Facebook parent company Meta unveiled a new supercomputer capable of building next-generation artificial intelligence infrastructure and foundational technology to boost research in the field of metaverse.
Called AI Research SuperCluster, the new machine is among the fastest AI supercomputers running today and will be the fastest AI supercomputer in the world when it is complete in mid-2022, Meta said.
The company’s researchers are using RSC to train large models in natural language processing. It will be used to build better AI models that can learn from trillions of examples, develop new augmented reality tools, work across hundreds of languages and seamlessly analyse text, images and videos together.
Using the new supercomputer, Meta aims to build a new set of AI systems that can power real-time voice translations to large groups of people, each speaking a different language, so they can collaborate on a research project or play an augmented reality game together, the company said.
Supercomputers are far more powerful than general-purpose computers and are typically used to address the most demanding problems in the world. These include developing medicine, fighting the Covid-19 pandemic, exploring oil and gas reserves, simulating nuclear weapon explosions, and forecasting weather.
“Ultimately, the work done with RSC will pave the way for building technologies for the next major computing platform – the metaverse, where AI-driven applications and products will play an important role,” Kevin Lee, RSC’s technical programme manager, and Shubho Sengupta, software engineer in the company’s AI research infra team, wrote in a blog on Monday.
To expedite its research on Metaverse, Meta launched a $50 million fund last September.
Metaverse is a digital space that allows users to communicate and move virtually in their three-dimensional avatars or digital representations. Hailed as a successor to the internet, it is a set of immersive spaces shared by users, in which they can interact, innovate and engage other people who are not in the same physical space.
RSC will let Meta fully realise the benefits of self-supervised learning and perform tasks that require the processing of large sets of data in a fraction of second. For instance, it will enhance speech recognition technology, which must work well even in challenging scenarios with lots of background noise, such as parties or concerts.
“NLP needs to understand more languages, dialects and accents … and advances in other areas, including robotics, embodied AI and multimodal AI, will help people accomplish useful tasks in the real world.
“High-performance computing infrastructure is a critical component in training large models, and Meta’s AI research team has been building these high-powered systems for many years,” said Mr Lee and Mr Sengupta.
US technology companies such as Microsoft, IBM and Google are competing to develop supercomputer technology.
In 2020, Group 42, an artificial intelligence and cloud computing company in Abu Dhabi, offered Artemis – the world’s 26th most powerful supercomputer – to researchers looking to contain the coronavirus.
Last year, Saudi Aramco, the world’s biggest oil company, and Saudi Telecom Company, the biggest telecoms operator in the kingdom by market value, launched a supercomputer, Dammam 7, that was featured among the world’s top 10 most powerful machines.
In May 2020, Microsoft announced it had built one of the top five publicly disclosed supercomputers.
Meta said it faced Covid-induced “unexpected challenges” in the development of RSC.
“Covid-19 and industry-wide wafer supply constraints brought supply chain issues that made it difficult to get everything from chips to components like optics and GPUs [graphics processing units], and even construction materials – all of which had to be transported in accordance with new safety protocols,” Mr Lee and Mr Sengupta said.
How safe is RSC?
Most of the RSC tasks, including building new AI models and creating new algorithms, involve the use of real-world data. RSC has been “designed from the ground up with privacy and security in mind”, the company said.
“RSC is isolated from the larger internet, with no direct inbound or outbound connections, and traffic can flow only from Meta’s production data centres … the entire data path is end-to-end encrypted.
“Before data is imported to RSC, it must go through a privacy review process to confirm it has been correctly anonymised,” Mr Lee and Mr Sengupta said.
RSC is up and running today, but its development is ongoing.
Currently, RSC involves 6,080 GPUs and the company aims to increase the number to 16,000. It will increase the AI training performance by more than 2.5 times.
A GPU is an electronic circuit that processes many pieces of data simultaneously.