.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
Minisymposium Presentation
Numerical Optimization Targeting Energy-Efficient Scientific Computing
Presenter
Roman Iakymchuk is an Associate Professor and a docent (equiv to habilitation) at the Division of Scientific Computing, Department of Information Technology, Uppsala University (UU), Sweden. He is also a part-time Associate Professor at the Department of Computing Science at Umeå University (UmU), Sweden. Roman is co-Principal Investigator of the EuroHPC JU Center of Excellence in Exascale CFD (CEEC) and leads a work package on Exascale Algorithms. He was a part of three other EU-funded projects such as EPEEC, Intertwine, and AllScale and led work packages on the latter two projects. Furthermore, Roman was granted a prestigious Marie Sklodowska Curie Actions Individual Fellowship for the Robust project at Sorbonne University, France. He conducts his research on energy-efficient and reliable numerical solvers, and their use in applications.
Description
Mixed-precision computing has the potential to significantly reduce the cost of exascale computations, but determining when and how to implement it in programs can be challenging.We propose a methodology for enabling mixed-precision with the help of computer arithmetic tools, roofline model, and computer arithmetic techniques. As case studies, we consider Nekbone, a mini-application for the Computational Fluid Dynamics (CFD) solver Nek5000, and a modern Neko CFD application. With the help of the VerifiCarlo tool and computer arithmetic techniques, we introduce a strategy to address stagnation issues in the preconditioned Conjugate Gradient method in Nekbone and apply these insights to implement a mixed-precision version of Neko. We evaluate the derived mixed-precision versions of these codes by combining metrics in three dimensions: accuracy, time-to-solution, and energy-to-solution. Notably, mixed-precision in Nekbone reduces time-to-solution by roughly 38% and energy-to-solution by 2.8x on MareNostrum 5, while in the real-world Neko application the gain is up to 29% in time and up to 24% in energy, without sacrificing the accuracy.