Scaling Sample-Based Quantum Diagonalization on GPU-Accelerated Systems using OpenMP Offload

TL;DR

The paper presents a portable, GPU-accelerated implementation of Sample-based Quantum Diagonalization (SQD) using OpenMP Target offload, focusing on accelerating the Davidson subspace diagonalization. By flattening nested data structures, implementing a device-resident Hamiltonian evaluation, and introducing a persistent configuration cache, the authors achieve up to ~95× per-node speedup on Frontier and notable gains on newer GPUs, while preserving CPU-equivalent MPI decomposition and numerical accuracy. The approach demonstrates strong scalability to large configurations (~10^9) and shows portability across diverse CPU/GPU architectures with minimal code changes. These results substantially reduce the classical bottleneck in SQD and broaden the practical reach of hybrid quantum-HPC electronic-structure calculations.

Abstract

Hybrid quantum-HPC algorithms advance research by delegating complex tasks to quantum processors and using HPC systems to orchestrate workflows and complementary computations. Sample-based quantum diagonalization (SQD) is a hybrid quantum-HPC method in which information from a molecular Hamiltonian is encoded into a quantum circuit for evaluation on a quantum computer. A set of measurements on the quantum computer yields electronic configurations that are filtered on the classical computer, which also performs diagonalization on the selected subspace and identifies configurations to be carried over to the next step in an iterative process. Diagonalization is the most demanding task for the classical computer. Previous studies used the Fugaku supercomputer and a highly scalable diagonalization code designed for CPUs. In this work, we describe our efforts to enable efficient scalable and portable diagonalization on heterogeneous systems using GPUs as the main compute engines based on the previous work. GPUs provide massive on-device thread-level parallelism that is well aligned with the algorithms used for diagonalization. We focus on the computation of ground-state energies and wavefunctions using the Davidson algorithm with a selected set of electron configurations. We describe the offload strategy, code transformations, and data-movement, with examples of measurements on the Frontier supercomputer and five other GPU accelerated systems. Our measurements show that GPUs provide an outstanding performance boost of order 100x on a per-node basis. This dramatically expedites the diagonalization step-essential for extracting ground and excited state energies-bringing the classical processing time down from hours to minutes.

Figures (3)

• Figure 1: Strong scaling performance of GPU accelerated code on 1 to 128 nodes, 8 to 1024 GPUs on Frontier.
• Figure 2: Comparison of performance per GPU across different GPU systems, using inputs for N$_2$ with $3.08\times10^8$ configurations problem. MI250X as the baseline. The figure shows three generations of GPUs: (MI250x, A100), (MI300X, H100), and (MI355X, GB200).
• Figure 3: MPI time-lines for a representative job on 32 nodes of Frontier.