Meta is building the ‘world’s fastest AI supercomputer’ to help Facebook in its transformation into a ‘metaverse’, it has announced today.
The tech giant claims that the AI Research SuperCluster (RSC) will become the fastest supercomputer on the planet when it’s complete by 2022.
RSC will pave the way toward building technology for ‘the next major computing platform’ – the metaverse.
Facebook (the company, not the product) renamed itself Meta in October, as part of its long-term project to turn its social media platform into a metaverse.
The social media platform will become accessible in the metaverse via smart glasses and virtual reality headsets.
Meta claims that RSC will assist its scientists in building better AI models, which can be used across many languages and develop new AR tools. It also helps them seamlessly analyze text, video, and images simultaneously.
It could lead to new AI systems capable of powering real-time voice translators to large amounts of people in metaverse. Each person speaks a different language so they can easily collaborate on a project or play AR together.
Meta announced today that it is building California’s “world’s fastest AI computer” to assist Facebook with its transition into the’metaverse”.
Meta’s supercomputer, AI Research SuperCluster (pictured), will become the most powerful supercomputer on the planet when it is fully constructed by the middle of 2022.
RSC will pave the way toward building technology for ‘the next major computing platform’ – the metaverse (pictured)
In a blog published today, Kevin Lee, Meta’s director of product management, and Shubho Sengupta, a software engineer for AI Research, said the new supercomputer is already partly operational but will be fully built by the middle of this year.
MailOnline has reached out to Meta for more information about its precise location.
RSC can train an AI model with many billions parameters in less than three weeks. It is also a third faster than what Meta takes to train it.
Meta said that to create the next generation in advanced AI requires powerful, new computers capable running quintillions per second.
‘Today, Meta is announcing that we’ve designed and built the AI Research SuperCluster (RSC) – which we believe is among the fastest AI supercomputers running today and will be the fastest AI supercomputer in the world when it’s fully built out in mid-2022.
‘We wanted this infrastructure to be able to train models with more than a trillion parameters on data sets as large as an exabyte – which, to provide a sense of scale, is the equivalent of 36,000 years of high-quality video.’
According to them, meta researchers already use RSC for large-scale models of natural language processing (NLP), as well as computer vision in research.
NLP is the development of machines that can understand text and voice data and then respond with speech or text in their own words – much like human beings.
They say that RSC has already been used by meta researchers to train large models for natural language processing (NLP), and computer vision research.
The AI Research SuperCluster (RSC) is a walkthrough for meta engineers. The new supercomputer is party built but will be fully built by the middle of this year
RSC contains equipment made by Nvidia (US technology company), including Nvidia high-end graphics processor units (GPUs). These electronic circuits break down complex problems into millions or even billions of tasks, and then work out the solutions simultaneously.
AI supercomputers can be built using multiple graphics processing units (GPUs), which are combined into compute nodes. These are connected via a strong network that allows for fast communication among the GPUs.
‘RSC today comprises a total of 760 Nvidia DGX A100 systems as its compute nodes, for a total of 6,080 GPUs – with each A100 GPU being more powerful than the V100 used in our previous system,’ Sengupta and Lee say.
“We’ll be working to raise the number GPUs from 6,080 – 16,000 by 2022. This will boost AI training performance 2.5 times.”
Because of the coronavirus pandemic, RSC began as a completely remote project that the team took from a simple shared document to a functioning ‘cluster’ – a set of computers that work together – in about a year and a half.
‘Covid-19 and industry-wide wafer supply constraints also brought supply chain issues that made it difficult to get everything from chips to components like optics and GPUs, and even construction materials – all of which had to be transported in accordance with new safety protocols,’ they say.
“To create this cluster efficiently we had to design the entire thing from scratch. This involved creating several new Meta-specific conventions as well as rethinking existing ones.
According to Meta experts, the development of next-generation advanced AI requires powerful computers that can perform quintillions per second.
Meta has built high-powered computing infrastructure for large-scale models. For years, the Meta AI Research Team has worked to build these systems.
Meta states that it has made long-term investments in AI since the creation of its Facebook AI Research laboratory in 2013.
It’s been a step beyond the original focus with Mark Zuckerberg’s goal to make the metaverse a reality.
“Metaverse”, a term used to refer to shared immersive spaces that can be accessed on multiple platforms, where physical and digital intersect, was coined by the dystopian novel “Snow Crash” in 1992.
Zuckerberg was the co-founder of Facebook, and he described it as an “embodied internet” when he created it in his Harvard University dormitory in 2004.
Facebook users can use the platform via a VR headset within years.
Mark Zuckerberg (pictured) plans for social media platform to be a metaverse’.
Rather than swiping the screen of a device, they could potentially meet up with a Facebook friend in a virtual shared space – like an ultra-realistic simulation of another planet or an idyllic garden – and vocally chat to each other’s avatars.
According to the company, “It will be marked by social presence and the feeling that one is right there with another person no matter where they happen to be,”
“The metaverse may still be a long way off but some parts are here, and more are coming soon.”