The chip helps turn smartphones into supercomputers

Researchers at MIT have successfully developed a type of 'memristors' artificial synapses with the most powerful processing performance ever.

Made of silicon with tens of thousands of artificial synapses , the chip simulates the activity of transmitting brain information to process complex AI algorithms while keeping it small and energy efficient.

This micro-nerve chip with thousands of memristors inside can capture large amounts of data, or recreate and blur image details. (Image: Getty Images).

The special feature of this 'neural microprocessor' , compared to previous versions, is that it does not need to be connected to the data center and can still operate stably, thereby bringing effective memory and storage. High resolution image.

The team aims to expand the artificial neural network based on software and powerful GPU capable of multi-tasking. As a specialized hardware, they can be integrated into mobile devices such as phones and cameras.

Unlike many traditional semiconductor chip lines that can only switch between two data states (0 or 1), artificial synapses can provide a range of values ​​equivalent to brain memory. It can learn working statuses and quickly reproduce the old signal for the same current flowing many times.

Reportedly, the team has applied a concept in the field of metallurgy: When metallurgists want to change the metal properties, they will combine with another substance with the desired properties to create alloys.

Similarly, the researchers used a chemical element in combination with silver, the material used to make memristor anodes, to increase the ability to continuously move ions in a thin channel.

Overall, this area still has a long way to go, but partly sees the future of artificial neural computers, which can do a lot of complex tasks on supercomputing scales without consuming a lot of power. Quality and network connectivity.