IBM unveils brain-like chip prototype for more energy-efficient artificial intelligence

According to global technology leader IBM, artificial intelligence (AI) could become more energy efficient when it enters a prototype "brain-like" processor.

Aug 12, 2023 - 14:50
 0  34
IBM unveils brain-like chip prototype for more energy-efficient artificial intelligence

According to global technology leader IBM, artificial intelligence (AI) could become more energy efficient when it enters a prototype "brain-like" processor. Emissions from computer warehouses operating artificial intelligence systems have been in the spotlight.

IBM noted that the prototype could lead to more efficient, battery-saving AI circuits for smartphones. He argued that its effectiveness came from parts that worked similarly to the connections in the human brain.

Physicist Thanos Vasilopoulos, who works at the IBM Research Center in Zurich, Switzerland, said that compared to traditional computers, the human brain can achieve significant performance while consuming little energy. He says that greater energy efficiency would allow "larger and more complex workloads to be performed in low-power or battery-constrained environments," such as in cars, cell phones and cameras.

In addition, he continued, "cloud providers can use these chips to reduce their energy costs and carbon footprint."

It shows how advances in hardware design can contribute to the energy efficiency of AI, which is critical to reducing the environmental impact and operating costs of AI systems.

Most chips are digital, meaning that data is stored as 0s and 1s, but the new chip uses memristors, which are analog components that can store different numbers.

The difference between digital and analog can be compared to the difference between a light switch and a light switch.

Since the human brain is analog, the function of memristors is analogous to synapses in the brain. According to Ferrante Neri, a professor at the University of Surrey, memristors are "inspired by nature" and computers mimic the brain.

Like a synapse in a biological system, a memristor can potentially "remember" its electrical past.

"Individual memristors can form a network that resembles a biological brain," he argued.

He expressed cautious optimism about the potential of chips using this technology, saying that "this development shows that we may see brain-like chips in the near future."

However, he warned that creating a memristor-based computer is not easy, and there are many obstacles to widespread adoption, such as high material costs and complex manufacturing processes. Using these parts of the new chip increases the energy efficiency of the AI, but also includes digital components. This makes it easy to integrate the chip into current AI systems.

Today, many phones are equipped with artificial intelligence processors that help, for example, process photos. For example, the iPhone has a "neural engine" chip.

According to IBM, processors in future phones and cars could be more powerful, offering longer battery life and new uses. In the future, chips like IBM's prototype could help save a lot of energy when they replace processors in banks of computers running powerful artificial intelligence systems.

Water needed to cool energy-consuming data centers can also be reduced. A large data center requires as much energy to run as a medium-sized city. Data centers require huge amounts of electricity to operate.

James Davenport, a professor of IT at the University of Bath, called IBM's findings "potentially interesting" but warned that the chip was a "potential first step" rather than a "simple-to-use" solution to the problem.

What's Your Reaction?

like

dislike

love

funny

angry

sad

wow