News from the field
Transcription
News from the field
ITEA2: a potential framework for developing industrial HPC applications and infrastructures Gérard Roucairol - Vice-Chairman ITEA 2 ITEA2: a EUREKA ICT Cluster Success Patterns ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• • Bottom up : Vision, Strategic Research Agenda & Roadmap • Core Group of Global Players • Large Community of SMEs, Universities and Academic Research Institutes • Well defined Organisation (Industry // Public Authorities): Board//DC, Board Support Group//AC, Steering Group, Office • Well defined Processes: Calls, POs, FPPs, Labelling, Funding, Reviews, Exhibition & Award Competition ITEA 2 - 2 ITEA 2 Symposium 2006 The initiators of ITEA 2 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• F B I FIN D NL E • The ITEA 2 Board members are: Airbus, Alcatel, Barco, Bosch, Bull, DaimlerChrysler, European Federation of High Tech SMEs, Nokia, Philips, Siemens, Telvent, Thales and Thomson. • The combined 2004 totals of this group are: – Total turn-over: €380 billion – Total R&D spend: €29 billion – Total no. of employees: 1.5+ million of which 210.000 in R&D (12% of Europe’s total number of researchers) ITEA 2 - 3 ITEA 2 Symposium 2006 There’s growing interest in ITEA Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• Person-years over time Project Outline Proposals (in person-years) 1600 1432 1400 1272 1261 1200 1581 1164 1062 1096 5969 913 1000 800 3094 600 400 212 200 0 1999 2000 2001 2002 2003 2004 2005 2006 2007 Forecast September 2006 ITEA 2 - 4 ITEA 2 Symposium 2006 Growing ITEA community Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• 25 countries countries 25 540 partners partners in in 86 86 projects projects 540 26% (20%) 28% (64%) 46% (16%) SME Research institutes & universities Large industries ( ) : effort Status September 2006 ITEA 2 - 5 ITEA in numbers ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• Effort top countries Exploitation 94% total 450 France 31 % Netherlands Spain Finland Germany Belgium Italy 19 % 11 % 10 % 9% 8% 7% No. of expected product references 185 No. of expected OEM references No. of expected results used for internal purposes Standardisation 150 120 No. of expected licenses to be sold by partners 60 No. of open source 30 No. of standardisation actions launched 20 No. of actions in progress 100 No. of actions issued or published Dissemination 55 30 Publications & Conferences 1,650 Project websites 55 ITEA 2 - 6 ITEA 2 Symposium 2006 Vision Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• ‘Europe to maintain leadership in Software-intensive Systems and Services building on key European strengths and industries’ Solutions / Services Differentiation Software Communications Infrastructure Real-Time Infrastructure Recognising and responding to: • the move from a product-oriented to a services-oriented world • ‘time-to-market’, i.e. INNOVATION, is make or break • the even more crucial role of SMEs and Academia • strong ERA cooperation being mandatory (seizing FP7 opportunity) • digital convergence is name of the game ITEA 2 - 7 ITEA 2 Symposium 2006 Executing the vision: the roadmap Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• Domains Cyber Enterprise Home Services & Software Creation Intermediation Services & Infrastructures Nomadic Scenarios Scripts Scenario: Knowledge Management Scenario: The Connected Home Aspect: The fully and seamlessly self-configured Connected Home Selective interoperability Dynamic allocation of functions "Follow me" at presentation High bandwidth access network Simple query Global p-2-p Adaptation to network and device characteristics Integrated remote control Medium- Term Internal ad-hoc networking Plug & play of appliances Knowledge retrieval and di ffusion according to context Wireless Networks Early protocols Intelligent, self organising adaptive KM system embedding the user Advanced neararea networks Collaboration Mechanisms Privacy and security techniques Enabling technologies outside ITE A Stored knowledge is entirely trustworthy/ certified Short- Term Long- T erm Network of functionalities Power saving techniques Partly automated electronic capture and retrieval of knowledge. Context and profile-aware specific KM Streaming to the world Symmetric broadband Short- Term Low-level collaboration Capturing knowledge from human behaviours. Semantic analysis tools Standardisation to visualise existing and needed KM Knowledge capture from business processes and external sources (internet) "Follow me" of applications Handover in home Scenario: Area Networks Wireless interconnection Aspect: Scenario: Service composition External ad-hoc networking Islands of dynamically composed friendly (i.e. easy to use) services Services built individually by human integration Plug & play of resources Smart agents Ambient KM embedding the user Network of appliances +Remote video High bandwidth wireless HN Software upgrades Partially automated KM for specific domains Knowledge ontologies, KM standardisation Capture of information from legacy support Secure DRM solution Limited DRM solution Local video Remote audio & pictures Intergrated system/human, automated