An analytical model for GPU power identifies architecture- and workload-dependent transition frequencies beyond which energy efficiency degrades, based on benchmarks on A40/A100/H100/H200 GPUs.
In: 2013 International Green Computing Confer- ence Proceedings
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
-
Modeling and Chasing the Energy-Efficiency Sweet Spots in Modern GPUs
An analytical model for GPU power identifies architecture- and workload-dependent transition frequencies beyond which energy efficiency degrades, based on benchmarks on A40/A100/H100/H200 GPUs.