MFEM Workshop 2025 | A Guided Tour of MFEM GPU Kernel Optimization Techniques

Описание: The MFEM (Modular Finite Element Methods) project provides high-order mathematical calculations for large-scale scientific simulations. Held on September 10-11, 2025, the fifth annual MFEM community workshop brought together users and developers for a review of software features and the development roadmap, a showcase of technical talks and applications, student lightning talks, an interactive Q&A session, and a visualization contest.

LLNL collaborator John Camier’s presentation explores a variety of kernel optimization strategies for MFEM, leveraging its GPU abstraction layer to achieve high performance on AMD and NVIDIA GPUs. Focusing on the Cascadia application code, he detailed the journey toward optimal implementations, highlighting techniques to navigate compiler possibilities and to unlock architectural opportunities. The discussion provides practical insights for enhancing finite element computations in high-performance computing environments. Joint work with Veselin Dobrev (LLNL), Stefan Henneking (UT Austin), Tzanio Kolev (LLNL), and Jiqun Tu (NVIDIA).

Learn more about MFEM at https://mfem.org/. LLNL-VIDEO-2012298