NuMagSANS – Runtime Experiment

This dataset contains the input data, code and benchmark results of a runtime scaling experiment performed with the GPU-accelerated SANS simulation framework NuMagSANS.The benchmark is based on an analytically defined vortex-like magnetization configuration inside spherical iron nanoparticles with a diameter of $40 \; \mathrm{nm}$ and uniform nuclear scattering-length density. Each nanoparticle is discretized on a simple cubic lattice with a cell size of $2\;\mathrm{nm}$, resulting in $4169$ magnetization cells per particle. The total system size is varied by increasing the number of identical nanoparticles from $N=10$ up to $N=25{,}600$, corresponding to approximately $4\times10^4$ to $10^{8}$ magnetization cells in total. All simulations were executed on a single NVIDIA RTX 3090 GPU using identical numerical settings. The dataset documents the resulting execution times as a function of system size. This repository includes a complete simulation pipeline. A Python script is provided to generate the magnetization configurations and the corresponding nuclear scattering-length density data for particle ensembles containing $N$ identical nanoparticles. A shell script iterates over the selected values of $N$, invokes the Python-based data generation, and subsequently executes NuMagSANS to perform repeated simulations for each system size. The execution times are not provided as a separate dataset. Instead, for each simulation corresponding to a particle ensemble of size $N$, the complete NuMagSANS output is included. Each output directory contains a NuMagSANS log file with time-stamped entries covering the full execution from start to completion. The reported runtimes therefore correspond to the total wall-clock (gross) execution time without any post-processing or reduction. Using the provided scripts, the full benchmark workflow can be reproduced and executed on a local system. The data are intended to support reproducibility, performance assessment, and comparative benchmarking of large-scale micromagnetic SANS simulations.