Periodic Reporting for period 1 - PHOENICS (Photonic enabled Petascale in-memory computing with Femtojoule energy consumption)
Okres sprawozdawczy: 2021-01-01 do 2022-06-30
Modern societies and economies increasingly depend on the massive generation of data resulting from the exponential growth of internet applications. Drastically enhanced computational performance is in particular needed for a plethora of
applications in artificial intelligence (AI) which necessitate unprecedented processing power, memory and communication bandwidth. This demand cannot be met by modern digital electronic technologies that are rapidly approaching their physical limits. The PHOENICS consortium will break through these barriers and lay the foundation for a disruptive neuromorphic compute platform based on hybrid photonic integrated circuits. By providing access to parallelized neuromorphic processing using wavelength division multiplexing, the PHOENICS consortium will harness exceptional scaling potential not available to
electronic systems and will deliver multiply-accumulate (MAC) performance at 3.2 PetaMAC/s at an energy cost of 50 FemtoJoule/MAC. Building on a hybrid architecture with substantial potential for future upscaling, the PHOENICS project aims at implementing a disruptive architecture which outperforms state-of-the-art electronic neuromorphic hardware. The consortium partners have shown the significant technological potential that a photonic approach can offer by establishing a new brain-inspired computing paradigm using phase-change-materials. By implementing scalable systems based on foundry
processing for creating bio-mimicking material platforms, the PHOENICS consortium will provide a new generation of photonic hardware accelerators for neuromorphic processing and develop a strong ecosystem for photonic computing.
applications in artificial intelligence (AI) which necessitate unprecedented processing power, memory and communication bandwidth. This demand cannot be met by modern digital electronic technologies that are rapidly approaching their physical limits. The PHOENICS consortium will break through these barriers and lay the foundation for a disruptive neuromorphic compute platform based on hybrid photonic integrated circuits. By providing access to parallelized neuromorphic processing using wavelength division multiplexing, the PHOENICS consortium will harness exceptional scaling potential not available to
electronic systems and will deliver multiply-accumulate (MAC) performance at 3.2 PetaMAC/s at an energy cost of 50 FemtoJoule/MAC. Building on a hybrid architecture with substantial potential for future upscaling, the PHOENICS project aims at implementing a disruptive architecture which outperforms state-of-the-art electronic neuromorphic hardware. The consortium partners have shown the significant technological potential that a photonic approach can offer by establishing a new brain-inspired computing paradigm using phase-change-materials. By implementing scalable systems based on foundry
processing for creating bio-mimicking material platforms, the PHOENICS consortium will provide a new generation of photonic hardware accelerators for neuromorphic processing and develop a strong ecosystem for photonic computing.
In reporting period 1 the foundations for the hybrid PHOENICS platform were laid on each individual material platform. This was achieved in the individual work packages as follows:
WP1 Management and Coordination: WWU was responsible for the overall Coordination and Management of PHOENICS.
WP2 Hybrid integration and input modulation: The overarching goal of WP 2 is the design and fabrication of input modulation units for the PHOENICS project. These devices are based on a hybrid photonic integration of InP-based amplifier-modulator arrays and polymer-based arrayed waveguide gratings for the (de-)multiplexing of the optical signals.
WP3 Multi-Frequency Comb driver chip:The overarching objective of WP3 is to design, fabricate and develop the packaging of soli-ton microcombs based on Si3N4 microresonators directly pumped by InP high power laser diode.
WP4 Matrix multiplier: The objective in WP4 is to develop matrix multiplier which can be implemented with the demultiplexer and detection units developed in WP5 to realize chip-scale MVM systems that operate at multiple wavelengths in parallel.
WP5 The objective of this WP is to design the demultiplexers and the detectors on the silicon photonics chips and integrate them with the matrix from WP4 (Figure WP5.2).
WP6 Analog convolution processing: The general objective of Work Package 6 is to demonstrate modules based on the hybrid neuromorphic platform in an accelerator for convolution layer processing in Convolutional Neural Networks.
WP7 Packaging: The overall objective of this work package is to develop assembly and packaging strategies for the heterogeneous integration of the PHOENICS platforms.
WP1 Management and Coordination: WWU was responsible for the overall Coordination and Management of PHOENICS.
