Recently, Ningchang information industry (Beijing) Co., Ltd. (hereinafter referred to as “Ningchang”) said that it has been invited to participate in NVIDIA’s 2020 GTC China online summit, which will be held from December 15 to 17, when it will bring a number of AI servers.
Photo: page of Ning Chang’s participation in GTC China online summit 2020
As an important partner of NVIDIA in China, Ningchang is currently carrying NVIDIA V100 ™ TensorCoreGPU、NVIDIAT4 ™ Ningchang AI server of nvida products such as tensorcore GPU has been widely used in well-known Internet and cloud computing companies to support AI scenarios such as semantic recognition, image analysis, machine translation, training and reasoning.
Ning Chang engineer said that Ning Chang AI server products have won the recognition of many IT companies in a short time, mainly relying on its customized services to meet the multiple needs of users for operation and maintenance, configuration, application scenario optimization, etc.
Customize AI server to facilitate IDC operation and maintenance
With the rapid development of AI (Artificial Intelligence) industry, GPU server has occupied a large number of IDC space. Compared with the general CPU server, AI server’s high calorific value, high IO and other characteristics are making it the “focus” of IDC operation and maintenance personnel.
In the face of thousands of servers, the platform based “intelligent management” has become the mainstream mode of IDC operation and maintenance supervision. How to bring heterogeneous AI servers into the “unified” management platform is a problem that every Internet manufacturer must consider when purchasing AI servers.
According to the introduction of Ning Chang engineer, in view of the pain point of “operation and maintenance management” of users, Ning Chang can configure BIOS, BMC and many other supervision and operation and maintenance interfaces according to the operation and maintenance standard of user data center when AI server leaves the factory. Users only need to connect the power supply and network cable to bring the batch of Ningchang AI servers into the existing intelligent management platform, which is simple and convenient.
In addition to software and interface standard customization of the bottom layer, Ningchang AI server can also reasonably arrange the location of hard disk and network card according to the actual layout and configuration of customer IDC when it leaves the factory, and synchronously optimize the cooling air duct, so as to achieve the goal of reducing the user’s TCO (total cost of ownership) by more than 10%.
In addition to providing customized services, relying on a comprehensive product, technology and engineering platform, Ningchang AI server has the ability to flexibly adapt to different AI scenarios, and can meet the needs of users for “flexible topology” computing architecture.
Won 30 mlperf performance first
Ning Chang engineer introduced that taking Ning Chang X640 series artificial intelligence server as an example, it takes into account super computing power, extremely high storage and strong data exchange capability. It can realize flexible GPU topology through hardware link re layout, and optimize GPU interconnection for different application scenarios and models.
Picture: Ningchang artificial intelligence X640 series server
Ningchang X640 can support 8 NVIDIA V100 for training applications ™ TensorCoreGPU/NVIDIA®QuadroRTX ™ 6000/NVIDIA®QuadroRTX ™ 8000 full speed 7 * 24h. For reasoning applications, support 16 NVIDIA T4 ™ Tensor core GPU provides online big data processing.
After fully equipped with various high-density GPUs, Ningchang X640 still has rich PCI-E interfaces for network, storage and other IO expansion, and further optimizes the operation and maintenance efficiency. The professional management platform can realize the identification, monitoring, fan speed control support, fault alarm and other functions of various mainstream GPUs in the market.
On October 22 this year, it was equipped with NVIDIA V100 / A100 ™ Ningchang x640g30ai server has achieved 30 world first results in more than 10 benchmarks such as resnet50, SSD, bet and DLRM of mlperf, the industry’s authoritative AI benchmark performance evaluation platform.
Illustration: comparison of test scores of x640g30 in 16 card configuration
Nowadays, Ningchang AI service products have won many well-known enterprise projects in IT industry, covering Internet, telecom, cloud computing, finance and other fields.
If you are interested in Ning Chang customized AI server, please call 400-898-1688 or send an email to: [email protected] . Nettrix Ningchang information industry (Beijing) Co., Ltd., a server fine customization manufacturer, is looking forward to serving you.