The firm additionally plans a brand new AI-optimised knowledge centre design and the second part of its 16,000 GPU supercomputer for AI analysis.
Meta is constructing its first-generation customized silicon chip for working synthetic intelligence (AI) fashions, saying its AI compute wants will develop dramatically over the following decade as we break new floor in AI analysis.
Meta (previously Facebook) is constructing its first-generation customized silicon chip for working synthetic intelligence (AI) fashions, saying its AI compute wants will develop dramatically over the following decade as we break new floor in AI analysis.
Called MTIA (Meta Training and Inference Accelerator), the in-house, customized accelerator chip will present larger compute energy and effectivity than CPUs, and is customised for inside workloads.
“By deploying each MTIA chips and GPUs, we’ll ship higher efficiency, decreased latency and larger effectivity for every workload,” said Santosh Janardhan, VP and Head of Infrastructure at Meta.
The company also plans a new AI-optimised data centre design and the second phase of its 16,000 GPU supercomputer for AI research.
“These efforts — and additional projects still underway — will enable us to develop larger, more sophisticated AI models and then deploy them efficiently at scale,” Janardhan added.
The next-generation knowledge centre will likely be an AI-optimised design, supporting liquid-cooled AI {hardware} and a high-performance AI community connecting 1000’s of AI chips collectively for knowledge centre-scale AI coaching clusters.
“It may also be quicker and less expensive to construct, and it’ll complement different new {hardware} resembling our first in-house-developed ASIC resolution, MSVP (Meta Scalable Video Processor), which is designed to energy the continuously rising video workloads at Meta,” Janardhan informed.
Meta’s Research SuperCluster (RSC) AI supercomputer, which the company believes is one of the fastest AI supercomputers in the world, was built to train the next generation of large AI models to power new augmented reality tools, content understanding systems, real-time translation technology and more.
It features 16,000 GPUs, all accessible across the 3-level Clos network fabric that provides full bandwidth to each of the 2,000 training systems.
“Custom-designing much of our infrastructure enables us to optimize an end-to-end experience from the physical layer to the virtual layer to the software layer to the actual user experience,” mentioned Meta.
(This story has not been edited by News18 employees and is revealed from a syndicated information company feed – IANS)
