Nvidia's interest in buying ARM has to do with the expected growth of edge computing. Traditionally, Nvidia's graphic processors have been used in centralized cloud data centers. But edge computing will shift many processing chores to devices or premises-based servers or close to the network edge, where low power draw and small footprint are important.
Graphics processing unit improvement seems to be occurring faster than Moore’s Law would predict. Nvidia’s GPUs today are 25 times faster than five years ago. If they were advancing according to Moore’s Law, they would have increased their speed only by a factor of 10, says Nvidia CEO Jensen Huang.
Between November 2012 and May 2020, Nvidia’s chip performance increased 317 times for a major class of AI computing, says Bill Dally, chief scientist and senior vice president of research at Nvidia. On average, the performance of these chips has more than doubled each year.
No comments:
Post a Comment