Intel Unveils Hala Point: The World's Largest Neuromorphic System for Advancing Brain-Inspired Artificial Intelligence

Wednesday - April 24 2024, 04:42 UTC - 1 year ago

On April 24, 2024, Intel launched Hala Point, the world's largest neuromorphic system, powered by the Loihi 2 processor. It supports 20 petaops, performs 20 times faster than a human brain and 100 times more energy-efficient than traditional CPU and GPU systems. The system opens up possibilities in various industries and promises a more sustainable and efficient future for AI.

On the current date of Wednesday, April 24, 2024, Intel announced the launch of Hala Point, the world's largest neuromorphic system designed to advance brain-inspired artificial intelligence. Code-named after a location in Maui, Hawaii, Hala Point is set to support research for future AI development while tackling efficiency and sustainability challenges faced by modern AI systems. This groundbreaking technology, powered by Intel's Loihi 2 processor, has already been deployed at Sandia National Laboratories and is expected to have a major impact on the field of AI.

With an impressive capacity of up to 20 petaops (20 quadrillion operations per second), Hala Point boasts a phenomenal efficiency of 15 trillion 8-bit operations per second per watt (TOPS/W) when running traditional neural networks. This is a major achievement that rivals and even surpasses existing architectures built on graphics processing units (GPU) and central processing units (CPU). The capabilities of Hala Point open up new possibilities for real-time continuous learning in AI applications such as scientific and engineering problem-solving, logistics, smart city infrastructure management, large language models (LLMs), and AI agents.

Hala Point is a massive six-rack-unit data center housed in a microwave oven-sized chassis, containing 1,152 Loihi 2 processors. Produced on Intel's 3 process node, the system can support up to 1.15 billion neurons and 128 billion synapses distributed over an impressive 140,544 neuromorphic processing cores. To support ancillary computations, Hala Point also features over 2,300 embedded x86 processors.

One of the most notable features of Hala Point is its integration of processing, memory, and communication channels in a massively parallelized fabric. This allows for a whopping 16 petabytes per second (PB/s) of memory bandwidth, 3.5 PB/s of inter-core communication bandwidth, and 5 terabytes per second (TB/s) of inter-chip communication bandwidth. Such immense capabilities not only enhance the overall performance of the system but also enable it to execute bio-inspired spiking neural network models 20 times faster than a human brain with neuron capacity of 1.15 billion. The impressive speeds at lower capacities of Hala Point also match up to 200 times faster rates of a human brain. It is important to note that the system is not intended for neuroscience modeling, although its neuron capacity is similar to that of an owl brain or the cortex of a capuchin monkey.

Hala Point's advanced Loihi-based processors have also proven to be extremely energy-efficient compared to traditional CPU and GPU architectures. They can perform AI inference and solve optimization problems using 100 times less energy, while also running up to 50 times faster. This breakthrough technology paves the way for a future where AI systems are powered by neuromorphic processors, making them more energy-efficient, cost-effective, and overall more sustainable.

In summary, Hala Point is a major breakthrough in the field of neuromorphic computing, aiming to advance brain-inspired artificial intelligence. With its remarkable neuron capacity, impressive energy efficiency, and unparalleled performance, the system is set to revolutionize various industries and pave the way for a more sustainable and efficient future.