Intel has created a chip for processing neural networks

Last year, processor giant Intel introduced the Nervana platform for artificial intelligence applications. Well, by the end of this year, Intel promises to show a new family of processors belonging to the platform, Intel Nervana Neural Network Processor (NNP), designed for processing neural networks. NNP chips will speed up the process of training deep learning models. With their help, it will be possible to develop completely different classes of AI applications and maximize the amount of data processed. The foundations of NNP did not originate in Intel laboratories. Recently, the processor giant has been actively buying up companies known for their development of hardware for artificial intelligence. Intel's successful acquisitions include Mobileye and Nervana Systems. It was Nervana Systems projects that became the basis for Neural Network Processors, formerly known as Lake Crest. Intel calls its brainchild “the industry’s first neural network processing chip,” but in reality this is not entirely true. The workloads created by modern AI objects are much better handled by GPUs than by classic CPUs. Today, AI data processing products are produced by Nvidia and ARM. Google didn’t want to lag behind either, developing its own neural network processor Tensor Processing Unit . So far, Intel has not provided any information when the new product will become available to potential buyers.