The European Excessive-Efficiency Computing Joint Endeavor (EuroHPC JU) has introduced a significant new funding name to reshape how high-performance computing (HPC) methods are benchmarked and evaluated throughout Europe.
Backed by a €4m funds beneath the Horizon Europe programme, the initiative seeks to ascertain sturdy, clear, and future-proof benchmarking practices that may hold tempo with quickly evolving computing applied sciences.
Open for submissions until 24 March 2026 at 17:00 CET, the decision invitations researchers, business gamers, and know-how organisations to contribute to a brand new era of benchmarking frameworks for each classical and rising hybrid computing environments.
The significance of high-performance computing
Excessive-performance computing refers to using highly effective computing methods able to processing huge quantities of information and performing complicated calculations at extraordinarily excessive speeds.
HPC methods are important for duties that exceed the capabilities of typical computer systems, similar to local weather modelling, drug discovery, aerospace simulation, synthetic intelligence coaching, and superior supplies analysis.
As Europe invests closely in exascale computing and AI-driven workloads, the potential of HPC continues to develop.
Effectively-designed benchmarking ensures that this potential is absolutely realised by serving to customers match the correct methods to the correct purposes, whereas additionally encouraging innovation in energy-efficient and sustainable computing architectures.
Constructing a unified benchmarking normal for HPC methods
On the coronary heart of the decision is the ambition to create a unified, modular, and extensible benchmarking framework for HPC methods.
EuroHPC JU goals to interchange fragmented approaches with a well-documented, constantly up to date suite of benchmarks that may be broadly adopted throughout Europe.
Benchmarking performs a essential position in high-performance computing, enabling truthful and repeatable comparisons of {hardware} and software program platforms.
A typical framework will enable customers to validate efficiency claims, evaluate methods objectively, and make knowledgeable selections based mostly on real-world workloads fairly than remoted metrics.
For system operators, the initiative additionally helps the optimisation of throughput and power effectivity – two priorities which might be changing into more and more vital as HPC methods scale towards exascale efficiency.
Focus space one: Benchmarking exascale and HPC-AI methods
The primary matter beneath the decision, A European HPC-centric Benchmarking Framework, targets exascale computing and HPC-AI convergence.
Proposals are anticipated to outline and standardise benchmarks that allow reproducible, repeatable, and replicable efficiency evaluations throughout completely different HPC methods.
Key targets embrace the event of fine-grained efficiency metrics, significantly these associated to power effectivity, in addition to the standardisation of benchmarking inputs and outputs.
Candidates should additionally design a structured workflow that captures the complete benchmarking lifecycle – from deployment to knowledge assortment and evaluation.
Focus space two: Benchmarks for hybrid quantum-classical computing
The second matter shifts consideration to the longer term, addressing benchmarking for hybrid quantum-classical computing.
As quantum applied sciences mature, there may be rising curiosity in integrating quantum processors with conventional HPC methods.
Proposals on this space ought to outline hardware-agnostic benchmarks that measure throughput, latency, power consumption, and the depth of integration between HPC and quantum computing.
Candidates are additionally anticipated to stipulate a strategic roadmap that anticipates technological advances and helps various qubit modalities.
Strengthening Europe’s HPC ecosystem
By investing in superior benchmarking frameworks, EuroHPC JU is reinforcing Europe’s management in HPC methods whereas guaranteeing transparency, comparability, and sustainability.
The decision represents a big alternative to form how efficiency is measured – not simply at this time, however within the hybrid and quantum-enabled HPC environments of tomorrow.
