CHINA TOPIX

11/21/2024 09:04:33 pm

Make CT Your Homepage

Facebook Open Sources Design for AI Server Big Sur

Facebook

(Photo : Getty Images) Facebook's Free Basics service has come under fire in India.

Social networking giant Facebook has announced that it is releasing the hardware design of a server that it uses to train artificial intelligence software. This will allow other companies to explore their potential in terms of building A.I. systems.

The server dubbed "Big Sur" is used by Facebook in order to run its machine learning programs. These type of programs have the capability to learn and can improve in performance over a certain period of time.

Like Us on Facebook

Facebook will release Big Sur to the Open Computer Project, an initiative launched by the social networking company in order to allow other companies to share their designs for new hardware.

In a statement obtained by Wired, Facebook engineering director of AI group Serkan Piantino said, "Our commitment to open source is something that individuals who work here are passionate about. Having that be a part of our culture is a benefit when it comes to hiring."

Machine learning technology is also used by tech giants like Microsoft and Google. One field where machine learning can be applied is in image recognition. In this application, machine learning software studies a photo or video in order to identify the object within the frame. Aside from image recognition, the technology is also being applied to large data sheet in order to detect fraud and email spams.

Unlike traditional computers that rely heavily on CPU computing power, Big Sur relies on the computing power of GPUs or Graphics Processing Unit. The server can house up to eight high-performance GPUs that can each consume up to 300 watts.

According to Computer World, Big Sur uses GPUs manufactured by Nvidia, one of the leading company in the graphics card market. Facebook claims that GPU-based system, like Big Sur, is twice as fast as previous generation of hardware. Big Sur also uses the Tesla Accelerated Computing Platform.

Real Time Analytics