September 9, 2021
(Reuters) – Facebook Inc develops machine learning chips to handle tasks such as recommending content to users. The information was reported Thursday, citing two people familiar with the project.
According to the report, the company has developed another chip for video transcoding to improve the viewing experience of recorded and livestreamed videos in the app. https://www.theinformation.com/articles/facebook-develops-new-machine-learning-chip
Facebook’s move is increasingly abandoning traditional tech providers such as Apple Inc Amazon.com Inc and Alphabet Inc’s Google to design their own chips to save money and improve performance. Because it is. (Https://reut.rs/3E0NlVN)
In the 2019 blog https://engineering.fb.com/2019/03/14/data-center-engineering/accelerating-infrastructure, Facebook will handle AI inference and video transcoding to improve performance He said he is building a custom chip design specifically aimed at. , The power and efficiency of its infrastructure. At that time, it served 2.7 billion people on all platforms.
The company also said it would work with semiconductor players such as Qualcomm Inc, Intel Corp and Marvell Technology to build these custom chips. A general purpose processor alone is not enough to manage the amount of workload that Facebook’s system handles.
However, according to a report from The Information, Facebook designed these chips entirely in-house and is not backed by these companies.
“Facebook is constantly looking for ways to drive higher levels of computing performance and power efficiency, both with our silicon partners and through our internal efforts,” said a spokeswoman for Facebook.
(Report by Chavi Mehta in Bangalore, edited by Anil D’Silva)
Facebook Develops Machine Learning Chip-Information
Source link Facebook Develops Machine Learning Chip-Information