Researchers in China have introduced the world’s largest computer chip, and you’ve probably never heard of it. Dubbed “Darwin Monkey” or “Wukong”, the system is modelled on the neural structure of the macaque brain and consists of more than 2 billion artificial neurones and over 100 billion synapses. Scientists say the machine could provide a critical platform for efforts to build what’s known as artificial general intelligence (AGI) — an aspirational level of machine smarts which, like human intelligence, would be general in the sense that it could apply to a variety of tasks.
China’s Darwin Monkey Supercomputer Mimics Brain Signals While Using Minimal Power
According to a report, the system architecture is built on spiking neural networks that closely mimic the way neurones are communicating in biological brains. And rather than handling continuous binary states, SNNs communicate bursts of electrical activity—spikes—which are fired only when sufficient input has been accepted. This architecture enables data to be processed in parallel and savings in energy. Its developers claim it draws just 2,000W from the wall, and that it is hanging off 960 Darwin III neuromorphic chips – each with millions of spiking neurones.
China has created a supercomputer called “Darwin Monkey” or “Wukong,” which is the biggest brain-inspired computer in the world. This system has over 2 billion artificial neurons and more than 100 billion synapses, similar to the brain structure of a macaque monkey.
Scientists in China developed this supercomputer based on the brain-like structure of a monkey.
It is named Darwin Monkey or “Wukong” and has more than 2 billion artificial neurons and over 100 billion synapses, which makes it similar to the neural structure of a macaque monkey.
The researchers believe this system can help neuroscientists study the brain and also help move closer to creating artificial general intelligence (AGI).
AGI is an AI system that can think and reason like a human.
Unlike regular computers that use binary code and process data in a linear way, Darwin Monkey uses a different approach called spiking neural networks (SNNs).
These networks work like the way neurons in a mammal’s brain send signals. Neurons send electrical signals in bursts, which is how information is passed around in the brain.
Artificial neurons in SNNs do the same thing: they send signals only when they get enough electrical input.
This is similar to how real neurons fire. Software-based neural networks use algorithms that try to copy the human brain, but SNNs are more like a real brain. This makes them more powerful than traditional computers because they can handle data at the same time.
SNNs might also use less energy.
After sending a signal, artificial neurons take a short break, which means they don’t fire as often. This helps save power. Darwin Monkey uses only 2,000 watts of power, which is about the same as a kettle or hairdryer. It uses 960 Darwin III neuromorphic chips, each of which can support up to 2.35 million spiking neurons.
Before Darwin Monkey, the largest neuromorphic system was Intel’s Hala Point, which had 1.15 billion artificial neurons and 128 billion artificial synapses across 140,544 processing cores.
Intel claims it can do 20 quadrillion operations per second, but comparing neuromorphic computers to regular supercomputers is tricky because they work differently.
The team behind Darwin Monkey, from Zhejiang University and Zhejiang Lab, a joint research institute between the Zhejiang government and Alibaba, said the system has already shown it can handle tasks like logical reasoning, content creation, and math problems using an AI model from a Chinese startup called DeepSea.
