Auto-translation used

Supercomputers of the world

Against the background of the news about the beginning of the creation of the Kazakh supercomputer

Modern supercomputers are unique research and computing tools capable of performing an unimaginable number of computational operations. They are colossal machines designed to process huge amounts of data and perform complex calculations in record time intervals.

The power of these supercomputers is measured in trillions and quadrillions of computing operations per second, which provides them with the ability to solve problems of science and engineering at unprecedented levels of efficiency. Such high performance allows supercomputers to be used in a wide variety of fields, from modeling complex climatic phenomena to the study of quantum chemistry and biomedical calculations.

The term "trillions" and "quadrillions" of computing operations reflects the grand scale of their work, making supercomputers an integral part of cutting-edge scientific research and technological advances. They become catalysts for innovation, helping to solve complex research problems that require outstanding computing power and efficiency.

So what kind of Supercomputers are there now? 

Consider the TOP 5 supercomputers in the world:

1 . OLCF-5 (Hewlett Packard Enterprise Frontier) - also known as Frontier, is one of the most powerful supercomputers in the world. It was designed to carry out complex scientific research and engineering calculations. The OLCF-5 is the first supercomputer in the world with an exaflops performance level, which means that it is capable of performing more than a quadrillion computing operations per second. Located in the USA in the data center  Oak Ridge Leadership Computing Facility, located in Oak Ridge, Tennessee.

2. Aurora (Intel), a supercomputer with a power exceeding 1 exaflops, is the second most powerful supercomputer in the world. The Aurora is powered by 63,744 Intel Data Center GPU Max processors of the Ponte Vecchio series and 21,248 Intel Xeon CPU Max processors of the Sapphire Rapids series. The supercomputer also includes more than 1024 data storage nodes with a total capacity of 220 petabytes and a maximum throughput of 31 terabytes per second. Located in the USA.

3. Eagle (Microsoft) - powered by Intel Xeon Platinum 8480 °C processors and NVIDIA H100 accelerators, and has reached the HPL index of 561 petaflops/s.

4. Fugaku - the project is almost entirely funded by the Japanese government — the annual budget is about 90 billion yen ($800 million). Fugaku's power is about 442 petaflops, although the energy consumption is much higher than Frontier — almost 30 MW. The entire system is based on the PRIMEHPC FX1000 — a kind of "node", which is a closed rack with blade servers inside.   Inside there is a Fujitsu A64FX central processor based on the ARM architecture, combining 52 processor cores and several auxiliary ones. There are about 150,000 such cores in the entire Fugaku.  

5. LUMI (Large Unified Modern Infrastructure) – the term "lumi" in Finnish translates as "snow". Despite its fifth place in the global ranking, LUMI is the leading among all European supercomputers.

The LUMI supercomputer is located in the extensive complex of the SCS IT center in the small town of Kayaani, Finland. This scientific center belongs to the state and is serviced by higher educational institutions of the country.

Inside, LUMI is equipped with 32 terabytes of RAM, 7 petabytes of flash memory, as well as 80 petabytes of traditional storage, in all cases the Lustre file system is used, designed for cluster computing with a high level of parallelism.

The additional 30 petabytes are intended for a data management service based on the Ceph file system, which also effectively divides file space between cluster nodes.

The total amount of LUMI memory is 117 petabytes with a total I/O bandwidth of 2 terabytes per second.

An example of research using LUMI is the artificial intelligence project implemented by the Tampere University research group aimed at the diagnosis and classification of prostate cancer.

Thanks to the use of LUMI, it was possible to reduce the calculation time from three days to three hours, significantly speeding up the testing and research processes.

The LUMI supercomputer is fully powered by the local Ämmäkoski hydroelectric power station, and the heat generated is used to heat 20% of the centralized heat supply system of the city of Kajaani. In this regard, LUMI can be considered one of the most environmentally efficient supercomputers in the world.

And if you want to see even more examples of supercomputers, then I recommend the excellent TOP500 website.

Comments 1

Login to leave a comment