Manchester, Dresden, Globalfoundries produce SpiNNaker2 chip
The SpiNNaker2 – codenamed JIB2 – is a spiking neural networking chip intended to be used in 70,000 chip, 10 million ARM core ‘SpiNNcloud’ system at Technische Universität Dresden. It will also be offered for sale through a startup company in Dresden called SpiNNcloud.
A roadmap shown in 2018 called for JIB2 to contain 144 Cortex-M4F cores plus power management, low-swing serial I/O, four LPDDR4 memory interfaces connecting to 8Gbytes of LPDDR4 memory. JIB2 is designed in 22FDX from Globalfoundries, a 22nm fully depleted silicon-on-insulator (FDSOI) manufacturing process.
The SpiNNaker2 chip was realized using the adaptive body biasing IP platform (ABX) of Racyics GmbH. This enables JIB2 to operate at the minimum energy point of the processing elements at 0.5V with additional support of dynamic voltage and frequency scaling for instantaneous high performance on demand.
The DVFS was targeted at taking JIB2 from ultralow power performance at 0.45V and 100MHz clock frequency through 250MHz at 0.5V and up to 500MHz at 0.6V.
The complete chip design was done using makeChip, a turnkey hosted design environment for 22FDX offered by Racyics. This approach was key for the academic group to close the so-called lab-to-fab gap and to realize a complex system-on-chip for volume production.
The SpiNNaker2 system is hybridized development of the SpiNNaker unit in Manchester (see SpiNNaker neuromorphic supercomputer reaches one million cores). It supports event-based real-time processing of both classic AI and upcoming sparse, event-based compute. It has been developed within the European Union flagship project “Human Brain Project” (https://www.humanbrainproject.eu).
Next: Professors Furber and Mayr
There is still a great deal to learn from biology if we are to realize the full potential of AI in the future. SpiNNaker2 has been designed to bridge the gulf between realistic brain models and AI so that each may increasingly be informed by the other,” said Professor Furber.
The SpiNNcloud system will be used to explore applications that range from autonomous driving, handling of the data of a smart city in real time, tactile internet applications, up to biomedical processing.
SpiNNaker was originally proposed by Professor Furber at Manchester University and a first chip containing 18 32bit ARM cores was implemented in 130nm CMOS. Professor Furber was one of the two original designers of the Advanced RISC Machine or ARM processor.
The JIB2 SpiNNaker2 chip therefore represents a more than eight-fold increase incomplexity at the chip level.
“Our holistic sparsity-optimized system design, combined with a unique hybrid AI framework enables real-time AI at an unprecedented large-scale. We will do world-class research on novel machine learning via the SpiNNaker2 machine at TU Dresden, but we are proud to also offer this unique AI commercially through our startup SpiNNcloud Systems,” said Professor Christian Mayr (Technische Universität Dresden).
Related links and articles: