SYCL compute kernels for ExaHyPE

Loi, Chung Ming; Bockhorst, Heinrich; Weinzierl, Tobias

Abstract

We discuss three SYCL realisations of a simple Finite Volume scheme over multiple Cartesian patches. The realisation flavours differ in the way how they map the compute steps onto loops and tasks: We compare an implementation that is exclusively using a sequence of for-loops to a version that uses nested parallelism, and finally benchmark these against a version modelling the calculations as task graph. Our work proposes realisation idioms to realise these flavours within SYCL. The results suggest that a mixture of classic task and data parallelism performs if we map this hybrid onto a solely data-parallel SYCL implementation, taking into account SYCL specifics and the problem size.

Citation

Loi, C. M., Bockhorst, H., & Weinzierl, T. (2024, March). SYCL compute kernels for ExaHyPE. Presented at 2024 SIAM Conference on Parallel Processing for Scientific Computing (PP), Baltimore, MD

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2024 SIAM Conference on Parallel Processing for Scientific Computing (PP)
• Start Date: Mar 5, 2024
• End Date: Mar 8, 2024
• Acceptance Date: Dec 8, 2023
• Publication Date: 2024
• Deposit Date: Dec 11, 2023
• Publicly Available Date: Dec 31, 2024
• Publisher: Society for Industrial and Applied Mathematics
• Peer Reviewed: Peer Reviewed
• Pages: 90-103
• Book Title: Proceedings of the 2024 SIAM Conference on Parallel Processing for Scientific Computing (PP)
• ISBN: 9781611977967
• DOI: https://doi.org/10.1137/1.9781611977967.8
• Public URL: https://durham-repository.worktribe.com/output/2022863
• Related Public URLs: https://arxiv.org/abs/2306.16731