Home > News content

Accelerating high-performance computing revolution second generation AMD EPYC pushing industry into "computing power era"

via:驱动之家     time:2020/3/11 8:30:51     readed:513

During the research work of the sudden new coronavirus (COVID-19), researchers first need to sequence the virus samples in order to make it possible to use bioinformatics method to compare and search according to the sequencing results, to find the source and the maximum probability host of the virus transmission, and to determine the next prevention, control and treatment means according to the characteristics of the virus. With the development of high-throughput sequencing technology, various biological data show explosive growth, DNA sequencing data will double in an average of four ~ five months. In the face of such a rapid growth of large short-read sequence data sets, how to effectively manage, analyze and utilize them has become an urgent problem in bioinformatics development, and high-performance computing (HPC) will help researchers solve this challenge.

More broadly, computing-intensive applications such as oil, meteorology, CAE、 nuclear energy, pharmaceuticals, environmental monitoring and analysis, and system simulation; data-intensive applications such as libraries, banks, securities, taxation, decision support systems; and communications-intensive applications such as websites, information centres, search engines, telecommunications, streaming media HPC are helping humans to solve technical challenges that require powerful computing power through simulation, modelling and analysis, leading them to the next stage

With the continuous development of information and data, HPC has been adopted by research scientists who initially need to perform complex mathematical calculations, and now it is gradually popularized, winning the favor of more and more enterprise organizations. In the face of explosive data growth and booming data collection and analysis in the information age, the demand for HPC computing power in all fields of society is growing day by day.

High performance computing

nea

Many business and IT leaders believe that HPC systems are based on supercomputers. indeed, although HPC are widely used in supercomputers produced by companies such as Atos、IBM、HPE、Cray and Fujitsu, there is a more widely used way to integrate multiple minicomputers into interconnected clusters to provide high-performance computing functionality. This method is widely used in various social production fields, effectively improving the enterprise information processing capacity and data calculation rate.

In the face of exponential data growth, no matter how it works,

The second generation AMD epyc processor leads the new standard of high performance computing

The second-generation AMD EPYC processor, the world's first seven-tier server processor, offers HPC users an unparalleled combination of perfect performance, including record floating-point performance, ground-breaking architecture, ultra-high number of x86 cores, and super memory and I/O bandwidth. Judging from the performance and market feedback of the 140 world records it broke, the second generation EPYC of up to 64 cores is undoubtedly the most cutting-edge product in the current market that caters to the computing power of HPC users.

On the performance side, the second generation of AMD EPYC has brought HPC users

For the security that HPC users value, the second generation epyc also achieves the ultimate in security defense. Amd epyc enhances the security of hardware core through advanced security functions. By taking the lead in integrating the special security processor, it provides the foundation for security functions such as secure boot, secure memory encryption (SME) and secure encryption Virtualization (SeV). For example, 509 encryption keys that can be used for security encryption virtualization, AES-128 encryption engine built in the memory controller, and security memory encryption functions that can be implemented without changing the software through simple BIOS settings, reduce users' concerns about data risks and enable them to focus more on business operations.

In addition, the chip design architecture based on the amdinfinity fabric bus technology guarantees the continuous innovation of users with the help of epyc. The chip design adopted by the second generation of AMD epyc is a hybrid multi chip architecture, which has reached a new height in the application of the second generation of AMD epyc processor. Chip design architecture is divided into two parts: eight chips as the processor core, each chip has eight x86 computing cores, and one I / O chip is responsible for processor security and external communication. This flexible design not only enables CPU core to adopt advanced process technology, but also enables I / O circuit to develop according to its own laws. It is also because of this non integrated chip design that users can use up to 64 core ultra-high computing performance earlier with the help of epyc.

The powerful HPC performance attracts the world's leading high-performance computing users to join the AMD epyc camp. AMD is working with the Oak Ridge National Laboratory of the U.S. Department of energy to develop the supercomputer frontier, which is expected to become the most powerful supercomputer in the world when it is delivered in 2021. AMD, together with Lawrence Livermore National Laboratory (LLNL) and HPE, adopts the next generation of AMD epyc CPU and AMD radeon insight GPU, as well as the open-source AMD ROCM heterogeneous computing software. It is planned to deliver El Capitan system in early 2023, with more than 2 exaflops (10 billion times) of double precision performance, which is expected to become the fastest supercomputer in the new world after delivery. Atos, the global leader in digital transformation, is providing two bullsequana xh2000 supercomputers based on the second generation epyc to the French meteorological agency for dynamic weather forecasting and research in atmospheric, marine and Climate Sciences. In addition, HPE, Dell technology group, Zurich Federal Institute of technology, San Diego Supercomputing Center, etc. all announced the launch or use of second-generation epyc supercomputers.

Partner ecology continues to expand, helping users upgrade to cloud and AI supercomputing

With the progress of technology, HPC is combining with cloud computing, machine learning, big data analysis, artificial intelligence and other technologies to promote high-performance computing to a new stage. It is necessary to build new infrastructure to cross HPC with innovative services, which requires higher performance of processor and ecosystem.

Since the second generation of AMD epyc processor was released, it has received strong support from partners in various fields around the world. After the release, there have been original design suppliers such as Gigabyte and QCT, independent hardware suppliers (IHV) such as Broadcom, Meguiar and Xilinx, Microsoft and multiple Linux operating system suppliers, as well as mainstream cloud providers such as Tencent cloud, AWS, Microsoft azure, Google cloud and Oracle cloud.

For the HPC requirements in the cloud environment, the azure HB cloud instance based on the first generation of AMD epyc processor earlier in Microsoft azure has achieved a level of computational fluid dynamics (CFD) performance that previously could not be matched. Now azure hbv2 virtual machine for high-performance computing has provided preview, further breaking through the boundary of cloud high-performance computing. Google also recently launched a test version of the n2d virtual machine on Google's computing engine that uses the second generation AMD epyc (Xiaolong) processor, bringing customers a top-notch high-performance computing experience.

With the further development of cloud computing, container based and cloud based workloads gradually dominate the HPC environment. AMD's close cooperation with mainstream cloud providers will ensure that users can get HPC supercomputing performance at a lower cost in the cloud.

On the other hand, in view of the higher requirements of the development of artificial intelligence for computing power, AMD is also sparing no effort to cooperate with hardware partners to meet the needs of HPC system heterogeneous computing platform. Compared with traditional CPU, heterogeneous computing has higher parallel floating-point computing efficiency, higher peak processing capacity, higher throughput, and lower latency. The preparation in this respect also leaves a great imagination space for AMD to play in HPC field.

In China, the world's largest semiconductor consumer market, huge computational power is the underlying basis for supporting information technology upgrading. In the face of the explosive data growth in the information age and the booming demand for data collection, analysis and so on, the demand for computing power in various industries has never been higher. China's researchers have made remarkable achievements in a short period of time, and have made great contributions to controlling and eventually defeating the epidemic.

China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments