It would be the world’s first zetta-class supercomputer.
Japan is building a super-fast computer, the first of its kind, that will be a thousand times faster than any computer we have now. It will be ready for use in 2030 and could cost more than $780 million to make. This new computer will help Japan remain at the forefront of artificial intelligence (AI) development.
Development of a successor to the country’s flagship supercomputer, Fugaku, will begin in 2025, according to Japan’s Ministry of Education, Culture, Sports, Science and Technology (MEXT). The supercomputer could reach speeds on a zetaFLOPS scale, which has never happened before. achieved previously.
According to Living science, “Floating-point operations per second (FLOPS) is used to measure how quickly computers can solve problems, where one floating-point operation is a single calculation. A supercomputer with a speed of 1 zetaFLOPS could create one trillion (1 followed by 21 zeros). ) calculations per second. Today’s most powerful supercomputers have only just broken the exaFLOPS barrier, meaning they can perform just over a trillion (1 followed by 18 zeros) calculations per second.
The Japanese news site Nikkei stated in a translated article that the decision to create such a powerful machine was made “to keep pace with the development of scientific research using artificial intelligence.”
According to ScienceAlert, supercomputers have proven consistently useful to scientists, helping researchers simulate black holes, discover new materials, model the Earth’s future, and explore the foundations of mathematics. As these machines become more and more powerful, we should see their capabilities increase as well. Unlike quantum computers, supercomputers are not much different from the desktops and laptops we all use every day; they just scaled up to an incredibly high level. They are still based on processors, memory and storage, but taken to the extreme.
An upgraded zetta-class machine would be able to train with more data faster and produce results that are more detailed, accurate, and comprehensive. If all goes well, there should be a new supercomputing standard in six years.