Up to 50% Energy Optimization
for AI
& HPC Workloads
Power is expensive, finite, and heavily regulated. EAR - Energy Aware Runtime, is a system software that helps AI and HPC data centers achieve more performance per watt by optimizing energy use in real time.
- What we do
Unlock More Performance Per Watt
As HPC and AI workloads grow, so does their energy demand. Data centers need smarter tools to gain efficiency, not just in hardware, but in how systems run. EAR (Energy Aware Runtime) is a software that monitors, analyzes, and dynamically optimizes your infrastructure in real time. It connects workloads, servers, and cooling systems to gain energy you didn’t know you had.
Designed for AI factories and HPC data centers, EAR helps you:
Monitoring & Analysis
⇨ Power, Energy, CO2, Temperature at system and job level. ⇨ CPU and GPU Performance metrics ⇨ Hints for better resource utilization
Energy Optimization
⇨ Reduce Electricity Bill, OPEX, TCO and CO2. ⇨ Increase Performance per Watt, up to 50% of energy gains ⇨ Report energy saved and potential energy savings
Smart Power Capping
⇨ Maintain DC power/energy within a given limit. ⇨ Maximize Performance under a power limit ⇨ Eliminate power peaks
- Working process
Turn Efficiency into a Competitive Advantage
Designed for AI factories and HPC data centers, EAR helps you:
Save up to 50% energy
Reduce Total Cost of Ownership
Lower operational costs OPEX and extend hardware lifespan
Meet power budgets and environmental goals
Prevent overloads and power spikes
Get fast ROI: see savings from year one
Integrate easily: EAR works with Intel, AMD, ARM, NVIDIA, SLURM, K8s, and more
How EAR Works
Designed for AI factories and HPC data centers, EAR helps you:
Holistic Monitoring: CPU, GPU, I/O, memory, cooling
Data Telemetry and analysis
Automatic Optimization: Up to 20% better performance per watt
Energy Hints & Manual Tuning: Reach up to a total of 50% energy savings
Smart Power Capping: Prevent overloads and power spikes
Visual Dashboards: Track energy and performance KPIs
- Working process
Why Choose EAR?
Modular and vendor-agnostic (Intel, AMD, ARM, NVIDIA, Slurm, PBS…)
Transparent and non-intrusive
Live at runtime with no disruption to users
Helps you meet sustainability & ESG goals
Proven impact: up to 50% energy savings on real deployments
No code recompilation
Holistic and agnostic
- Case Studies
Case Studies
LRZ Super MUC-NG
Reducing peak power from 3 MW to 2 MW with minimal performance loss on one of Europe’s most efficient supercomputers.
SURF - Snellius
Partnering to explore software-driven energy efficiency for hybrid AI/HPC systems.
BSC – MareNostrum 5
Optimizing energy usage on a 314 PFlops pre-exascale system across CPU and GPU partitions.
EDF CRONOS
Gaining visibility and control of energy use by project, code, and user across 2 MW of internal HPC compute.
- A Team of Experts
Backed by Experts
EAR was co-developed with the Barcelona Supercomputing Center in 2016, and is already powering top European systems. We offer:
Installation & configuration
Training & advanced tools
Ongoing support & deep integration consulting
Trusted by Leading HPC Centers
Stay Updated
Get energy-saving insights, case studies, and updates from the EAR team.