GPU offload on the tested system yields 4-12x throughput and up to 15x energy-efficiency gains over CPU-only execution for gromacs, lammps, OpenGadget3, AthenaK and dealii-X, with gains sensitive to problem granularity.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Node-Level Performance and Energy Characterization of Flagship Science Applications on SuperMUC-NG Phase 2
GPU offload on the tested system yields 4-12x throughput and up to 15x energy-efficiency gains over CPU-only execution for gromacs, lammps, OpenGadget3, AthenaK and dealii-X, with gains sensitive to problem granularity.