Intel Neomorphic System: "Inspired by the Human Brain"

Intel's creation, called Hala Point, is only the size of a microwave oven, but boasts 1.15 billion artificial neurons. It's a huge step up from the 50 million neuron capacity of its predecessor, Pohoiki Springs

by Sededin Dedovic
Intel Neomorphic System: "Inspired by the Human Brain"
© Justin Sullivan / Getty Images

In neuromorphic computing, the focus is on simulating the structure of the human brain to provide more efficient data processing, including faster speeds and higher accuracy, and that's a current topic. Hala Point boasts 1.15 billion artificial neurons.

Many universities and technology companies are working on this, including scientists at Intel, who have built the largest "brain-based computer system" for the Sandia National Laboratories in New Mexico, as reported by TechRadar.

Intel's creation, called Hala Point, is the size of just a microwave but boasts 1.15 billion artificial neurons. This is a huge leap from the capacity of 50 million neurons of its predecessor, Pohoiki Springs, which appeared four years ago.

There's a theme in Intel's naming, in case you were wondering - they're locations in Hawaii. Hala Point is ten times faster than its predecessor, 15 times denser, and with a million cores on one chip. Pohoiki Springs had only 128,000.

Equipped with 1,152 Loihi 2 research processors (Loihi is a volcano in Hawaii), the Hala Point system will be tasked with harnessing the power of extensive neuromorphic computing. "Our colleagues at Sandia have consistently applied our Loihi hardware in ways we never imagined, and we look forward to their research with Hala Point that will lead to breakthroughs in the scale, speed, and efficiency of many significant computing problems," said Mike Davies, director of the Neuromorphic Computing Lab at Intel Labs.

Since a neuromorphic system of this size has not existed before, Sandia has developed special algorithms to ultimately harness the full potential of the computer. "We believe that this new level of experimentation - the beginning, we hope, of large neuromorphic computers - will help create brain-based systems with unparalleled processing, response, and learning capabilities from real data," said Sandia's lead researcher, Craig Vineyard.

His colleague, research associate Brad Eymon added, "One of the main differences between brain-like computing and conventional computers we use today - both in our brains and in neuromorphic computing - is that processing is distributed across many neurons in parallel, rather than long serial processes that are an inevitable part of conventional computing.

As a result, the more neurons we have in a neuromorphic system, the more complex computations we can perform. We see this in real brains too. Even the smallest mammal brains have tens of millions of neurons; our brains have about 80 billion.

We see this in today's AI algorithms too. Bigger is far better." Currently, Hala Point is a research prototype that will enhance the capabilities of future commercial systems. Intel predicts that such lessons will lead to practical advancements, such as the ability for LLM to continuously learn from new data.

Such progress promises a significant reduction in the unsustainable burden of training widespread AI implementations. Researchers from Sandia National Laboratories plan to use Hala Point for advanced brain-level research. The organization will focus on solving scientific computing problems in device physics, computer architecture, computer science, and informatics.

Recent trends in scaling deep learning models to trillions of parameters have exposed daunting sustainability challenges in artificial intelligence and highlighted the need for innovations at the lowest levels of hardware architecture.

Neuromorphic computing is fundamentally a new approach that relies on insights from neuroscience that integrate memory and computation with very granular parallelism to reduce data movement. Advancing beyond its predecessor, Pohoiki Springs, Hala Point now brings neuromorphic performance and efficiency improvements to major conventional deep learning models, especially those processing real-time workloads such as video, speech, and wireless communications.

A secret service officer watches the crowd prior to President Joe Bidens remarks at Intel Ocotillo Campus on March 20, 2024 in C© Rebecca Noble / Getty Images

Intel will receive $8.5 billion in non-repayable assistance and an $11 billion loan from the U.S.

government to expand its manufacturing capacity. The aim of this assistance is for the company to regain its leadership position in chip manufacturing and compete with major rivals such as Taiwan's TSMC and South Korea's Samsung.

In addition to the promised assistance, Intel has committed to investing an additional $100 billion in chip manufacturing factories over the next five years. They also expect to receive a tax return of 25 percent of the invested amount.

The U.S. government also expects this move to provide 30,000 new jobs. The first tranche of money could be allocated to Intel as early as this year, immediately after the agreement is finalized. The goal is to ensure that by the end of this decade, 20 percent of the most advanced chips are produced in the U.S.

Intel has a significant advantage over competitors when it comes to computer processors. Although it faced financial difficulties during 2022 and 2023, Intel essentially did a great job during that period, according to ExtremeTech.

According to the analyst firm Canalys, this company has maintained a dominant position, with a 78 percent market share in this segment. It is followed by AMD with 13 percent, while all others, Apple, Qualcomm, Arm, and MediaTek, together hold 9 percent of the market.