Performance Evaluation of Mixed-Precision Runge-Kutta Methods
- Resource Type
- Conference
- Authors
- Burnett, Ben; Gottlieb, Sigal; Grant, Zachary J.; Heryudono, Alfa
- Source
- 2021 IEEE High Performance Extreme Computing Conference (HPEC) High Performance Extreme Computing Conference (HPEC), 2021 IEEE. :1-6 Sep, 2021
- Subject
- Communication, Networking and Broadcast Technologies
Computing and Processing
Performance evaluation
Energy consumption
Runtime
Additives
Codes
Conferences
Ordinary differential equations
Mixed-precision
multiprecision
Runge-Kutta
numerical methods
- Language
- ISSN
- 2643-1971
Additive Runge-Kutta methods designed for preserving highly accurate solutions in mixed-precision computation were proposed and analyzed in [8]. These specially designed methods use reduced precision for the implicit computations and full precision for the explicit computations. We develop a FORTRAN code to solve a nonlinear system of ordinary differential equations using the mixed precision additive Runge-Kutta (MP-ARK) methods on IBM POWER9 and Intel x86_64 chips. The convergence, accuracy, runtime, and energy consumption of these methods is explored. We show that these MP-ARK methods efficiently produce accurate solutions with significant reductions in runtime (and by extension energy consumption).