[ad_1]
January 24, 2022 – Meta Platforms gave an enormous thumbs as much as NVIDIA, selecting the applied sciences for what it believes will probably be its strongest analysis system to this point.
The AI Analysis SuperCluster (RSC), introduced , is already coaching new fashions to advance AI. As soon as absolutely deployed, Meta’s RSC is anticipated to be one of many largest buyer set up of NVIDIA DGX A100 techniques.
“We hope RSC will assist us construct totally new AI techniques that may, for instance, energy real-time voice translations to giant teams of individuals, every talking a special language, so they might seamlessly collaborate on a analysis venture or play an AR sport collectively,” the corporate says in a weblog.
Coaching AI’s fashions
When RSC is absolutely constructed out, later this 12 months, Meta goals to make use of it to coach AI fashions with greater than a trillion parameters. That would advance fields resembling natural-language processing for jobs like figuring out dangerous content material in actual time. Along with efficiency at scale, Meta cited excessive reliability, safety, privateness and the pliability to deal with “a variety of AI fashions” as its key standards for RSC.
Beneath the hood
The brand new AI supercomputer at the moment makes use of 760 NVIDIA DGX A100 techniques as its compute nodes. They pack a complete of 6,080 NVIDIA A100 GPUs linked on an NVIDIA Quantum 200Gb/s InfiniBand community to ship 1,895 petaflops of TF32 efficiency.
Regardless of challenges from COVID-19, RSC took simply 18 months to go from an concept on paper to a working AI supercomputer thanks partially to the NVIDIA DGX A100 know-how on the basis of Meta RSC.
Penguin Computing is our NVIDIA Accomplice Community supply companion for RSC. Along with the 760 DGX A100 techniques and InfiniBand networking, Penguin offered managed providers and AI-optimised infrastructure for Meta comprised of 46 petabytes of cache storage with its Altus techniques. Pure Storage FlashBlade and FlashArray//C present the extremely performant and scalable all-flash storage capabilities wanted to energy RSC.
20x efficiency positive factors
It’s the second time Meta has picked NVIDIA applied sciences as the bottom for its analysis infrastructure. In 2017, Meta constructed the primary technology of this infrastructure for AI analysis with 22,000 NVIDIA V100 Tensor Core GPUs that handles 35,000 AI coaching jobs a day.
Meta’s early benchmarks confirmed RSC can practice giant NLP fashions 3x quicker and run pc imaginative and prescient jobs 20x quicker than the prior system.
In a second section later this 12 months, RSC will develop to 16,000 GPUs that Meta believes will ship a whopping 5 exaflops of blended precision AI efficiency. And Meta goals to develop RSC’s storage system to ship as much as an exabyte of information at 16 terabytes per second.
A scalable structure
NVIDIA AI applied sciences can be found to enterprises of any dimension. NVIDIA DGX, which features a full stack of NVIDIA AI software program, scales simply from a single system to a DGX SuperPOD operating on-premises or at a colocation supplier. Prospects also can hire DGX techniques by NVIDIA DGX Foundry.
Touch upon this text under or through Twitter: @IoTNow_OR @jcIoTnow
[ad_2]
