Skip to main content
Go to the home page of the European Commission (opens in new window)
English en
CORDIS - EU research results
CORDIS

Transparent Resource Utilisation for Scientific and Enterprise Workloads

Project description

Using up all the idle power in high-performance computing

As high-performance computing, data analytics and AI workloads grow more complex, much of today’s computing hardware remains underused due to inefficiencies in how applications share resources. This leads to wasted energy, higher costs and reduced performance across scientific and industrial systems. The EIC-funded TRU-STACK project will develop uHyper, a user-space hypervisor that allows multiple parallel computing applications to run concurrently on the same hardware without interfering with each other or requiring code changes. By coordinating how workloads share resources, it aims to reduce idle time, improve efficiency and significantly cut both execution time and energy consumption. Ultimately, uHyper will turn underused computing capacity into measurable performance and cost gains, boosting high-performance computing.

Objective

HPC, HPDA, and AI/ML workloads power modern science and industry, yet their growing complexity often leaves hardware underutilised. Real workloads interleave compute-heavy and serial phases, causing idle resources and inefficiencies. One way to unlock more parallelism is to execute multiple independent applications concurrently and fill these gaps. However, today’s operating systems and hypervisors lack mechanisms to coordinate multiple parallel runtimes safely on the same node. As a result, operators avoid co-location, sacrificing throughput, wasting energy, and inflating costs.

uHyper is a breakthrough user-space hypervisor that enables concurrent execution of parallel applications on shared nodes without requiring code changes. Positioned between the OS and runtimes such as MPI, OpenMP, or Python, uHyper orchestrates their execution to prevent interference and unlock latent performance. Early results consistently show more than 30% shorter execution times and at least 25% lower energy use compared to best-practice baselines.

These gains translate directly into economic impact. In public cloud environments, uHyper reduces billable instance-hours, while on-premise systems benefit from lower energy consumption (OPEX) and smaller infrastructure needs (CAPEX). Even modest adoption in the multi-billion-euro parallel workloads market generates significant value. As a software-only, drop-in solution, uHyper can be deployed rapidly and at scale.

By turning idle capacity into results, uHyper supports the European Green Deal by lowering energy per unit of work, strengthens digital sovereignty through a vendor-neutral orchestration layer, and enhances competitiveness in sectors such as pharmaceuticals, automotive, and advanced engineering. The project will culminate in a dedicated spin-off company to commercialise the technology through an open-core model and alliances with OEMs and integrators, ultimately delivering more science and more public value per euro.

Keywords

Programme(s)

Topic(s)

Funding Scheme

HORIZON-EIC - HORIZON EIC Grants

See all projects funded under this funding scheme

Call for proposal

(opens in new window) HORIZON-EIC-2025-TRANSITIONOPEN

See all projects funded under this call

Coordinator

BARCELONA SUPERCOMPUTING CENTER CENTRO NACIONAL DE SUPERCOMPUTACION
Net EU contribution
€ 1 026 833,28
Address
CALLE JORDI GIRONA 31
08034 BARCELONA
Spain
Region
Este Cataluña Barcelona
Activity type
Research Organisations
Links
Total cost
No data

Participants (1)