How China Built an Exascale Computer Using Old Tech

China has built an exascale computer using a mix of old and new technology, and it’s an impressive feat of engineering.

Checkout this video:

Introduction

China is set to achieve a remarkable feat in the world of high-performance computing. The country is building the first exascale supercomputer, which will be capable of a billion-billion calculations per second.

This is an impressive feat of engineering, but it’s also worth noting that China is using old technology to achieve it. The heart of the system will be based on chips that are four years old.

The chips in question are Sunway TaihuLight SW26010 processors. These are 64-bit ARMv8 chips that were first released in 2016. Each chip has 256GB of DDR4 memory and 16GB of flash memory.

The Chinese Exascale Computer

China has built the world’s fastest computer, an exascale supercomputer that is twice as fast as the previous record holder. The Chinese Exascale Computer is made up of old, slower hardware that has been repurposed and linked together to create a powerful machine.

The Design

The Chinese Exascale Computer is a supercomputer that was designed and built by the Chinese Academy of Sciences. It is currently the world’s most powerful supercomputer, with a peak performance of 1 exaflops.

The computer is based on a Chinese-designed microprocessor called the Loongson-3A, which is a derivative of the MIPS architecture. The Loongson-3A was first used in the Tianhe-1A supercomputer, which became operational in 2010 and was ranked as the world’s second most powerful supercomputer at that time.

The Chinese Exascale Computer is housed in a data center in Wuxi, China. It consists of 40,960 Loongson-3A processors, arranged in 256 racks. Each rack contains 160 processors, for a total of 40,960 processors.

The computer has a memory capacity of 1 petabyte (1 million gigabytes). It uses a custom-designed interconnect called the “Sunway Dragonfly” to connect the nodes. The Sunway Dragonfly has a bandwidth of 100 gigabits per second and a latency of 1 microsecond.

The Chinese Exascale Computer was designed for general purpose computing, with an emphasis on scientific and engineering applications. It is being used for astrophysics research, climate modeling, earthquake simulations, and other scientific research tasks.

The Hardware

The Chinese Exascale Computer is a supercomputer that was completed in 2019. It is made up of four cabinets, each containing 1,024 processors. The processors are based on a 32-bit architecture and run at 3 GHz. The supercomputer also has 1 TB of DRAM and 16 PB of flash storage.

The Exascale Computer is powered by an AC power supply and uses cooling fans to keep the processor cabinets cool. The supercomputer is housed in a Building 35,000 square meter data center.

The U.S. Exascale Computer

The U.S. is years behind China in the development of an exascale computer, which is a machine that can perform one billion billion calculations per second. The U.S. Department of Energy has been working on an exascale computer for years, but the project has been plagued by delays. China, on the other hand, has been able to build an exascale computer using older technology.

The Design

The U.S. Exascale Computing Project (ECP) is a Department of Energy (DOE) initiative to deliver a capable exascale computing ecosystem by the end of 2023 that will enable scientific leadership for the nation and accelerate the delivery of approximately 50 exascale systems to U.S. research laboratories and universities.

The project’s overarching goal is to ensure continued U.S. leadership inHigh Performance Computing (HPC) by delivering an exascale ecosystem that is fully integrated, more interoperable, and more productive than anything that has been previously built or operated. This will provideAmerican researchers with the computational power and software tools needed to address many of the most complex scientific and engineering challenges of our time, like developing new clean energy solutions, improving disease diagnosis and treatment, designing safer aircraft and automobiles, and advancing scientific understanding of Earth’s climate.

To achieve its goals, ECP is working with key partners across nine DOE National Laboratories, five DOE Office of Science User Facilities, three DOE Strategic Partnerships for Research Areas (SPRA) programs, as well as academia and industry on three primary fronts: hardware technology development; system software development; and applications enablement.

The Hardware

The hardware powering China’s exascale computer is a departure from the hardware used in the U.S. The U.S. has long favored using graphics processing units (GPUs) from NVIDIA Corporation to speed up scientific calculations. But in China, GPU chips from Nvidia’s main rival, Advanced Micro Devices, are being used instead.

Instead of using custom-made chips, China is using a type of chip called an FPGA, or field-programmable gate array. FPGAs can be reconfigured to perform different tasks, so they’re well suited for research because scientists can quickly adapt them for new projects. China has also been working on its own FPGA chips, which it plans to use in future exascale computers.

But the use of FPGAs isn’t the only difference between the U.S. and Chinese exascale computers. The U.S. government has strict rules about sharing technology with China, which has made it difficult for American companies to do business there. As a result, most of the hardware and software used in China’s exascale computer was developed and built in China.

Comparison

When China announced last year that it had built the world’s first exascale supercomputer, it was a major milestone in the country’s quest to catch up with the united states in high-performance computing.

But while China may have been first to exascale, it did not achieve this feat with the most cutting-edge technology. Instead, the country relied on a type of processor that is years behind the state of the art.

The processor in question is called an Arm chip, and it is used in many of our smartphones and other devices. But while Arm chips are powerful enough for tasks like browsing the web and playing games, they are not as powerful as the chips used in high-end supercomputers.

While China’s use of Arm chips may seem like a strange choice, there are actually several reasons why the country chose to use them. For one thing, Arm chips are much cheaper than the alternatives. They also use less power, which is important for a country like China that has limited resources.

But perhaps the most important reason why China chose to use Arm chips is because they are easy to mass produce. This is because Arm chips are made using a process called “TSMC 28nm” which is a older and less expensive manufacturing process.

While China’s decision to use old technology may seem counterintuitive, it actually makes perfect sense when you consider the country’s goals. For China, building an exascale supercomputer was never about building the most powerful machine possible. Instead, it was about proving that the country could catch up with the United States in high-performance computing. And by using old technology, China was able to do just that.

Conclusion

Though China may not have the same technology as the United States when it comes to building an exascale computer, they have been able to utilize older technology in order to create a machine that is just as powerful. This is a huge accomplishment, and it goes to show that China is a force to be reckoned with in the world of computing.

Scroll to Top