and standardised KM Dynamically composed services, negotiation based on use r, context and service profiles Short- T erm Sensors and actuator technologies Medium- T erm Clusters of friendly services Pre-configured portals Network of autonomous cooperative actors Open system architectures Services ontology (functional and non-functional) Software evolution infrastructure Serve r, client & agents containers Ontology for profiles and context Data mining and profiling Context-aware multimedia search Semantic search engines Technologies to fulfil legal requirements Learning and adaptation technologies Self-organisation techniques Medium- Component paradigm evolution Development time documentation (Abstract component description) Component Context-aware and personalised intelligent services Self-organising distributed peer-to-peer infrastructure Fusion of data from profiles with recognition of user's habit Term Deployable Evolvable Components Abstraction Evolution Component based methodologies Kobra RU P UM L 2 Standards Domain standardisation Notation systems Domain analysis machine readable notations XML/XMI Automatic setup Semantic mappings Long- Term Version management Long- Term Short- Term Medium- Term Standard components Deploytime code (setup code) Intelligent brokering adapted to user and context Services Certification Short- Term Components & software supplier market Runtime Binaries (Plug & play) Agent swarms Dynamic interoperability Plug-&play Scenario: Nomadic Partly automated KM + fusion/crossfertilisation from di f ferent domains Smart appliances Ad-hoc limited KM Function registration and discovery Wired & wireless home network Dedicated control Dynamic allocation of functions Networked appliances Integrated KM is an ad-hoc information system repository + independent human resources management. No standardised methodologies Dynamic reconfiguration Selective interoperability Content exchange Medium- Term Long- Term Long- Term Technologies List Clusters Network Services 56 ContentAcquisition& Processing 30 Now ST MT Rendezvous of all the object in the environment Rendezvous domainspeci Þc generic Regulation and standards High-precision open-space localisation Localisation in buildings New or extended intelligent sensors ferent sensors to information/knowledge domainspeci Þc GPS Individual Synthetic Triangulation in GSM/UMTS Everywhere, incl. in-door In a local network In a WAN network Home and enterprise Home and enterprise Adaptive knowledge based auto-configuration Agent-based configuration of network elements Rendezvous Security and privacy Dynamic filtering and transformation for adaptation to Identity management and profiling Speci Þc and management (e.g. groups and communities) Pattern matching of media data Media interpretation algorithms off-line near real-time real-time off-line near real-time real-time Identity management Standardised ontologies and construction of proÞles from user habits Monolithic, proprietary Users ’ personal patterns of behaviour when on-line established through ‘avatars ficient compression algorithms / bit-rate reduction (transmission time (isochronous – asynchronous), cost, ’59 V oice commands, noisy environments in speci Þc domains environments in multiple domains Enhanced natural language understanding, speaker independence Games H.264 Monotonic Natural-sounding with intonation Understanding content, some emotions Monitoring, directions (games) Interpretation, pointing Facial movements, some expressions Eye movement (to control applications); eye-ball tracking Calibration, focus control Understand some emotions Multi-user interfaces; v 2D and 3D symbolic (games and simulators) Full 3D and context support irtual and augmented reality Usability engineering Rendezvous Compression with scalable complexity for optimising coding Legal aspects for inter-operation (e.g. CPU resources, and power) Generating knowledge from data Learning user interfaces Work ßow Pro Þles within closed or Context ProÞle Privacy and security Advanced web services allowing secure and transactional coordination between applications Support for multi-display/device/HSI systems End-to-end Support for accountable events Þles) Simple (presence, location) Groups, more context, detailed domainspeci Þc Adapting menus, mainly single users, simple collaboration Full context and situation, some emotions Multi-diverse environment Understand some human behaviour Basic Multi-diverse environments, trust aspects Full support Basic Session roaming, context support, complex systems Full adaptation and support Basic Multi-diverse environment Te chnologies for Human-System Interaction / generic T echnologies for Network Services / Derive high-level from low-level knowledge or data W eb-based development environment Full support T able 4-1 Tool support Open source best practice 81 Distributed and collaborative engineering Distributed engineering support technologies Shared work space E.g. Groove Concurrent engineering support technologies Version management, E.g. CVS, GForce Collaborative decision making, groupware solutions Proprietary Integration of non-software engineering disciplines Cyber rooms and processes