Meta Broadcasts New Particulars on Information Facilities for AI Tasks

Meta Broadcasts New Particulars on Information Facilities for AI Tasks

Meta Platforms on Thursday shared new particulars on tasks it was pursuing to make its knowledge facilities higher suited to supporting synthetic intelligence work, together with a customized chip “household” that it stated it was growing in-house.

The Fb and Instagram proprietor stated in a sequence of weblog posts that it designed a first-generation chip in 2020 as a part of the Meta Coaching and Inference Accelerator (MTIA) program, which was aimed toward bettering effectivity for the suggestions fashions it makes use of to serve adverts and different content material in information feeds.

Reuters beforehand reported that the corporate was not planning to deploy its first in-house AI chip broadly and was already engaged on a successor. The weblog posts portrayed the primary MTIA chip as a studying alternative.

“From this preliminary program, we’ve got realized invaluable classes that we’re incorporating into our roadmap,” it wrote.

The primary MTIA chip was centered completely on an AI course of known as inference, by which algorithms skilled on big quantities of information make judgments about whether or not to indicate, say, a dance video or a cat meme as the subsequent publish in a person’s feed, the posts stated.

A Meta spokesperson declined to touch upon deployment timelines or elaborate on the corporate’s plans to develop chips that would prepare the fashions as nicely.

Meta has been engaged in a large mission to improve its AI infrastructure this previous yr, after executives realized it lacked the {hardware} and software program wanted to assist demand from product groups constructing AI-powered options.

As a part of that, the corporate scrapped plans for a large-scale rollout of an in-house inference chip and began work on a extra formidable chip able to performing each coaching and inference, in response to the Reuters reporting.

Meta acknowledged in its weblog posts that its first MTIA chip stumbled with high-complexity AI fashions, though it stated the chip dealt with low- and medium-complexity fashions extra effectively than competitor chips.

The MTIA chip additionally used solely 25 watts of energy — a fraction of what market-leading chips from suppliers reminiscent of Nvidia devour — and used an open-source chip structure known as RISC-V, Meta stated. 

Along with detailing its chip work, Meta offered an replace on plans to revamp its knowledge facilities round extra fashionable AI-oriented networking and cooling techniques, saying it could break floor on its first such facility this yr.

The brand new design could be 31 p.c cheaper and might be constructed twice as rapidly as the corporate’s present knowledge facilities, an worker stated in a video explaining the modifications.

© Thomson Reuters 2023

Google I/O 2023 noticed the search big repeatedly inform us that it cares about AI, alongside the launch of its first foldable cellphone and Pixel-branded pill. This yr, the corporate goes to supercharge its apps, companies, and Android working system with AI expertise. We focus on this and extra on Orbital, the Devices 360 podcast. Orbital is out there on Spotify, Gaana, JioSaavn, Google Podcasts, Apple Podcasts, Amazon Music and wherever you get your podcasts.
Affiliate hyperlinks could also be routinely generated – see our ethics assertion for particulars.

Supply hyperlink


Please enter your comment!
Please enter your name here