APS Logo

Performance Boosting Portable Acceleration of SISSO++ for Symbolic Descriptor Learning

ORAL

Abstract

Sure-independence screening and sparsifying operator (SISSO) is a powerful, artificial intelligence tool for identifying symbolic descriptors and predictive models [1]. It has been successfully applied for discovering new optimal materials, e.g. [2]. Recently, we have developed SISSO++, a new implementation of SISSO that uses both OpenMP and MPI to approach linear-scaling parallel performance on CPUs [3]. Here, we present an updated algorithm that uses the Kokkos performance-portable programming model to offload the performance-critical region of our algorithm to accelerators, such as Nvidia or AMD GPUs [4]. We demonstrate the performance of these updates by using the prediction of thermal conductivity over rock salts and chalcopyrites as an example and highlight the opportunities opened by the improvement.



[1] R. Ouyang et al. Phys. Rev. Mater. 2, 083802 (2018).

[2] T. Purcell et al. submitted to npj Comput. Mater. arXiv:2204.12968 (2022).

[3] T. Purcell et al. Journal of Open Source Software 7, 3960 (2022).

[4] Ch. Trott et al. IEEE Transactions on Parallel and Distributed Systems 33, 805 (2022).

–

Presenters

  • Yi Yao

    The NOMAD Laboratory at the FHI-MPG and IRIS-Adlershof of HU, Berlin, Germany, The NOMAD Laboratory at the Fritz Haber Institute of the MPG

Authors

  • Yi Yao

    The NOMAD Laboratory at the FHI-MPG and IRIS-Adlershof of HU, Berlin, Germany, The NOMAD Laboratory at the Fritz Haber Institute of the MPG

  • Matthias Scheffler

    The NOMAD Laboratory at the FHI-MPG and IRIS-Adlershof of HU, Berlin, Germany, The NOMAD Laboratory at the Fritz Haber Institute of the MPG, The NOMAD Laboratory at the FHI of the Max-Planck-Gesellschaft and IRIS-Adlershof of the Humboldt-Universität zu Berlin, Germany

  • Christian Carbogno

    The NOMAD Laboratory at the FHI-MPG and IRIS-Adlershof of HU, Berlin, Germany, Fritz Haber Institute of the Max Planck Society, The NOMAD Laboratory at the FHI of the Max-Planck-Gesellschaft and IRIS-Adlershof of the Humboldt-Universität zu Berlin, Germany

  • Luca M Ghiringhelli

    1. The NOMAD Laboratory at the FHI-MPG and IRIS-Adlershof of HU, Berlin, Germany 2. Physics Department and IRIS-Adlershof of HU, Berlin, Germany, Physics Department and IRIS-Adlershof of HU, Berlin, Germany and The NOMAD Laboratory at the FHI-MPG and HU, Berlin, Germany, NOMAD Laboratory at the FHI of the Max-Planck-Gesellschaft and IRIS-Adlershof of the Humboldt-Universität (HU) zu Berlin; Physics Department and IRIS-Adlershof of HU zu Berlin

  • Thomas A Purcell

    The NOMAD Laboratory at the FHI-MPG and IRIS-Adlershof of HU, Berlin, Germany, The NOMAD Laboratory at the FHI of the MPG and IRIS-Adlershof of the Humboldt-Universität zu Berlin