation progressively more difficult, the availability of fast and scalable quantum circuit simulators becomes vital for quantum algorithm developers, as well as quantum hardware engineers focused on the validation and optimization of quantum devices. The cuQuantum SDK was created to accelerate and scale up quantum circuit simulators developed by the quantum information science community by enabling them to utilize efficient scalable software building blocks optimized for NVIDIA GPU-based platforms. The functional building blocks provided cover the needs of both state vector- and tensor network- based simulators, including approximate tensor network simulation methods based on matrix product state, projected entangled pair state, and other factorized tensor representations. By leveraging the enormous computing power of the latest NVIDIA GPU architectures, quantum circuit simulators that have adopted the cuQuantum SDK demonstrate significant acceleration, compared to CPU-only execution, for both the state vector and tensor network simulation methods. Furthermore, by utilizing the parallel primitives available in the cuQuantum SDK, one can easily transition to distributed GPU-accelerated platforms, including those furnished by cloud service providers and high-performance computing systems deployed by supercomputing centers, extending the scale of possible quantum circuit simulations. The rich capabilities provided by the cuQuantum SDK are conveniently made available via both Python and C application programming interfaces, where the former is directly targeting a broad Python quantum community and the latter allows tight integration with simulators written in any programming language.

I. Introduction

Quantum circuit simulators are a critical part of quantum algorithm and application development workflows. Today's quantum computers are prohibitively small, error-prone, hard to access, capacity-constrained, and, at times, expensive. Even as they scale, this will not likely change until Fault-Tolerant Quantum Computing (FTQC) systems are broadly deployed and no longer capacity-constrained. Therefore, researchers and developers rely on quantum circuit and analog simulators as a critical tool in their toolbox. Many of these simulators are based on state vector (SV) and tensor network (TN) simulation methods, both of which rely heavily on linear algebra and matrix/tensor multiplications. Graphics Processing Units (GPUs) have traditionally been great computational engines for these types of problems, given their ability to utilize thousands of threads to efficiently parallelize computations. For these reasons, NVIDIA introduced the cuQuantum SDK with its two main component libraries, euState Vec and euTensorNet (Fig. 1). Our strategy is focused on accelerating and scaling up all quantum circuit simulators on GPUs. By working to improve GPU kernels and provide other performance enhancements, in addition to enabling advanced simulation techniques, we have provided simulator developers around the world with the ability to perform quantum circuit simulations at scales and speeds previously not available to them. Fig. 1.

Overview of the NVIDIA cuQuantum SDK. QPU stands for quantum processing units.