Configurable methodologies and process standards Executable process definitions Multi-diverse environment presence Seamless and interchangeable HSIs Multi-device HSIs Data sharing based on object management methodologies Methodologies for per-to-peer engineering Rendezvous LT Data sharing Complete data transfer Object management with versioning of schemas E.g. Eclipse Con Þgurable, E.g. RU P Domain-specific CAD technologies MT E.g. XMI Based on Þle paradigm Tool integration frameworks and platforms Configurable, extensible and modular engineering Configurable workflows Þles ST E.g. MOF Standardised tool data transfer formats Integrated navigation and linking of engineering elements, including data management Disruptions for various environments Accountable event generation and collection fic data) Now Integration and interoperation of engineering tools Common meta-model for engineering artefacts Disruptions Dynamic and roamable pro proprietary environments General support for application mobility Profile and network context-based service orchestration Disruptions Evolution of current systems Evolution of current systems Intelligent, context-aware and adaptive HSIs User profiles Context-awareness (with sensors and pro Service-level agreements (SLA) Self-adapting learning algorithms for content & context Descriptive coding of context evolution in space and time 3-D realistic, other domains Evolution of current approaches Platforms for HSI development HSI usability test systems Generic 80 Domain-specific CAD technologies interpretation, understanding content, some emotions Gestures (to control applications) Brokering of services parameters with respect to overall performances of the transmission channel and terminal node capabilities Natural language understanding, application related semantic Text to speech Service coordination H.261/H.263 space, quality (perceived quality - full integrity) Behaviour prediction based on actual and historical data Fusion of proÞles across applications Federated Individual to-URN resolution Universal identification (e.g. camera images, position and digital maps) Engineering process support LT Rendezvous Voice commands noisy WAN WAN Standardised device interface (sensors / actuators) and 58 application model Profiles for users and user clusters session context Integration of information Algorithms for media integration Real-time projection algorithms for full windscreen projection (e.g. derived delay times from traf MT Speech recognition, speech to text Synthetic Service-based self-managing devices context fusion; integration of context Efficient analysis of data Rendezvous ST Simple, self-explaining and easy-to-use multimodal HSIs Automatic resource discovery in a network generic Context capturing and management Ef ficient & standardised context exchange (e.g. user profiles) standarisation Now LT Individual Manually con Þgured Seamless distributed network capabilities Sensor fusion to integrate raw physical data from dif Software architectures of collaborative sensory systems Infrastructure, MT T ransparent access to distributed data (individual sensor / actuator data or synthetic data) within the local network (e.g. in home, building, plants) Remote access to local systems and services (individual or synthetic) 57 Positioning Low-cost location positioning with increasing precision More ef ST Ambient intelligence Standardised exchange of positioning information Smart tags will allow identification Human-System Interaction 67 Now LT Digital Sensory System Challenge Process description language Executable Tool con Þguration for process support Requirements driven process management Requirements management and change management Requirements linking and tracking E.g. DOORS Requirements re-use techniques and guidelines Requirements based measurement Requirements, meta-models and specification languages From document-centred to object-oriented Re-use by reference Template Concept of uncertain requirements Requirements verification and validation Formalised description Formalised description Requirements based code generation techniques, test generations Knowledge based engineering Ontology , standardised knowledge models, taxonomies 82 Data mining, software mining and profiling Speci Þc De Þnitions and matching algorithms Focused and adapted generic methods, e.g. pattern matching based on DN A analysis Ta ble 3-2 T echnologies for engineering process support / T able 5-3 (e.g. enhanced route-finding algorithms, taking into account additional attributes such as user preferences and security aspects) T echnologies for Content Acquisition & Processing / Table 2-1 Rendezvous Content Infrastructures & Basic Services Human-System Interaction Engineering ITEA 2 - 8 Welcoming industrial HPC applications and infrastructures ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• Domains Cyber Enterprise Home Services & Software Creation Intermediation Services & Infrastructures Nomadic Scenarios Scripts Scenario: Knowledge Management Scenario: The Connected Home Aspect: The fully and seamlessly self-configured Connected Home Selective interoperability Dynamic allocation of functions "Follow me" at