SpiNNaker2: ten million cores to simulate the brain

SpiNNaker2 arrives as the second generation of neuromorphic processors, designed to emulate large-scale biological neural networks in real time. With ten million ARM cores integrated, this system aims to replicate the activity of parts of the human brain, offering a tool for researchers in neuroscience and computing.

Neuromorphic chip wafer glowing with millions of ARM processor cores arranged in neural network patterns, data streams flowing between cores like biological synapses, researchers analyzing real-time brain simulation on holographic displays, robotic manipulator connecting wafer to cooling system, blue and orange electrical traces pulsing with activity, engineering visualization style, ultra-detailed silicon die structures visible, dramatic laboratory lighting with dark background, photorealistic technical render, high contrast metallic surfaces

Massively parallel architecture for biological simulation 🧠

Each ARM core in SpiNNaker2 functions as a virtual neuron, communicating via events similar to synaptic spikes. The platform uses a mesh interconnection network that allows data traffic between nodes without bottlenecks. The design prioritizes energy efficiency, consuming less than a traditional supercomputer, and focuses on learning models such as STDP (spike-timing-dependent plasticity).

The silicon brain that still can't make coffee ☕

With ten million cores, SpiNNaker2 can simulate an ant colony or a piece of cerebral cortex, but it still can't remember where you left your keys. Researchers celebrate that the system processes information like a biological brain, though with one advantage: it doesn't complain when you ask it to run the same loop ten thousand times. Artificial intelligence still has a lot to learn from biology, but at least it doesn't ask for vacations.