The 10 fastest supercomputers in the world

The computers we use in our daily lives are getting faster every year. But those with features ridiculous enough to fall into the supercomputer category will always surpass them.

While technology is constantly evolving, the list of the world’s fastest supercomputers has also changed due to various innovations. Soon, The list you can see below is based on data listed on the Top500 website.The system, which works by measuring the power of computers, has been updated until June 2023.

10.Tianhe-2A

Tianhe-2A was developed by the National University of Defense Technology of China and It is one of the oldest supercomputers on this list. It was launched in 2013 and remains active at the National Supercomputing Center in Guangzhou, China.

Its components include Intel Xeon processors and Matrix-2000 coprocessors, and it can reach a processing capacity of 61.44 petaFLOPS. Another important point is that its construction was carried out with the efforts of approximately 1,300 scientists and engineers.

Those responsible for the supercomputer even tried to improve its capabilities in 2015, but the North American government blocked Intel from supplying its hardware to Chinese organizations.

9.Selene

Unlike other supercomputers that appear on the list, One of Selene’s differences is this: Thiswith Built during the COVID-19 pandemic. However, it has made some improvements since then, reaching ninth place in the Top500 list.

Selene can currently reach speeds of 63.46 petaFLOPS. Additionally, it features NVIDIA SuperPod A100 DGX modular architecture as well as AMD Epyc processors and 80 GB A100 GPUs.

Another important detail is that it was produced by NVIDIA in order to deepen studies in the field of artificial intelligence and help chip design.

8.Perlmutter

In the new ranking prepared by Top500, we have Perlmutter in eighth place. It is located at the US National Energy Research Center and was created as a result of a partnership between HPE, NVIDIA and AMD.

It is considered one of the most powerful artificial intelligence supercomputers in the world. Among its components we find around 1,500 AMD EPYC Milan processors in addition to 6,159 NVIDIA A100 GPUs.

In terms of speed, it has a processing capacity of 70.87 petaFLOPS.

7. Sunway Taihu Light

We continue our relationship with Sunway TaihuLigt. This system, developed by the National Research Center for Parallel Computer Engineering and Technology in China, is located at the National Supercomputer Center in Wuxi.

This supercomputer features Sunway’s 260-core SW26010 processors that help it achieve 93,014 petaFLOPS.

It is also worth noting that Recognized as the world’s fastest supercomputer between June 2016 and June 2018. However, when the Summit was implemented, he was dethroned.

6.Sierra

Created by IBM, Sierra is used at the Lawrence Livermore National Laboratory in California, United States. It is a computer focused on Homeland Security issues that is used to check the safety and effectiveness of nuclear weapons in Uncle Sam’s territory.

It is currently capable of reaching 94,640 petaFLOPS using two Power9 processors and four NVidia V100 GPUs. Another point is that these video cards are connected to each of the 4,320 nodes.

5. Summit

Created by IBM, Summit is being installed at Oak Ridge National Laboratory, where another supercomputer on this list is also located. Due to its features, it was the fastest equipment in the world until June 2020, but it has already been surpassed by four rivals.

This supercomputer It is used in applications focused on energy, artificial intelligence and human health.It has a processing capacity of 148,600 petaFLOPS. It also has two Power9 processors and six Tesla V100 GPUs made by NVidia.

Finally, a curiosity: Due to its size, the summit takes up approximately 55 meters of space and uses approximately 219 kilometers of cable to provide all the necessary connections.

4. Leonardo

The fourth place in the Top500 supercomputer list is Leonardo, located in Bologna, Italy. It is operated by Cineca, a consortium of 65 universities, 27 public research institutions and two Italian ministries.

The device, which is used to create applications for sectors such as artificial intelligence, medicine, renewable energy and bioengineering, has a processing capacity of 238.70 petaFLOPS.

The system of this supercomputer is based on the BullSequana XH200 architecture, it also has two computing modules, a total of 3,500 Intel Xeon processors and 14 thousand Nvidia A100 model GPUs.

3. LUMI

Located in Finland, LUMI is the third supercomputer on the list compiled by the Top500 website. It belongs to the European High Performance Computing Consortium EuroHPC and is considered a pre-exascale system.

In the evaluation made by the Top500 team, it can reach 309.10 petaFLOPS. Its components include the HPE Cray EX235a architecture, as well as optimized third-generation AMD EPYC processors and AMD Instinct MI250X GPU.

An interesting fact about LUMI is that Works with 100% hydroelectric energyIn addition to having an advanced cooling system. Another important point is that some of the heat it produces is used to heat some areas in the area where it is installed.

2. Fugaku

Fugaku has long been known as the world’s fastest supercomputer. And when you look at its specifications, you can understand why it has been at the top of this list for a long time: It is capable of reaching speeds of 442,010 petaFLOPS.

Manufactured by Fujitsu, this system features the 48-core SoC A64FX manufactured by Fujitsu, making it the first system powered by ARM processors to reach the top. It has a total of 7,630,848 total cores.

Fugaku is currently installed at Riken’s computational science center in Kobe, Japan.

1. Border

Developed by Oak Ridge National Laboratory, Frontier is currently known as the world’s fastest supercomputer. To give an idea of ​​the magnitude of its power, it was the first to reach the “exascale” limit and could perform 1.1 quintillion operations per second.

As for its specifications, it is known to use the HPE Cray EX235a architecture in addition to AMD EPYC processors and AMD Instinct GPU. It has a total of 8,699,904 cores and an energy efficiency rating of 52.23 gigaflops/watt.

“Based on the latest HPE Cray EX235a architecture and powered by AMD EPYC 64C 2 GHz processors, the system has 8,699,904 total cores, a power efficiency rating of 52.23 gigaflops/watt, and features gigabit Ethernet for data transfer. Frontier is currently being integrated and “The TOP 500 group is being tested at ORNL in Tennessee, USA, which will be operated by the Department of Energy,” he said.

So, have you ever dreamed of having so many supercomputers? To follow the content TecMundo, follow here and follow us on our social network. To the next one!