presentation High bandwidth access network Simple query Global p-2-p Adaptation to network and device characteristics Integrated remote control Medium- Term Internal ad-hoc networking Plug & play of appliances Knowledge retrieval and di ffusion according to context Wireless Networks Early protocols Intelligent, self organising adaptive KM system embedding the user Advanced neararea networks Collaboration Mechanisms Privacy and security techniques Enabling technologies outside ITE A Stored knowledge is entirely trustworthy/ certified Short- Term Long- T erm Network of functionalities Power saving techniques Partly automated electronic capture and retrieval of knowledge. Context and profile-aware specific KM Streaming to the world Symmetric broadband Short- Term Low-level collaboration Capturing knowledge from human behaviours. Semantic analysis tools Standardisation to visualise existing and needed KM Knowledge capture from business processes and external sources (internet) "Follow me" of applications Handover in home Scenario: Area Networks Wireless interconnection Aspect: Scenario: Service composition External ad-hoc networking Islands of dynamically composed friendly (i.e. easy to use) services Services built individually by human integration Plug & play of resources Smart agents Ambient KM embedding the user Network of appliances +Remote video High bandwidth wireless HN Software upgrades Partially automated KM for specific domains Knowledge ontologies, KM standardisation Capture of information from legacy support Secure DRM solution Limited DRM solution Local video Remote audio & pictures Intergrated system/human, automated and standardised KM Dynamically composed services, negotiation based on use r, context and service profiles Short- T erm Sensors and actuator technologies Medium- T erm Clusters of friendly services Pre-configured portals Network of autonomous cooperative actors Open system architectures Services ontology (functional and non-functional) Software evolution infrastructure Serve r, client & agents containers Ontology for profiles and context Data mining and profiling Context-aware multimedia search Semantic search engines Technologies to fulfil legal requirements Learning and adaptation technologies Self-organisation techniques Medium- Component paradigm evolution Development time documentation (Abstract component description) Component Context-aware and personalised intelligent services Self-organising distributed peer-to-peer infrastructure Fusion of data from profiles with recognition of user's habit Term Deployable Evolvable Components Abstraction Evolution Component based methodologies Kobra RU P UM L 2 Standards Domain standardisation Notation systems Domain analysis machine readable notations XML/XMI Automatic setup Semantic mappings Long- Term Version management Long- Term Short- Term Medium- Term Standard components Deploytime code (setup code) Intelligent brokering adapted to user and context Services Certification Short- Term Components & software supplier market Runtime Binaries (Plug & play) Agent swarms Dynamic interoperability Plug-&play Scenario: Nomadic Partly automated KM + fusion/crossfertilisation from di f ferent domains Smart appliances Ad-hoc limited KM Function registration and discovery Wired & wireless home network Dedicated control Dynamic allocation of functions Networked appliances Integrated KM is an ad-hoc information system repository + independent human resources management. No standardised methodologies Dynamic reconfiguration Selective interoperability Content exchange Medium- Term Long- Term Long- Term Technologies List Clusters Network Services 56 ContentAcquisition& Processing 30 Now ST MT Rendezvous of all the object in the environment Rendezvous domainspeci Þc generic Regulation and standards High-precision open-space localisation Localisation in buildings New or extended intelligent sensors ferent sensors to information/knowledge domainspeci Þc GPS Individual Synthetic Triangulation in GSM/UMTS Everywhere, incl. in-door In a local network In a WAN network Home and enterprise Home and enterprise Adaptive knowledge based auto-configuration Agent-based configuration of network elements Rendezvous Security and privacy Dynamic filtering and transformation for adaptation to Identity management and profiling Speci Þc and management (e.g. groups and communities) Pattern matching of media data Media interpretation algorithms off-line near real-time real-time off-line near real-time real-time Identity management Standardised ontologies and construction of proÞles from user habits Monolithic, proprietary Users ’ personal patterns of behaviour when on-line established through ‘avatars ficient compression algorithms / bit-rate reduction (transmission time (isochronous – asynchronous), cost, ’59 V oice commands, noisy environments in speci Þc domains environments in multiple domains Enhanced natural language understanding, speaker independence Games H.264 Monotonic Natural-sounding with intonation Understanding content, some emotions Monitoring, directions (games) Interpretation, pointing Facial movements, some expressions Eye movement (to control applications); eye-ball