Share This Article
Meta Platforms plans to develop the RSC supercomputer by the end of the year
Meta has joined forces with Nvidia to create the RSC supercomputer. According to the developers’ plans, the machine will be the most powerful among all existing ones and will be engaged in research in the field of artificial intelligence.
The supercomputer will be housed in Meta’s data center and should be completed by the end of 2022. Nvidia Corporation is supplying the equipment to build the RSC. The computer is already training new models based on artificial intelligence. According to Meta, the RSC will soon build more advanced solutions using AI, which will be trained on trillions of patterns. The supercomputer will analyze huge amounts of data in multiple languages. It will be able to process video, audio, and text content and detect malicious information. Thus the machine will be used to provide a high level of security for the users of Meta products.
In addition to excellent performance, RSC will be a reliable assistant for various AI models. For this, Meta has engaged Nvidia, which is one of the best manufacturers of graphics processors. The companies have been cooperating in a number of areas for a long time.
The creators of the supercomputer are confident that they will be able to use it to create real-time voice translation. This will eliminate restrictions for communication or interaction between people from different countries.
There are 760 systems from Nvidia already in use for the supercomputer, acting as compute nodes. In total, the machine requires 6080 GPUs, which are included in the single Nvidia Quantum network. So far, it has been possible to reach the maximum performance level of the computer at 1.9 exaflops. However, the developers plan to increase it to 5 exaflops by the end of this year. This is quite an ambitious plan, given that at the moment the most powerful supercomputer is Fugaku from Japan, its performance is 2.15 exaflops.
Meanwhile, the demand for such machines continues to grow. Supercomputers are used in many areas – both for scientific developments, and to improve the efficiency of production processes. For example, the U.S. Department of Energy has signed an agreement to supply a supercomputer called Polaris, for the creation of which Nvidia and AMD chips were used. In addition, back in 2019, the U.S. government signed a contract with Intel Corporation to develop a supercomputer for the national energy laboratory. However, so far the machine has not begun to function – the reason was a global shortage of microchips and failures in the supply of necessary materials. Due to these force majeure circumstances, the commissioning has been postponed until this year.