EcoSched jointly selects GPU counts via lightweight profiling and coschedules jobs with a score-based policy plus NUMA placement, delivering up to 14.8% energy savings, 30.1% makespan reduction, and 40.4% EDP improvement on H100/A100/V100 systems.
Improving gpu energy efficiency through an application-transparent frequency scaling policy with performance assurance
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.DC 2verdicts
UNVERDICTED 2representative citing papers
OPEN framework predicts performance in heterogeneous CPU-GPU systems with up to 98.29% accuracy by combining an offline-built predictor with online collaborative filtering to reduce profiling overhead.
citing papers explorer
-
Towards Energy Efficient Co-Scheduling in HPC
EcoSched jointly selects GPU counts via lightweight profiling and coschedules jobs with a score-based policy plus NUMA placement, delivering up to 14.8% energy savings, 30.1% makespan reduction, and 40.4% EDP improvement on H100/A100/V100 systems.
-
Coordinated Power Management on Heterogeneous Systems
OPEN framework predicts performance in heterogeneous CPU-GPU systems with up to 98.29% accuracy by combining an offline-built predictor with online collaborative filtering to reduce profiling overhead.