AMPERE · Project

Software Tools That Cut Energy Use and Speed Up Parallel Computing Development

digitalPilotedTRL 6

Imagine you need to build software for a powerful computer that has many different types of processors working together — like coordinating a kitchen where every chef speaks a different language and uses different tools. Right now, programming these systems is painfully slow and error-prone, and getting them to be both fast and energy-efficient feels like trying to solve a Rubik's cube blindfolded. AMPERE built a smart toolbox that lets engineers describe what they need at a high level, and then automatically generates optimized code that runs efficiently across all those different processors. They proved it works in real automotive and railway systems where safety and timing are life-or-death matters.

By the numbers

consortium partners across the project

countries represented in the consortium

58%

industry participation ratio in the consortium

total deliverables produced

reference application domains validated (automotive and railway)

demo deliverables including ecosystem releases and runtime components

The business problem

What needed solving

Programming modern parallel processors is extremely complex and expensive, especially when you need the software to be simultaneously fast, energy-efficient, safe, and reliable. Automotive and railway companies face mounting pressure to deploy compute-heavy applications on heterogeneous hardware while meeting strict certification requirements, but their development tools were not built for this level of multi-criteria optimization.

The solution

What was built

AMPERE built a complete model-driven development ecosystem that automatically generates optimized parallel code from high-level system descriptions. The final release includes code synthesis tools, runtime libraries with dynamic energy management, scheduling algorithms for predictable computation, segregation and resilience mechanisms, and integrated GPOS/RTOS/PikeOS hypervisor support — all producing 39 deliverables validated in automotive and railway applications.

Audience

Who needs this

Tier-1 automotive suppliers developing ADAS and autonomous driving platformsRailway signalling companies building safety-critical control systemsEmbedded systems integrators working with heterogeneous multi-core processorsAerospace and defense contractors needing certified parallel computing solutionsIndustrial automation companies deploying real-time edge computing

Business applications

Who can put this to work

Automotive embedded systems

enterprise

Target: Tier-1 automotive suppliers developing ADAS or autonomous driving software

If you are an automotive supplier struggling with the complexity of programming heterogeneous computing platforms for advanced driver assistance — this project developed a complete model-driven ecosystem that automatically generates parallel code from high-level system descriptions, validated in automotive use cases. It addresses performance, energy efficiency, real-time response, and safety constraints simultaneously, reducing the manual effort of porting software across different processor architectures.

Railway signalling and control

enterprise

Target: Railway technology companies building safety-critical onboard or trackside systems

If you are a railway technology provider dealing with the cost and complexity of certifying software on modern parallel hardware — AMPERE delivered operating system and hypervisor support (including PikeOS) with built-in segregation and resilience mechanisms, validated in railway applications. The correct-by-construction approach from 2 reference domains means fewer costly verification cycles when deploying on heterogeneous platforms.

Industrial edge computing

mid-size

Target: Companies deploying compute-intensive applications on energy-constrained edge devices

If you are deploying AI or signal processing workloads at the edge and battling energy budgets — AMPERE built dynamic energy management policies and runtime reconfiguration that balance performance against power consumption on parallel heterogeneous hardware. The 39 deliverables include ready-to-use runtime libraries with monitoring capabilities that optimize energy use without sacrificing real-time guarantees.

Frequently asked

Quick answers

What would it cost to adopt these tools?

The project was publicly funded as an RIA (Research and Innovation Action), so core research results are typically available under open or favorable licensing terms. Specific commercial licensing for integrated components like the PikeOS hypervisor would depend on the partner SYSGO. Contact the consortium for exact pricing.

Can this scale to our production environment?

AMPERE delivered a final integrated ecosystem validated in 2 reference applications from automotive and railway — both demanding industrial domains. The 12-partner consortium with 58% industry participation suggests the tools were designed with production constraints in mind, though deployment readiness will vary by component.

What is the IP and licensing situation?

As an EU-funded RIA project, results follow Horizon 2020 IP rules where each partner owns their foreground IP. The consortium includes 7 industry partners who likely hold commercial rights to their specific contributions. Access terms should be negotiated directly with individual partners or through the coordinator.

How does this integrate with our existing development workflow?

The AMPERE ecosystem is built around model-driven engineering, meaning engineers describe system requirements in domain-specific modeling languages that then feed into automated code synthesis. Based on deliverable descriptions, the toolchain includes OS-level support (GPOS, RTOS, PikeOS hypervisor) and runtime libraries, suggesting integration at both the development and deployment stages.

What safety and certification standards does this support?

The project explicitly addresses safety through segregation mechanisms and correct-by-construction code generation, validated in automotive and railway domains that fall under standards like ISO 26262 and EN 50128. The PikeOS hypervisor component from the consortium is already certified in safety-critical markets.

What is the technology maturity level?

AMPERE delivered a final release of the complete ecosystem with integrated runtime energy support, operating systems, and hypervisor — all validated in 2 industrial use cases. Based on available project data, this places the technology between demonstration and pre-commercial stage, with some components closer to market than others.

Is ongoing support available?

The project closed in June 2023, but 7 industry partners across 7 countries remain active in their respective markets. The Barcelona Supercomputing Center as coordinator can direct inquiries to the right partner. Continued development likely depends on each partner's commercial roadmap.

Consortium

Who built it

The AMPERE consortium is a strong 12-partner team spanning 7 European countries (Spain, France, Germany, Italy, Portugal, Switzerland, Czech Republic), led by the Barcelona Supercomputing Center — one of Europe's top HPC institutions. With 58% industry participation (7 out of 12 partners), this is not an academic exercise: the industrial partners brought real-world constraints from automotive and railway sectors. The remaining 4 universities and 1 research organization provided the scientific depth in model-driven engineering and parallel computing. Notably, zero SMEs participated, which suggests the technology targets established companies with complex embedded system needs rather than startup-scale operations.

BARCELONA SUPERCOMPUTING CENTER CENTRO NACIONAL DE SUPERCOMPUTACIONCoordinator · ES
SYSGO SROparticipant · CZ
EVIDENCE SRLparticipant · IT
UNIVERSITA DEGLI STUDI DI SIENAthirdparty · IT
ROBERT BOSCH GMBHparticipant · DE
THALES ITALIA SPAparticipant · IT
SYSGO GMBHthirdparty · DE
INSTITUTO SUPERIOR DE ENGENHARIA DO PORTOparticipant · PT
SCUOLA SUPERIORE DI STUDI UNIVERSITARI E DI PERFEZIONAMENTO S ANNAparticipant · IT
HITACHI RAIL GTS ITALIA SRLparticipant · IT
EIDGENOESSISCHE TECHNISCHE HOCHSCHULE ZUERICHparticipant · CH
THALESparticipant · FR

How to reach the team

Barcelona Supercomputing Center (BSC-CNS), Spain — reach out to their technology transfer office or the AMPERE project lead

Order a report on this consortium See BARCELONA SUPERCOMPUTING CENTER CENTRO NACIONAL DE SUPERCOMPUTACION profile →

Next steps

Talk to the team behind this work.

Want an introduction to the AMPERE team for licensing their parallel computing tools or exploring a pilot? SciTransfer can connect you directly with the right partner for your use case.

Order a report on this topic More in Digital & ICT