Intel introduced the first processor for artificial intelligence systems

For operation of artificial intelligence, as we know, the huge computing power required. Now, as the electronic components used are often the most common components (with the exception of the most powerful systems like IBM Watson). As a rule, used Intel Core solutions from consumer to Xeon. But all of them were created for several other tasks that do not involve working with neural networks. Therefore the company has decided to develop a processor that will be sharpened it to work with AI systems.

Intel introduced the first processor for artificial intelligence systems

The first dedicated processor for the AI ​​from Intel called Nervana Neural Processor (NNP), and the release of the first models is planned for the end of this year. The new processor is Nervana ASIC-chip (application-specific integrated circuit), and it aims primarily to perform tasks in the field of machine learning and deep learning. At the core of the processor data bus are providing bi-directional exchange of large volumes of information. Thus, it will be possible to combine several Nervana processors into one with little or no loss of performance. This approach enables the processor to operate with the multi-level neural networks. In addition, as stated by representatives of Intel,

"The main feature of Nervana NNP processor is that at the heart of his work is math with integers reduced accuracy Flexpoint. The advantage of its use is the performance gains when performing tasks, reducing the delay and increase the data transmission rate of the external buses. In addition, Nervana can execute in hardware function matrix addition, finding intersections and a number of other operations. A special algorithm allows to dynamically change processor memory so that its architecture is optimal for a particular operation. "

Despite the huge experience in the field of Intel processors creation, when entering the chip market for AI, it will have to face the likes of Qualcomm, IBM and the NVIDIA, which have long been engaged in the production of components in this area. So watch this race will be very interesting.