As high-performance computing (HPC) scales up to the exascale era, the demand for massive computing power has never been higher. For projects like HiDALGO2, which relies on extreme-scale simulations to address critical global challenges—such as urban air quality, wildfire propagation, and renewable energy optimisation—computational performance is non-negotiable.
However, as our systems grow more powerful, managing their environmental footprint becomes equally vital.
In a thought-provoking article originally published by our partners at Bull (previously part of Atos), industry experts—including Pélagie Alves (Communication & Dissemination Leader for European HPC initiatives like EUPEX) alongside Franck Thomas, Donato Raimondi, and Michel Quaix—explore the crucial intersection of computing power, energy efficiency, and responsible resource management.
The Three Pillars of Sustainable HPC
The feature breaks down how the HPC ecosystem is actively shifting toward sustainability without sacrificing the raw performance needed for complex scientific simulations:
Next-Gen Infrastructure & Cooling: Traditional air cooling is no longer enough. The article highlights how Direct Liquid Cooling (DLC) using warm glycol water can capture heat right at the component level. This allows modern supercomputers to reach a Power Usage Effectiveness (PUE) close to 1.0, with the added benefit of recovering up to 100% of the waste heat to warm local district heating networks.
Aligning Power with Workload Usage: Energy efficiency is highly dependent on software behavior. By utilizing advanced telemetry tools (like BullSequana ARGOS), operators gain granular visibility into how resources are utilized, allowing them to balance raw execution time with actual energy consumption.
The Power of Code Optimization: Infrastructure is only half the battle; the software must be optimized for the hardware. The article spotlights the achievements of the Centre of Excellence for Performance and Programming (CEPP). Their optimization work on demanding physical simulation codes (such as the Gordon Bell Prize-winning WarpX plasma simulation) successfully reduced execution times by a factor of four, directly slashing the overall energy footprint.
Why This Matters for HiDALGO2
At HiDALGO2, we firmly believe that scientific excellence and environmental responsibility must go hand in hand. The code optimization and co-design principles outlined by the Bull team mirror our own efforts to benchmark, scale, and optimize global-challenge simulation pilots for the most energy-efficient execution possible on modern EuroHPC architectures.
“Managing the environmental footprint is an integral part of the design… demonstrating that high performance and scientific excellence can go hand in hand.”
Read the full article to discover the technical details behind the June 2026 Green500 rankings, advanced liquid cooling workflows, and the future of sustainable exascale computing.
