The reason for seeking AI chips in mobile phones is simple. Just like the desktop computers we use every day, ordinary cpus are very bad for machine learning; This aspect requires a lot of small computing, but the CPU can only use a limited number of cores to compute, and usually, this work needs to rely on nearly a thousand cores of the graphics card to complete. But with AI, you can increase the number of times the CPU cores work in sync. In this regard, Gary Brotman, head of AI at Qualcomm, believes that \"parallelization is definitely the key to the future core, and CPU execution will become powerful and efficient as a result.\"
< br >
As consumers, in theory, AI chips can make mobile phones have better performance and longer battery life. At present, many machine learning services (such as voice assistants) need to send your data to the cloud for analysis and calculation, and the data exchange in the middle of the process hides the user's privacy information. The iPhone X introduced by Apple said that because of the AI chip, these calculations can be carried out directly on the phone, saving this step, which means reducing the risk of data leakage. In addition, the original server computing in the cloud can also free up some energy to do some more useful calculations.
AI chips are also known as AI accelerators or compute cards, which are modules specifically designed to handle a large number of computational tasks in AI applications (other non-computational tasks are still handled by the CPU). At present, AI chips are mainly divided into GPU, FPGA, and ASIC.
al chip refers to a processor that can handle general AI tasks and has core intellectual property (IP). It is a common processor that integrates operational AI algorithms; It is a processor that efficiently improves the efficiency and iteration ability of one or more speech and image.