A few days ago, the US Department of Commerce announced that American enterprises were prohibited from selling electronic components, equipment and software technology to ZTE within seven years, which once again led the industry to focus on the development of domestic chips. Today, Alibaba Dharma Institute revealed to the media that it is currently studying a neural network chip called Ali NPU.

According to the researchers of Dharma Institute, The performance of Ali NPU is ten times that of mainstream CPUs The cost and power consumption of GPU based AI chips are only half of them, and the cost performance ratio is more than 40 times. Alibaba can apply this chip to various industries through the output of cloud computing capacity.

It is reported that at present, the chip R&D team of the institution is distributed in the United States and Shanghai, with a total of dozens of people, which should increase to 100 by the end of the year. Ali has also invested in several chip companies. The research and development of Ali NPU will accelerate the application of AI intelligence in business scenarios, improve efficiency and reduce costs.

It should be noted that, Ali NPU is a neural network chip that specializes in AI reasoning and operation problems such as image and video recognition, cloud computing, etc., and cannot replace CPU for the time being, GPU and other general-purpose chips have played a role in boosting the "China Chip", but the road for domestic chips is still long.

Follow our Weibo @ Love Computer

Follow our WeChat official account: playphone

Of course, we also pay attention to our Bilibili account: love computer