Intel announced a few hours ago that it has made significant progress in the field of neural or neuromorphic computing processors. The second-generation Loihi chip is about 10 times faster in processing algorithms and applications, and the chip is smaller than a fingertip.
To embed. Through neuromorphic chips, Intel hopes to circumvent Moore's Law, which has become a barrier to processor development. Neuromorphic processing is based on the science of neurotransmitters and how the human brain works, and the makers of these chips want to improve the learning of various systems by making them.
These chips are inspired by the neural networks of the human brain. Computer processing dates back to a few decades ago, but the neural networks in the brain are still unique because of their power and speed. This new chip is the product of 3 years of research in Intel Neuroscience Research.
Intel wants to emulate From the human brain, higher processing speeds are possible along with a significant reduction in energy consumption. Studies have shown that the chip consumes less than 1 watt of power, which is far less than existing GPUs and GPUs. Intel predicts that neuromorphic processing can be useful for search engines. Use the robotics industry and solve complex problems.
In addition to improving the chip's architecture, Intel also unveiled the Lava software framework. Intel says this is an open, modular framework that allows developers to use a common set of tools, methods, and libraries. Apart from the robotics industry, self-driving cars can also benefit from such powerful chips, and Ford Motor Corp., for example, is one of the chip's customers.
- What is machine learning and why is it so important? li>
Source: Digital Trends