HPC

  • Why Most HPC Systems Use InfiniBand Interconnection

    In addition to the well-known Ethernet, there are many other categories of network architecture. For server-side connection scenarios, InfiniBand (IB) technology is valued and used for its inherent characteristics. It has a nearly dominant position, especially in High-Performance Computing (HPC), large data center storage and other scenarios. So what’s the difference between IB and Ethernet? Why most HPC systems use IB interconnection? What InfiniBand Is and Where It Is Used IB is a “cable switching” technology that supports multiple concurrent connections, and it is the I/O standard of the new generation server hardware platform. With the rapid development of CPU performance, the performance of I/O system has become a bottleneck restricting the performance of servers. The PCI bus architecture used in the past does not conform to the new application trend. To overcome the inherent drawbacks of PCI, Intel, Cisco, Compaq, EMC, Fujitsu and other companies jointly launched the IB architecture, the core of which is to separate the I/O system from the server host. At present, only a few companies, such as Mellanox, Intel, Qlogic, can provide IB products….

    April 11, 2019 0 0 0
  • QSFP-DD Might Be the Mainstream Form-factor of 400G Optical Transceivers

    Time to enter 2019, when 400G has become a hot topic in the optical communications industry, the world’s leading optical transceiver manufacturers have launched their own 400G optical modules. When we list the form-factors of these manufacturers’ 400G optical modules (as shown in the figure below), we found that all the manufacturers except the Finisar (acquired by II-VI) have adopted the QSFP-DD form-factor — the market seems to have recognized QSFP-DD as the first choice for form-factors of 400G optical modules, though some manufacturers have also introduced 400G optical modules with OSFP and CFP8 form-factors. 400G Form-factors of Mainstream Optical Transceivers Manufacturers Tips: QSFP-DD is a high-speed pluggable module form-factor defined by the QSFP-DD MSA group. The QSFP-DD MSA group has defined the next generation, high-density, high-speed pluggable module form factor that addresses the industry need for high-density, high-speed networking solutions in a backward compatible form factor. The QSFP-DD Specification has been developed and refined by many companies within the QSFP-DD MSA group and released to the public. Why do mainstream manufacturers choose the QSFP-DD form-factor? Does this mean that the…

    March 6, 2019 1 0 11
  • The New Darling of HPC Data Center — 100G QSFP28 CLR4 Optical Module

    With the advent of 5G era and the rapid development of AI industry, HPC data centers supporting intelligent algorithms and in-depth learning will be more applied to the business field. For data center interconnection, the research and development of optical transceiver modules applied in HPC scenario has to be paid attention by practitioners. As a global optical interconnection design innovator, the Gigalight company recently launched such a HPC optical transceiver module — 100G QSFP28 CLR4. The Background of the Birth of 100G CLR4 The 100G QSFP28 CWDM4 transceiver is currently the most widely used optical module in data centers, such as Facebook, Alibaba and Tencent, as the preferred option for optical interconnection within 500m to 2km distances. Due to the low cost of CWDM4 lasers and Mux/Demux devices, the 100G CWDM4 solution is more competitive in 2km transmission scenario than 100GABSE-LR4 solution. However, the application of CWDM4 in the field of HPC is often stretched, because HPC data centers not only need optical modules to meet the requirements of large bandwidth, but also the characteristics of low delay and strict signal transmission…

    November 14, 2018 0 0 0
  • HPC Data Center Is Heading for a New Era

    Recently, AMD released the world’s first 7nm data center CPU “Rome”. Subsequently, Amazon announced that it would provide related virtual machines from now on, claiming that its cost-effective ratio is 10% higher than that of Intel virtual machines. Prior to this, AMD has already scramble to release the world’s first 7 nanometer GPU before NVIDIA. Aiming at the data center market, AMD’s two attacks are tantamount to a more direct competition for Intel and NVIDIA at the same time. Dr. Lisa Su, president and CEO of AMD, said, “Our multi-year investment in data center hardware and software roadmaps has pushed our CPU and GPU to be adopted by more and more cloud computing, enterprise computing and high performance computing customers. In the next quarter, we believe that this trend will accelerate with the introduction of CPU and GPU product portfolio, the most extensive and powerful data center with industry-leading 7nm process technology.” AMD believes that they will have a $29 billion opportunity in data centers by 2021. As the industry turns to artificial intelligence and machine learning, GPU is playing…

    November 12, 2018 0 0 0
  • HPC On Cloud? HPC On Cloud!

    In recent years, two brothers, High Performance Computing (HPC) and cloud computing, have always appeared in various occasions at the same time, and a new term has emerged — HPC On Cloud. What is HPC On Cloud? Why is there HPC On Cloud? Let’s talk about it together. Cloud Computing vs. HPC Let’s take a look at what their concepts are. Cloud computing is a kind of usage mode that distributes according to demand and charges according to usage. Users can put hard disks and CPUs on their computers or company servers on the Internet and make unified dynamic calls. They only need to pay by hour or even minute, instead of buying computers and servers by themselves. For example, cloud computing is similar to tap water at home, so we don’t need to build a water plant in our own home but just need to turn on the tap and distribute it on demand by paying. HPC is applied the area of large-scale data processing. Take Baidu’s project team as an example, HPC is used in many departments, such as speech…

    November 9, 2018 0 0 0
  • What Are Supercomputing, High-Performance Computing and Cloud Computing?

    In recent years, the terms “supercomputing”, “high performance computing” and “cloud computing” have often been talked about, but most people do not seriously understand what they are and what the difference is. This article will answer these questions in detail for you. Supercomputing is based on supercomputer. Supercomputers are computers capable of supercomputing. Supercomputing uses computers to research, design products and support complex decisions. In addition to the most advanced computing hardware systems, it also includes software systems and testing tools, and algorithms to solve complex computing. At present, China’s supercomputers are in the leading position in the world. In last year’s supercomputer 500 strong selection, China’s “Sunway TaihuLight” won the championship. Although such a giant has been developed, the specific application of supercomputers in China is still in a relatively backward position compared with the United States and Germany. High Performance Computing, referred to as HPC, is the application of high-performance computers. High-performance computers are parallel computers composed of traditional ultra high-speed computers and multiple CPU. However, generally speaking, HPC is almost equivalent to supercomputing, mainly for scientific computing,…

    November 5, 2018 0 0 3

Contact Us

Email: sunwf@gigalight.com