a supercomputer capable of simulating the full human

Moore's Law, which dates back many decades, states that a microchip's transistor count quadrupled roughly every two years. Due to this, our computers are now more affordable and smaller.

Transistor sizes are now getting close to the atomic scale, nevertheless. Excessive heat generation and a phenomena known as quantum tunnelling, which obstructs the transistors' ability to operate, are issues at these small scales. Transistor miniaturization will finally come to an end due to this slowing down.

Scientists are investigating novel computing strategies to address this problem, beginning with the human brain, which is a potent computer that is concealed inside each of us. The computer model proposed by John von Neumann does not describe how our brains function. They lack discrete memory and processing sections.

Rather, they function by establishing connections between billions of nerve cells, which exchange electrical impulses that carry information. A connection known as a synapse allows information to be transferred from one neuron to the next. The brain's network of synapses and neurons is adaptable, efficient, and scalable.

Thus, memory and computation are controlled by the same neurons and synapses in the brain, unlike in a computer. Researchers have been working on this paradigm since the late 1980s, hoping to apply it to computers.

Life imitating

The foundation of neuromorphic computers consists of complex networks of basic processors that function similarly to the neurons and synapses in the brain. This has the primary benefit that these devices are "parallel" by nature.

This implies that almost every processor in a computer has the capacity to function and communicate with every other processor at the same time, much like neurons and synapses do.

Additionally, the energy consumption is orders of magnitude lower since individual neurons and synapses conduct extremely basic computations compared to typical computers. Despite the fact that synapses are frequently thought of as memory units and neurons as processing units, they really contribute to both processing and storing. To put it another way, the data is already there where the calculation needs it.

As a result, the brain processes information more quickly overall since memory and processor are integrated, unlike in traditional (von Neumann) computers, which slow down processing. However, it also avoids the need to use a significant amount of energy and execute a specialized process of accessing data from a primary memory component, which occurs in conventional computer systems.

DeepSouth is mostly inspired by the ideas we just discussed. There are more neuromorphic systems in use right now than this one. The Human Brain Project (HBP), which is supported by an EU program, is noteworthy. The Human Brain Project (HBP) ran from 2013 to 2023 and produced the Heidelberg, Germany-based BrainScaleS system, which simulates the function of neurons and synapses.

BrainScaleS is able to replicate the "spike" that occurs when an electrical impulse passes through a neuron in the human brain. Because of this, BrainScaleS would be a perfect fit to study the mechanics of cognitive functions as well as the mechanisms behind neurological and neurodegenerative illnesses in the future.

Because neuromorphic computers are designed to resemble real brains, this might be the start of a revolution. They provide economical and sustainable processing power and enable researchers to assess neural system models, making them a perfect platform for a variety of uses. They may contribute to new developments in artificial intelligence as well as improve our knowledge of the brain.