A Performance-Portable SYCL Implementation of CRK-HACC for Exascale

• DOI: 10.1145/3624062.3624187
• Publication Date: 2023-11-10T18:53:39Z
• Authors: Esteban Miguel Ra...; Simon John Pennycook; Adrian Pope; Nicholas Frontiere; Zhiqiang Ma; Varsha Madananth
• Keywords: Performance (cs.PF); FOS: Computer and information sciences; D.1.3; J.2; Computer Science - Performance; Cosmology and Nongalactic Astrophysics (astro-ph.CO); Computer Science - Distributed, Parallel, and Cluster Computing; FOS: Physical sciences; D.2.7; D.2.8; D.1.3; J.2; D.2.7; Distributed, Parallel, and Cluster Computing (cs.DC); D.2.8; Astrophysics - Cosmology and Nongalactic Astrophysics

The first generation of exascale systems will include a variety of machine architectures, featuring GPUs from multiple vendors. As a result, many developers are interested in adopting portable programming models to avoid maintaining multiple versions of their code. It is necessary to document experiences with such programming models to assist developers in understanding the advantages and disadvantages of different approaches. To this end, this paper evaluates the performance portability of a SYCL implementation of a large-scale cosmology application (CRK-HACC) running on GPUs from three different vendors: AMD, Intel, and NVIDIA. We detail the process of migrating the original code from CUDA to SYCL and show that specializing kernels for specific targets can greatly improve performance portability without significantly impacting programmer productivity. The SYCL version of CRK-HACC achieves a performance portability of 0.96 with a code divergence of almost 0, demonstrating that SYCL is a viable programming model for performance-portable applications.<br/>12 pages, 13 figures, 2023 International Workshop on Performance, Portability & Productivity in HPC<br/>