Facebook’s parent company Meta today announced that it has built one of the world’s fastest supercomputers. In addition, the company says that this supercomputer, dubbed as the AI Research Supercluster (RSC), will be the ‘fastest in the world’ once fully built in mid-2022.
At its core, RSC has a total of 6,080 NVIDIA A100 GPUs, making a total of 760 NVIDIA DGX A100 systems. Meta says that these GPUs are more powerful than the V100 GPUs that it uses in its current systems. Each of these DGXs (NVIDIA’s servers and workstations that use GPUs for deep learning applications) communicate through an NVIDIA Quantum 1600 Gb/s InfiniBand two-layer Close Fabric. And for storage, the RSC has 175 petabytes of pure storage FlashArray, 46 petabytes of cache storage in the Penguin Computing Altus system, and 10 petabytes of pure storage FlashBlade.
When RSC is complete, InfiniBand Network Fabric will add 16,000 GPUs as endpoints (as opposed to 6,080 GPUs deployed now), which the company says will make it one of the largest such networks ever deployed . Additionally, the company says it plans to upgrade the caching and storage system to handle one exabyte of data per second, against the current 16 TB/s.
But why build such a machine?
Keeping rights aside, Meta has built RCS for two specific purposes. First, of course, is building the metaverse. Meta wrote in a blog post, “Our long-term investments in building the next generation AI infrastructure with self-supervised learning and RSC are helping us to create the foundational technologies that will power the Metaverse and also advance the broader AI community.” Will increase.”
In addition to helping develop future technologies, RCS will help the company tackle some of its current issues – misinformation and harmful content.
So far, Meta uses a combination of human supervision and AI models to identify misinformation and harmful content on its platforms, which include Facebook, Instagram and WhatsApp. But clearly this is not enough. Misleading content containing harmful information continues to slip on its platform despite monitoring. To tackle this, Meta seeks to develop AI models that can take into account millions of parameters to analyze content, including a combination of text, images, graphics, video and voice, and flag or any discourse. Removing what it finds dangerous. META’s current systems cannot handle such complex models in a way that they are capable of delivering results in record time, right now. This is where RCS comes into play.
RSC will help META’s AI researchers build better AI models that can learn from trillions of examples that use multimodal signals to determine whether an action, sound or image is harmful or benign. “This research … will help keep people safe on our services today,” the company said.
Which brings us to the most important question – what is a supercomputer?
In simple words, a supercomputer is a computer that can perform millions of tasks in a second. Not only can it handle millions of queries at a time but it can also solve complex problems that would otherwise take a general purpose computer years to solve within record time depending on the complexity of the problem).
IBM describes supercomputers as the world’s fastest computers made up of interconnections, I/O systems, memory, and processor cores. The company explains that unlike traditional computers, supercomputers use more than one central processing unit (CPU), grouped into compute nodes, which contain a single processor or a group of processors and a memory block. A supercomputer can contain thousands of nodes that communicate and collaborate with each other to solve specific problems.
But what about the GPU in Meta’s RCS?
Nowadays, companies are using GPU or graphics processing unit instead of CPU to make supercomputers. GPUs are already being used in smartphones, monitors and smart TVs and PCs to handle graphics well. Now, highly powerful versions of GPUs, also known as general purpose GPUs or GPGPUs, are being used to build supercomputers.
There is a reason for this change.
GPUs use parallel processing to perform any task. This means that the GPU can handle multiple tasks at a time without the success or failure of one process affecting the other. This parallel processing nature makes them extremely efficient in handling workloads and fast in performing computations and assigned tasks. Using GPU instead of CPU can reduce the time taken to perform any computation, which is why they are being used in supercomputers these days.
For comparison, a CPU-based supercomputer called Jaguar was upgraded in 2012 with NVIDIA’s Tesla K20 GPU. After the upgrade the name of the supercomputer was changed to Titan. Its performance went from 2.3 petaflops to over 20 petaflops (the unit in which supercomputer performance is measured). The Titan became about ten times faster and five times more energy efficient than the Jaguar, while fitting inside the same 200 cabinet as its predecessor.
What operating systems do supercomputers run?
Unlike general-purpose computers running Apple’s macOS or Windows or even Google’s ChromeOS, supercomputers mostly run the Linux operating system due to their open-source nature. Supercomputers are designed for specific purposes and vary greatly in configuration. Therefore, designing and maintaining a proprietary OS for such machines is time-consuming and costly.
Linux, on the other hand, is free to use and easy to customize, which is why it is used in the case of supercomputers.
What are supercomputers used for?
Supercomputers are used for research and development purposes in various fields, including weather forecasting, space research, testing the power of encryption, and even in the development of drugs for various diseases. For example, IBM is using a supercomputer – the Summit – consisting of 16 systems with more than 330 petaflops, 775,000 CPU cores and 34,000 GPUs, to help researchers understand COVID-19 and find its treatments and potential cures. be able to help.
The summit, which houses one of the most energy efficient supercomputers in the world, will help medical researchers understand America’s cancer population and develop anti-cancer drugs. In addition, IBM’s Sierra and Summit supercomputers are also being used by researchers to identify patterns in human proteins and cellular systems for developing drugs for diseases such as Alzheimer’s, heart disease and addiction. Sierra is also being used to assess the performance of nuclear weapon systems.
The Piz Dent supercomputer, developed by Cray and deployed at the Swiss National Supercomputing Center in Switzerland, is used for a variety of purposes, including analyzing data collected from the Large Hadron Collider.
Post Explained: What is a supercomputer and where is it used? Appeared first on BGR India.