Chipmaker Intel has announced the Loihi 2, the second generation of its so-called neural research chips. The chips and associated software depend on the neurons in the brain.
The idea behind neural computers is that they mimic the behavior of individual neurons in the brain. So the calculations are made through a whole series of smaller units communicating with each other via electrical peaks (“pins” in English), with voltage rising and falling. The terminology is similar to how we also talk about brains, where tiny neurons communicate via electrical impulses. This would make chips more suitable for artificial intelligence, for example.
So Intel is already working on the second generation of those neural chips: Loihi 2. With the new version, the processor gets an upgrade, but the hardware is also significantly modified. This should make it possible to run a whole range of new algorithms and discover patterns in data in more efficient ways. Its main advantage is its power consumption, which should be much lower for complex tasks compared to conventional chips.
Search
The chip consists of 128 cores, each of which consists of 8192 components. They send electrical impulses to each other to form a kind of neural network that addresses a specific problem. Information is sent between those “neurons” in the form of traditional binary spikes (zero and one) or via gradient (32-bit) spikes that mimic the impulses in our brains. Intel writes this in its comprehensive technical explanation which you can read here (PDF). The company also says that the Loihi 2 is ten times faster than its predecessor four years ago.
The chip is primarily intended for research, and has not yet been used in the workplace. Researchers can test it via the Intel Neuromorphic Research Cloud, along with the open source Lava framework, which Intel wrote for these chips.
The idea behind neural computers is that they mimic the behavior of individual neurons in the brain. So the calculations are made through a whole series of smaller units communicating with each other via electrical peaks (“pins” in English), with voltage rising and falling. The terminology is similar to how we also talk about brains, where tiny neurons communicate via electrical impulses. This would make the chips more suitable for artificial intelligence, for example, Intel is already working on the second generation of these neural chips: Loihi 2. With the new version, the processor will be upgraded, but the hardware will also be significantly modified. This should make it possible to run a whole range of new algorithms and discover patterns in data in more efficient ways. Its main advantage is its power consumption, which should be much lower for complex tasks compared to conventional chips. The chip consists of 128 cores, each of which consists of 8192 components. They send electrical impulses to each other to form a kind of neural network that addresses a specific problem. Information is sent between those “neurons” in the form of traditional binary spikes (zero and one) or via gradient (32-bit) spikes that mimic the impulses in our brains. Intel writes this in their extensive technical explanation, which you can read here (pdf). The company also says that the Loihi 2 is ten times faster than its predecessor four years ago. The chip is primarily intended for research, and has not yet been used in the workplace. Researchers can test it via the Intel Neuromorphic Research Cloud, along with the open source Lava framework, which Intel wrote for these chips.
