IBM develops an innovative computer chip that imitates the brain and promises to become the fastest in performing artificial intelligence tasks. The new chip called NorthPole is a digital architecture compatible with silicon chips used in modern computers.
northpole differs from traditional chips in that it combines computational ability with memory blocks that store information.
Researchers from IBM They demonstrated how Northpole is able to effectively perform image recognition tasks based on artificial intelligence with high energy efficiency and speed, ahead of this even the latest chips from the leading manufacturer of the NVIDIA graphic processors. In addition, he successfully copes with the problems of speech recognition and processing of the natural language. Northpole also provides better performance on a transistor – tiny electronic switches that underlie computing technology. The chip makes 22 billion transistors in space only 800 square millimeters.
However, such a specialized design of Northpole limits its capabilities, since it is not able to perform artificial intelligence and is not suitable for working with large models. Despite this, the Modha team plans to show how several Northpole chips can be used to support large language models.
These developments are important for future artificial intelligence and computer technology. On the one hand, they confirm that IBM continues to strive to increase the effectiveness of calculations, inspired by biological principles. On the other hand, they demonstrate a variety of approaches to the implementation of the hardware level of artificial intelligence and the lack of a universal solution suitable for all tasks. Perhaps in the future we will see more hybrid systems that combine various types of chips in order to better comply with the various needs of artificial intelligence.