tracking Calibration, focus control Understand some emotions Multi-user interfaces; v 2D and 3D symbolic (games and simulators) Full 3D and context support irtual and augmented reality Usability engineering Rendezvous Compression with scalable complexity for optimising coding Legal aspects for inter-operation (e.g. CPU resources, and power) Generating knowledge from data Learning user interfaces Work ßow Pro Þles within closed or Context ProÞle Privacy and security Advanced web services allowing secure and transactional coordination between applications Support for multi-display/device/HSI systems End-to-end Support for accountable events Þles) Simple (presence, location) Groups, more context, detailed domainspeci Þc Adapting menus, mainly single users, simple collaboration Full context and situation, some emotions Multi-diverse environment Understand some human behaviour Basic Multi-diverse environments, trust aspects Full support Basic Session roaming, context support, complex systems Full adaptation and support Basic Multi-diverse environment Te chnologies for Human-System Interaction / generic T echnologies for Network Services / Derive high-level from low-level knowledge or data W eb-based development environment Full support T able 4-1 Tool support Open source best practice 81 Distributed and collaborative engineering Distributed engineering support technologies Shared work space E.g. Groove Concurrent engineering support technologies Version management, E.g. CVS, GForce Collaborative decision making, groupware solutions Proprietary Integration of non-software engineering disciplines Cyber rooms and processes Configurable methodologies and process standards Executable process definitions Multi-diverse environment presence Seamless and interchangeable HSIs Multi-device HSIs Data sharing based on object management methodologies Methodologies for per-to-peer engineering Rendezvous LT Data sharing Complete data transfer Object management with versioning of schemas E.g. Eclipse Con Þgurable, E.g. RU P Domain-specific CAD technologies MT E.g. XMI Based on Þle paradigm Tool integration frameworks and platforms Configurable, extensible and modular engineering Configurable workflows Þles ST E.g. MOF Standardised tool data transfer formats Integrated navigation and linking of engineering elements, including data management Disruptions for various environments Accountable event generation and collection fic data) Now Integration and interoperation of engineering tools Common meta-model for engineering artefacts Disruptions Dynamic and roamable pro proprietary environments General support for application mobility Profile and network context-based service orchestration Disruptions Evolution of current systems Evolution of current systems Intelligent, context-aware and adaptive HSIs User profiles Context-awareness (with sensors and pro Service-level agreements (SLA) Self-adapting learning algorithms for content & context Descriptive coding of context evolution in space and time 3-D realistic, other domains Evolution of current approaches Platforms for HSI development HSI usability test systems Generic 80 Domain-specific CAD technologies interpretation, understanding content, some emotions Gestures (to control applications) Brokering of services parameters with respect to overall performances of the transmission channel and terminal node capabilities Natural language understanding, application related semantic Text to speech Service coordination H.261/H.263 space, quality (perceived quality - full integrity) Behaviour prediction based on actual and historical data Fusion of proÞles across applications Federated Individual to-URN resolution Universal identification (e.g. camera images, position and digital maps) Engineering process support LT Rendezvous Voice commands noisy WAN WAN Standardised device interface (sensors / actuators) and 58 application model Profiles for users and user clusters session context Integration of information Algorithms for media integration Real-time projection algorithms for full windscreen projection (e.g. derived delay times from traf MT Speech recognition, speech to text Synthetic Service-based self-managing devices context fusion; integration of context Efficient analysis of data Rendezvous ST Simple, self-explaining and easy-to-use multimodal HSIs Automatic resource discovery in a network generic Context capturing and management Ef ficient & standardised context exchange (e.g. user profiles) standarisation Now LT Individual Manually con Þgured Seamless distributed network capabilities Sensor fusion to integrate raw physical data from dif Software architectures of collaborative sensory systems Infrastructure, MT T ransparent access to distributed data (individual sensor / actuator data or synthetic data) within the local network (e.g. in home, building, plants) Remote access to local systems and services (individual or synthetic) 57 Positioning Low-cost location positioning with increasing precision More ef ST Ambient intelligence Standardised exchange of positioning information Smart tags will allow identification Human-System Interaction 67 Now LT Digital Sensory System