A high-performance and portable implementation of the SISSO method for CPUs and GPUs
Eibl S, Yao Y, Scheffler M, Rampp M, Ghiringhelli LM, Purcell TA (2025)
Publication Type: Journal article
Publication year: 2025
Journal
Book Volume: 6
Article Number: 047001
Journal Issue: 4
Abstract
Sure-independence screening and sparsifying operator (SISSO) is an artificial intelligence (AI) method based on symbolic regression and compressed sensing widely used in materials science research. SISSO++ is its C++ implementation that employs MPI and OpenMP for parallelization, rendering it well-suited for high-performance computing (HPC) environments. As heterogeneous hardware becomes mainstream in the HPC and AI fields, we chose to port the SISSO++ code to GPUs using the Kokkos performance-portable library. Kokkos allows us to maintain a single codebase for both Nvidia and AMD GPUs, significantly reducing the maintenance effort. In this work, we summarize the necessary code changes we did to achieve hardware and performance portability. This is accompanied by performance benchmarks on Nvidia and AMD GPUs. We demonstrate the speedups obtained from using GPUs across the three most time-consuming parts of our code.
Authors with CRIS profile
Involved external institutions
Rechenzentrum Garching der Max-Planck-Gesellschaft (MPCDF) / Max Planck Computing & Data Facility
Germany (DE)
Fritz-Haber-Institut der Max-Planck-Gesellschaft (FHI)
Germany (DE)
Germany (DE)
Fritz-Haber-Institut der Max-Planck-Gesellschaft (FHI)
Germany (DE)
How to cite
APA:
Eibl, S., Yao, Y., Scheffler, M., Rampp, M., Ghiringhelli, L.M., & Purcell, T.A. (2025). A high-performance and portable implementation of the SISSO method for CPUs and GPUs. Machine Learning: Science and Technology, 6(4). https://doi.org/10.1088/2632-2153/ae0ab3
MLA:
Eibl, Sebastian, et al. "A high-performance and portable implementation of the SISSO method for CPUs and GPUs." Machine Learning: Science and Technology 6.4 (2025).
BibTeX: Download