Session II: Experiments with Mixed Precision Algorithms in Linear Algebra

Low-precision floating-point arithmetic is a powerful tool for accelerating scientific computing applications, especially those in artificial intelligence. Here, we present an investigation showing that other high-performance computing (HPC) applications can also harness this power.  We consider solving a linear system Ax=b, with double precision A and b, by the use of an LU factorization computed in lower precision and mixed-precision iterative refinement.

Download Presentation

Location: Grand Ballroom C Date: March 27, 2019 Time: 12:10 pm - 12:30 pm Jack Dongarra, UTK/ORNL