Challenge Process description language Executable Tool con Þguration for process support Requirements driven process management Requirements management and change management Requirements linking and tracking E.g. DOORS Requirements re-use techniques and guidelines Requirements based measurement Requirements, meta-models and specification languages From document-centred to object-oriented Re-use by reference Template Concept of uncertain requirements Requirements verification and validation Formalised description Formalised description Requirements based code generation techniques, test generations Knowledge based engineering Ontology , standardised knowledge models, taxonomies 82 Data mining, software mining and profiling Speci Þc De Þnitions and matching algorithms Focused and adapted generic methods, e.g. pattern matching based on DN A analysis Ta ble 3-2 T echnologies for engineering process support / T able 5-3 (e.g. enhanced route-finding algorithms, taking into account additional attributes such as user preferences and security aspects) T echnologies for Content Acquisition & Processing / Table 2-1 Rendezvous Content Infrastructures & Basic Services Human-System Interaction Engineering ITEA 2 - 9 An illustrating example: the ParMA project ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• • Parallel programming for Multicore Architectures “New machines will be parallel machines, and this will require major changes in the way we develop software” Software and the Concurrency Revolution by Herb Sutter and James Larus, Microsoft ITEA 2 - 10 ParMA Objectives ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• The objective of the ParMA project is to fully exploit the power of multi-core architectures to: – deliver substantial performance improvements for: • conventional HPC (High Performance Computing) applications • mainstream applications, • embedded systems (MPSoC: Multi-Processor System-on-a-Chip) – enable the advent of power-intensive innovative applications by: – developing advanced technologies for exploiting parallel computing – exploring commonalities between HPC, Multi-Core and MPSoC programming environments – leveraging relevant technology and methodology between these domains ITEA 2 - 11 Project tasks ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• WP1: Project Management, Guidance, and Dissemination WP2: Evolve methods and tools to facilitate the development (or restructuring) of parallel applications and to make them much more efficient.: development environment WP3: Provide advanced tools to check, debug, measure, analyse, and optimise parallel applications. debugging and tuning environment WP4: Extend the Linux Operating System (NUMA API, Scheduler, etc.) and optimise libraries to enable parallel applications to fully exploit the power of multi-core / multithreaded architectures execution environment The ParMA Technology WP5: Experiment and demonstrate the ParMA technology in diverse application domains Feedback to Technology providers (WP2, WP3, WP4) ITEA 2 - 12 ITEA 2 Symposium 2006 The consortium (1/2) Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• • Developers of parallel applications • Platform provider – Bull (F) F – HPC D • Magma (D), Recom (D), GNS (D) • SW tool vendors – Embedded systems F E • Dassault aviation (F), Indra (S) – HPC • Caps enterprise (F), GWT (D), F D – Embedded systems • Allinea (UK) ITEA 2 - 13 ITEA 2 Symposium 2006 The consortium (2/2) Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• • Three large HPC centers – HLRS / Universität Stuttgart (D) D – ZIH / TU Dresden (D) – FZ Jülich / ZAM (D) • Academic and applied research labs – HPC • UVSQ (F) • INT (F) F – Embedded systems • CEA-LIST (F) • UAB (S) • Robotiker (S) F E ITEA 2 - 14 ParMA: A key investment ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• • Massively parallel programming to fully exploit the power of multi-core architectures is a very ambitious goal that requires a substantial research and experimentation effort • ParMA builds on a broad participation across countries because – The knowledge and skills contributed by the various partners are complementary and cannot be found within a single European country, – Such endeavor requires a huge effort to cover a large enough perspective, – The whole HPC European research and industry is at stake. • The expected results will strengthen the European Research expertise in this critical domain and, more importantly, the worldwide market position of European parallel programming tool vendors, multi-core processors-based systems manufacturers, HPC applications developers and distributors as well as high performance embedded systems suppliers ITEA 2 - 15 ITEA2: amplifying, extending System@tic/Teratec initiatives at European level ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• IOLS UVSQ Dassault Bull Ter@tec CEA Autres… System@tic Labos associés Exemple INT FAME2 Cariocas HLRS, FZ Jülich, Magma,GWT,GNS, RECOM, ZIH Dresden, INDRA, U BA, Allinea,... ITEA2 ∑ ITEA 2 - 16 ! 2023 ITEA 2 Symposium 2006 Technical state-of-the-union address ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• ITEA 2 - 17 Thank you for your attention