Photonic enabled petascale in-memory computing with femtojoule energy consumption (PHOENICS)
(Funded by the EU)
Project ID: 101017237
Duration: January 2021 - December 2024
Breaking Moore's Law: New photonic computing project aims to speed up artificial intelligence computing power to petascale processing levels
The EU's new 5.8 million euro project using light to provide the ultrafast computing rates needed by artificial intelligence (AI) kicked off on march 31st. The PHOENICS project brings together world leaders in neuromorphic photonic computing to achieve for the first time energy efficient petascale processing powers with ultra-high bandwidth. This is the processing power required by AI to reach its full potential.
The 4-year project is ajoint research effort of the coordinating University of Münster (WWU, Germany) in collaboration with:
- University of Exter (UK)
- École polytechnique fédérale de Lausanne (EPFL, Switzerland)
- Nanoscribe GmbH & Co.KG (Germany)
- University of Oxford (UK)
- Heinrich Hertz Institute of the Fraunhofer Gesellschaft (HHI, GErmany)
- University of Ghent (Belgium)
- IBM Research GmbH (Switzerland)
- MicroR System (Switzerland)
Applications employing AI pervade more and more of today’s digitized society but pose enormous challenges for electronic hardware in terms of computing power and storage capacity. AI needs processing power growing at rate more than 5x higher than given by Moore’s Law and this is not possible with current trends. The PHOENICS project aims to address these challenges with innovative hardware approaches for processing the enormous data volumes, which are needed by demanding AI applications. By moving away from electronic towards photonic approaches, the PHOENICS project will establish disruptive methods for ultrafast information processing.
As a new paradigm for AI computing, the development of photonic neuromorphic processors will be the key innovation, which promises to deliver unprecedented computing power and energy efficiency.
The acronym PHOENICS stands for “Photonic enabled petascale in-memory computing with femtojoule energy consumption” and sums up the three goals of the project: In contrast to traditional hardware, the concept of in-memory computing will allow data processing more similar to the human brain by removing the separation between computing and data storage units; photonic technology provides high speed data transport where current electronic systems face severe limitations; taken together, this will lead to significant energy advantages.
The PHOENICS architecture is based on the hybrid integration of three different chip-platforms: a frequency microcomb chip, which is developed jointly by EPFL and MicroR Systems, an InP (indium phosphide) active modulation unit developed by HHI, complemented with a silicon photonics processor developed by the Universities of Exeter, Oxford, and Ghent. WWU, Nanoscribe GmbH & Co. KG, and IBM will design the system architecture and join the chip platforms.
Over the 4-year funding period, the consortium plans to establish photonic computing as a competitive approach for machine learning.
The project has received funding from the European Union’s Horizont 2020 research and innovation programme under grant agreement No 101017237.
EU Commission's "FET Proactive" funding line
FET Procative nurtures emerging themes, seeking to establish a critical mass of European researchers in a number of promising exploratory research topics. This supports areas that are not yet ready for inclusion in industry research roadmaps, with the aim of building up and structuring new interdisciplinary research communities. FET Proactive is part of the EU’s “Horizon 2020” Framework Programme for Research and Innovation.
University of Münster (Germany), University of Exeter (UK), École polytechnique fédérale de Lausanne (EPFL, Switzerland), Nanoscribe GmbH & Co. KG (Germany), University of Oxford (UK), Fraunhofer Gesellschaft, Heinrich Hertz Institute (Germany), University of Ghent (Belgium), IBM Research GmbH (Switzerland), MicroR Systems (Switzerland)