TU Dresden, University of Manchester and GlobalFoundries announce SpiNNaker2, a breakthrough in AI cloud systems, bringing real-time AI with sub-millisecond latency and high power efficiency at cloud scale

TU Dresden, University of Manchester and GlobalFoundries announce SpiNNaker2, a breakthrough in AI cloud systems, bringing real-time AI with sub-millisecond latency and high power efficiency at cloud scale

Dresden − July 27, 2021 − Technische Universität Dresden, University of Manchester, Racyics GmbH and GlobalFoundries (GF), the world’s leading specialty foundry, today announced the tapeout of the SpiNNaker2 chip, a new brain-inspired artificial intelligence chip human. Based on the SpiNNaker infrastructure developed by the University of Manchester, the hybrid AI architecture created at TU Dresden, the ABX® adaptive body polarization IP platform by Racyics and using the 22FDX® solution by GF, SpiNNaker2 is a real-time neuromorphic AI processor with unparalleled efficiency and sub-millisecond latency for event-based systems. SpiNNaker2 is designed to scale up to a 70,000 chip cloud system while maintaining strict real-time operation. SpiNNaker2 is a key enabler for real-time AI at massive data rates. This will have a disruptive impact on AI applications such as smart cities, 5G, tactile internet, autonomous driving, which go beyond current hardware due to their triple requirements for low latency, high throughput and high energy efficiency. The complete 10 million core SpiNNaker2 machine, called “SpiNNcloud”, will be deployed at the Technische Universität Dresden for research purposes. In parallel, the start-up “SpiNNcloud Systems GmbH” will market SpiNNaker2.

AI is having an increasingly important impact on our daily lives. However, current AI hardware and algorithms are still only partially inspired by the main AI model, the human brain. In particular, even the best AI hardware is still a far cry from the 20W power consumption, low latency, and unprecedented large-scale, high-throughput processing offered by the human brain. SpiNNaker2 is a multi-processor system for real-time event-based processing of classical AI and upcoming sparse event-based computing paradigms inspired by the parsimonious, on-demand processing nature of the brain. It was developed within the framework of the flagship project of the European Union “Human Brain Project” (https://www.humanbrainproject.eu). SpiNNaker2 features a patented event-based compute architecture with power management and custom AI hardware acceleration. Its backbone is a lightweight, bespoke communications infrastructure that can scale from edge devices to large-scale server/cloud applications, without suffering the inherent slowdown that typically results from scaling. computing scale to the dimensions of the cloud.

As Professor Steve Furber (University of Manchester) says: “There is still a lot to learn from biology if we are to realize the full potential of AI in the future. SpiNNaker2 was designed to bridge the gap between realistic brain models and AI so that each can be increasingly informed by the other.”

The SpiNNaker2 chip is intended to build a 70,000 chip, 10 million arm ‘SpiNNcloud’ core at the Technische Universität Dresden, which will be used with local initiatives such as the CETI cluster of excellence, national partners and international partners to explore applications ranging from autonomous driving, real-time smart city data manipulation, touchscreen internet applications, to biomedical treatments.

“Our holistic system design optimized for parsimony, combined with a unique hybrid AI framework, enables real-time AI at unprecedented scale. We will do world-class research on new machine learning via the SpiNNaker2 machine at TU Dresden, but we are proud to also offer this unique AI commercially via our startup SpiNNcloud Systems (www.spinncloud.com),” says the professor. Christian Mayr (Technische Universität Dresden).

SpiNNaker2 was realized in the 22FDX ecosystem using the Adaptive Body Polarization (ABX®) from Racyics GmbH to exploit the full potential of 22FDX. It allows SpiNNaker2 to operate robustly at the minimum energy point of processing elements at 0.50V with additional support for dynamic voltage and frequency scaling for instant high performance on demand. Thus, SpiNNaker2 exploits the advanced energy saving features of 22FDX, the flagship technology offered by GF Dresden. The complete chip design was performed using makeChip (www.makechip.design), a turnkey hosted design environment for 22FDX offered by Racyics. This approach was essential for the university group to bridge the so-called gap between lab and manufacturing and successfully realize a highly complex system-on-chip for mass production.

22FDX®, GF’s unique 22nm FD-SOI transistor technology, with its world-class low standby power and low-cost optimized transistor performance, proves to be an ideal foundation at the cutting edge of true artificial intelligence. The 22FDX-based SpiNNaker2 chip shows the synergy between European world-class AI research and European world-class chip technology, beginning to bridge the huge gap between pure computing power AI and true AI brain type.

“As the world’s leading specialty foundry, GF enables world-class teams to achieve business breakthroughs. SpiNNaker2 is tangible proof of the exceptional ultra-low-power performance achieved by combining the latest advancements in brain-like artificial intelligence and our high-performance 22FDX technology,” says Dr. Manfred Horstmann, Director of GlobalFoundries Dresden.

Comments are closed.