New Advances in hypre 3.0 for Mixed Precision and Semi-Structured Problems
Inside Livermore Lab via YouTube
Google Data Analytics, IBM AI & Meta Marketing — All in One Subscription
Get 20% off all career paths from fullstack to AI
Overview
Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Explore the latest developments in the hypre 3.0 library through this 59-minute seminar presentation from the FEM@LLNL series. Learn about two major enhancements introduced in the September release: a comprehensive rewrite of the semi-structured code featuring a new semi-structured algebraic multigrid solver (SSAMG), and innovative mixed precision functionality that supports lower precision preconditioning of Krylov solvers. Discover how the semi-structured code overhaul extends functionality from square matrices to rectangular matrices with complete support for matrix-vector and matrix-matrix operations, enabling the implementation of SSAMG that leverages structure while building on existing PFMG and BoomerAMG solvers. Understand the mixed precision capabilities that allow runtime precision switching with minimal code changes and enable mixed-precision solver implementations. Examine performance results for the new solvers and implementations, gaining insights into how these advances enhance computational efficiency for finite element applications and scientific computing workflows.
Syllabus
FEM@LLNL | New Advances in hypre 3.0 for Mixed Precision and Semi-Structured Problems
Taught by
Inside Livermore Lab