Dawn and NVIDIA Collaborate to Launch Multiple GPU Servers

On December 13, 2012, Shuguang and NVIDIA released three GPU server products in Beijing, namely TC4600 GPU blade server, Tiankuo W580I-G10 server and Tiankuo I620-G15 server. Mr. Wang Zhengfu, Chief Operating Officer of Shuguang Co., Ltd., Dr. Dai Rong, Director of High Performance Products, Mr. Xu Tao, High Performance Product Manager, and Vice President of NVIDIA Corporation; Mr. Deng Peizhi, Senior Project Manager of Ashok Pandey, NVIDIACUDA and HPC in China, NVIDIA China Mr. Guo Aimin, sales manager, attended the new GPU server conference.

As we all know, Dawning has been deeply pursuing 16 years in the server market, continuously innovating independently, and developing high-, medium-, and low-end product lines. It is the leading HPC solution provider with outstanding characteristics and strong strength. In 2010, with GPU technology just emerging, with the powerful computing power of NVIDIA Tesla GPU, Shuguang developed the world’s first high-performance computer with measured performance exceeding petaflops and ranking second in the world’s TOP500. —Dawning "Nebula". Since then, GPU servers have become one of Dawning's innovations, and they have continuously integrated advanced design concepts and technical features to develop industry-leading GPU server series products to provide users with higher computing power - Chief Operating Officer of Shuguang. Mr. Wang Zhengfu said.

Three arrows and accurate positioning of the user's application needs According to report, the three GPU servers launched by Dawning are based on the upgrading and replacement of the original products. The new products can support the latest NVIDIA Tesla M2090/K10/K20 products. NVIDIA Tesla is the fastest and most efficient high-performance computing GPU in the world. With NVIDIA Tesla, product developers and researchers can achieve improved performance and significantly lower usage barriers. NVIDIA Tesla can help users quickly and accurately solve the world's most important computing problems in all walks of life. Currently, Dawning has formed a relatively comprehensive GPU product line that can meet the needs of users for high-performance computing.

In terms of products, the TC4600 GPU blade server launched by Dawning has high-performance computing as the main application target, incorporating many advanced design concepts and technical features, and is the industry's leading GPU blade server product. Based on the TC4600 platform independently developed by Sugon, it can support up to 8 GPU acceleration units (4 GPU blades) and 12 InteltelEmplex EP processors (4 GPU blades + 2 regular blades) in 5U space. At the same time, the TC4600 GPU blade server achieves high-performance, high-density, scalable, on-demand configuration, and flexible solution product design requirements to meet the diverse and dynamically changing application needs of users. TC4600 GPU service products are aimed at computationally intensive users, such as the processing of satellite remote sensing data, numerical simulation of meteorological and oceanic environments, and petroleum exploration data processing. Such applications are characterized by a high requirement for floating-point computing capabilities.

The Tiankuo I620-G15 server released at the same time is a high-performance dual-channel GPU server developed by Shuguang based on Intel's latest Romley platform, which can achieve up to 80% performance improvement over the previous generation. The I620-G15 can support a maximum memory capacity of 512GB. The internal cross-card design can meet the requirements of a full-height NVIDIA Tesla GPU accelerator. This powerful scalability is sufficient to support mission-critical operations and meet the needs of resource-intensive applications. More PCI-E slots and higher-speed PCI-E buses allow users to easily expand. The I620-G15 server is ideal for industrial data centers and remote enterprise environments where financial, securities, transportation, postal, telecommunications, energy, and other requirements are critical to server performance, scalability, and reliability.

The Tiankuo W580I-G10 workstation is a new generation of dual-station workstations with excellent performance, stable and reliable configuration and flexible configuration. It is also the best node for GPU high-performance computing. The excellent system architecture ensures that users A single system can support four mainstream NVIDIA Tesla computing accelerator cards in the industry, and supports high-speed interconnect networks such as 10G or Infiniband, providing a high-density, high-performance node platform for GPU computing clusters. Not only that, Tiankuo W580I-G10 workstation is also compatible with the current mainstream 32-bit and 64-bit applications, and specifically optimized for CAD, simulation, animation and CG rendering, GPU computing and other applications, mainly for movies, design, animation , Construction, machinery, oil and other high-end graphic processing industry users.

GPU computing has become a trend to provide users with a package of HPC GPU solutions It is reported that GPU heterogeneous computing has become the trend of high-performance computing. In the latest TOP500 list, the reporter observed that most of the machines are using NVIDIA Tesla GPU acceleration to achieve extremely high computing power, such as the fastest computer "Titan", with about 90% of its performance coming from over. Million NVIDIA Tesla K20x GPU Accelerator. In the face of technological trends, Dawning, as NVIDIA's only OEM partner in China, will make full use of this advantage and in-depth cooperation in design, R&D, technical cooperation, market development, etc., and timely launch new products to meet market demand. Dawning not only intensified efforts to develop GPU series products, but also applied GPU's powerful computing performance to the solution.

Development of aerospace equipment, satellite remote sensing data processing, weather forecasting, marine environment numerical simulation, petroleum exploration data processing, biomedical research, financial engineering data analysis, new material development and design, computational structural mechanics, life science calculation, fluid dynamics NVIDIA Tesla GPU computing is playing a powerful role in the areas where high floating-point computing capabilities are required, computer vision, data mining, and numerical analysis. Compared to the traditional CPU solution, the NVIDIA Tesla GPU solution has the advantages of low price and high speed, and it is undoubtedly more satisfying for users.

As the first group army that domestic HPC deserves, Dawning has won the domestic HPC TOP100 championship for four consecutive years and has helped China continue to refresh its achievements on the world stage. In the above fields, with its strong HPC technology capabilities and rich experience accumulated in HPC solutions, Dawning can provide users with a package of HPC GPU solutions and advance the users to solve practical problems.

With the continuous development of GPU technology, the application area of ​​GPU is also expanding. Many application software industries are slowly moving closer to the GPU. In the future applications, GPU virtualization will open up a new world for the field of three-dimensional stereoscopic earthquake protection. GPU's "cloud" concept.