[ad_1]
Robots with Intel Inside brains? That’s what Chipzilla has in mind with her Loihi 2 neuromorphic chip, which tries to mimic the human brain.
It’s Intel’s attempt to create smarter computers capable of efficiently discovering patterns and associations in data and learning to make increasingly smarter decisions.
It is not a processor in the traditional sense, and it is intended for experimentation rather than production use. As you can see in the datasheet [PDF], it consists of up to 128 cores each containing up to 8,192 components that act like natural spiked neurons that send messages to each other and form a neural network that tackles a particular problem.
Cores also implement synapses that transmit information between neurons, which can each send binary peaks (1s or 0s) or graduated peaks (32-bit payload value).
An overview of the architecture of the Loihi 2 chip … Source: Intel. Click to enlarge
Each neuron can be assigned a program written using a basic instruction set to perform a task, and the whole thing is driven by six normal processor cores that run software written in, say, C. There is also external I / O to communicate with other hardware, and interfaces to link multiple Loihi 2 chips together in a mesh as needed. There are other features, such as three-factor learning rules, which you should check out the tech brief for more details. The previous generation did not have graduated tips or programmable neurons.
The “strengths” of the Loihi 2 neuron instruction set … Source: Intel. Click to enlarge
There is a race to replicate the brain electronically to run powerful AI applications quickly and without making, say, a massive dent in the electric bill. Samsung just said it wants to put human-like brain structures on a chip. IBM is also developing hardware designed around the brain.
Intel’s latest Loihi is ten times faster than the previous generation component announced four years ago this month, it is claimed.
“Loihi 2 is currently just a research chip,” said Garrick Orchard, researcher at Intel Labs via email with The register. “Its core-based architecture is scalable and could enable future versions of the chip – when the technology matures – which could have a range of business applications from data center to edge devices. “
Each Loihi 2 chip potentially has more than a million digital neurons, compared to 128,000 in its predecessor. To put that in context, there are approximately 90 billion interconnected neurons in the human brain, which should give you an idea of the level of intelligence possible with this material right now.
Digital neurons calculate asynchronously in parallel and can be customized by programming. Loihi 2 supports a maximum of 120 million synapses, compared to over a trillion synapses in the human brain, and has 2.3 billion transistors in a 31mm2 matrix area. According to Intel, its digital circuits operate “up to 5,000 times faster than biological neurons.”
The chip is a first sample from the manufacturing node Intel 4, which is the semiconductor giant’s brand name for its highly retarded 7nm processing node, and uses extreme ultraviolet (EUV) to etch the chips. Loihi 1 was performed using a 14 nm process.
“With Loihi 2 manufactured with a pre-production version of the Intel 4 process, this underscores the health and progress of Intel 4,” an Intel spokesperson told us.
Intel, by the way, showcased the Intel 4 processor family board, codenamed Meteor Lake, for desktop and mobile PCs, at a press event in July, the first time it was released. ‘did. Chips using this microarchitecture are expected to ship in 2023, so Loihi 2 is a preview of what’s to come in terms of manufacturing.
Intel is working with the research community to provide applications for Loihi 2. Its predecessor was used to create systems for odor identification, robotic arm management, and rail scheduling optimization.
There are no ongoing projects with Loihi 2 yet, although partners who worked with the original Loihi “have expressed their enthusiasm for the new abilities in Loihi 2,” Orchard said.
One of those partners is the US National Laboratory in Los Alamos, which uses the first-generation Loihi chip as an artificial brain to understand the benefits of sleep.
An open source programming framework called Lava has been introduced alongside Loihi 2 with which developers can write AI applications that can be implemented in the on-chip neural network. The underlying tools will also support the robotic operating system (ROS), TensorFlow, Pytorch, and other frameworks.
The Lava framework is available for download on Github.
This neuromorphic material will be available to researchers through Intel’s Neuromorphic Research Cloud. Available components include the Oheo Gulch board, which includes a single Loihi 2 socket connected to an FPGA. A system named Kapoho Point with eight Loihi 2 chips will be available soon. ®
Our friends at The next platform have more analysis and information on Loihi 2 here.
[ad_2]
Source link