WP2 Hybrid integration and input modulation: The overarching goal of WP 2 is the design and fabrication of input modulation units for the PHOENICS project. These devices are based on a hybrid photonic integration of InP-based amplifier-modulator arrays and polymer-based arrayed waveguide gratings for the (de-)multiplexing of the optical signals.
WP3 Multi-Frequency Comb driver chip:The overarching objective of WP3 is to design, fabricate and develop the packaging of soli-ton microcombs based on Si3N4 microresonators directly pumped by InP high power laser diode.
WP4 Matrix multiplier: The objective in WP4 is to develop matrix multiplier which can be implemented with the demultiplexer and detection units developed in WP5 to realize chip-scale MVM systems that operate at multiple wavelengths in parallel.
WP5 The objective of this WP is to design the demultiplexers and the detectors on the silicon photonics chips and integrate them with the matrix from WP4 (Figure WP5.2).
WP6 Analog convolution processing: The general objective of Work Package 6 is to demonstrate modules based on the hybrid neuromorphic platform in an accelerator for convolution layer processing in Convolutional Neural Networks.
WP7 Packaging: The overall objective of this work package is to develop assembly and packaging strategies for the heterogeneous integration of the PHOENICS platforms.
PHOENICS will generate impact on a broad range of areas in our increasingly information-driven society. The project will lead to breakthrough developments in the key areas of AI, autonomous navigation, health, mobility and security . The high performance compute platform developed by the project will lead to dramatic leaps in the performance of sophisticated products developed by the industries active in these areas, including autonomous driving, AI-assisted diagnosis in medical imaging, cyber security and cloud services. Since the fabrication approach for PHOENICS systems is based on scalable mass fabrication, the innovative compute platform will find also applications in consumer electronics and mobile products.
Firstly, the high-throughput MVM systems will enable novel inference applications in data centers. By moving towards Petascale computation power, transformational changes in computation will be within reach.
Secondly, the PHOENICS platform will enable the implementation of CNN architectures with real-time processing capability in HDTV and 8k format. This will in particular enable the detection of micromotion and precision feature detection, which is of key importance in AI supported medical applications, such as neurosurgery and cancer diagnosis.
Thirdly, live processing of high volume image data with PHOENICS systems will provide transformational changes to object detection in autonomous navigation. High speed object detection will be of key importance to enable new applications in automatized driving and automatized aviation. In these key industries, high speed object recognition is a key technology driver which requires access to ultrafast compute hardware. The PHOENICS platform will provide the performance and bandwidth of enabling reliable object detection even at speeds required for airplanes and unmanned arial vehicles. The PHOENICS program thus provides crucial advances in an industry where Europe plays a world-leading role.
Fourth, PHOENICS will advance the state-of-the-art and our underpinning knowledge and understanding in a range of scientifically and technologically important areas, from nanophotonics and quantum technologies to memory technology and energy-efficient non-von Neumann computing.
Dissemination to the wider public audience is also important to aid the public understanding of science and engineering and to raise awareness of the beneficial use of public funds in the pursuit of scientific and technical advancement, and the consequent knock-on effects on economic and societal well being.
Firstly, the high-throughput MVM systems will enable novel inference applications in data centers. By moving towards Petascale computation power, transformational changes in computation will be within reach.
Secondly, the PHOENICS platform will enable the implementation of CNN architectures with real-time processing capability in HDTV and 8k format. This will in particular enable the detection of micromotion and precision feature detection, which is of key importance in AI supported medical applications, such as neurosurgery and cancer diagnosis.
Thirdly, live processing of high volume image data with PHOENICS systems will provide transformational changes to object detection in autonomous navigation. High speed object detection will be of key importance to enable new applications in automatized driving and automatized aviation. In these key industries, high speed object recognition is a key technology driver which requires access to ultrafast compute hardware. The PHOENICS platform will provide the performance and bandwidth of enabling reliable object detection even at speeds required for airplanes and unmanned arial vehicles. The PHOENICS program thus provides crucial advances in an industry where Europe plays a world-leading role.
Fourth, PHOENICS will advance the state-of-the-art and our underpinning knowledge and understanding in a range of scientifically and technologically important areas, from nanophotonics and quantum technologies to memory technology and energy-efficient non-von Neumann computing.
Dissemination to the wider public audience is also important to aid the public understanding of science and engineering and to raise awareness of the beneficial use of public funds in the pursuit of scientific and technical advancement, and the consequent knock-on effects on economic and societal well being.