Image: Diane Bryant, Executive Vice President, Intel Data Center Group.
Silicon Valley is not less enthusiastic about artificial intelligence, deep learning neural networks and machine learning, as the world's largest chip manufacturer, Intel's strength in this competition is rapidly increasing.
Intel President Diane Bryant recently announced that the company is developing its next-generation high-end server chip product, Xeon Phi, for AI applications.
Baidu's data center for the Deep Speech platform will use the Xeon Phi chip to maximize the resolution speed of speech in natural language.
Bryant says that by 2020, there will be more servers for data analysis than any other task. Intel's chips have long been known for processing numbers at high speeds, but in recent years, Nvidia graphics chips have far surpassed Intel's advantages in neural network servers. The former's chips are more efficient at processing unstructured data such as video and speech and recognition models.
As a result, Intel began to invest more resources in central processing units (CPUs) to enable it to handle more deep learning tasks. Intel promises that CPU boosts, especially the Xeon Phi, are their answer. The new chip, code-named Knights Mill, will be available in 2017.
Image: Intel's Diane Bryant and Baidu's Wang Jin
Last week, Intel acquired Nervana, a San Diego-based deep learning startup, for more than $350 million. The team will help Intel improve cloud applications and R&D frameworks for deep learning on multiple levels. Jason Waxman, corporate vice president of Intel Cloud Computing, said in an interview with VentureBeat that Nervana will fully power Intel's AI cause.
Intel believes that the performance level of the Xeon Phi chip will be comparable to Nvidia's graphics processing chip. Of course, Nvidia wants to be different and expressed that view in a https://blogs.nvidia.com/blog/2016/08/16/correcting-some-mistakes/ blog post the other day. Bryant says that the processor's speed of accessing memory has been significantly improved, resulting in significant improvements in CPU-based processing performance, which is important when the task volume increases.
Intel is also working with the National Center for Energy Research, Scientific Computing, which hopes to optimize machine learning capabilities at scale.
Baidu senior vice president Wang Jin said: The next era is the era of artificial intelligence, technology is the power to change human life, Baidu is very much looking forward to the use of AI for speech and natural language processing.
Image: The new Xeon Phi.