Projektbeschreibung
Ein neues Zeitalter des Hochleistungsrechnens dank eines neuen Ansatzes an supraleitfähige Logikgatter?
Taschenrechner waren eine große Verbesserung gegenüber Stift und Papier, doch Computer haben unsere Fähigkeiten zum Durchführen von Berechnungen grundlegend verbessert, da immer mehr und komplexere Aufgaben in einem Bruchteil der Zeit vorgenommen werden können. Supercomputer, die auf supraleitfähigen Quantenlogik-Gattern beruhen, haben noch mehr Möglichkeiten geschaffen. In zahlreichen Bereichen nehmen sie eine immer wichtigere Rolle ein, haben jedoch Schwierigkeiten, wenn es darum geht, die Leistung zu steigern und den Energieverbrauch gleichzeitig zu senken. Das EU-finanzierte Projekt SuperGate wird einen neuen Ansatz an supraleitfähige Logiken ausarbeiten, die eine mindestens gleichwertige, wenn nicht sogar bessere Leistung ermöglichen, während aktuelle Probleme gleichzeitig minimiert werden. Dadurch könnten Supercomputer die nächste Evolutionsstufe erreichen.
Ziel
Supercomputers are playing an increasingly important role for our society by performing calculations with a variety of implications ranging from weather forecasting to genetic material sequencing to testing of drugs for new diseases. Enhancing the performance of modern supercomputers, whilst minimising their energy losses, represent two contrasting but major needs that the information technology industry will have to address in the future.
The best solution proposed to date to reduce the energy costs of supercomputers without affecting their performance is based on hybrid computing architectures, where a semiconductor part based on complementary metal-oxide semiconductor (CMOS) technology and used for memory operations is combined at low temperatures with logic circuitry offering minimal energy losses thanks to the usage of superconductor (S) materials. Existing superconducting logics, which relies on rapid single flux quantum (RSFQ) technology, however, it is difficult to interface with CMOS and to scale up and it is sensitive to magnetic perturbations – which are the main reasons why hybrid platforms have not replaced CMOS systems despite their advantages.
Starting from our recent discovery that the logic state in some S devices can be controlled via the field effect (FE), in this project we propose to develop a new technology for superconducting logics that can offer performance at least comparable to that of RSFQ logics whilst overcoming all its drawbacks. We will adopt a systematic approach aiming at (i) understanding of the FE in a S, (ii) determining the S materials and device geometry with optimised performance, (iii) testing the dynamic response of optimised devices, (iv) developing logic circuits based on such devices and (v) testing a logic circuit in conjunction with a CMOS electro-optical modulator. We will also establish technology transfer and pave the way for the commercialisation of our technology, which can revolutionise the world of supercomputer.
Wissenschaftliches Gebiet
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsensorsoptical sensors
- natural sciencesphysical scienceselectromagnetism and electronicssemiconductivity
- natural sciencesmathematicspure mathematicsgeometry
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringcomputer hardwaresupercomputers
- natural sciencesphysical scienceselectromagnetism and electronicssuperconductivity
Schlüsselbegriffe
Programm/Programme
Aufforderung zur Vorschlagseinreichung
Andere Projekte für diesen Aufruf anzeigenUnterauftrag
H2020-FETOPEN-2018-2019-2020-01
Finanzierungsplan
RIA - Research and Innovation actionKoordinator
78464 Konstanz
Deutschland