Fuxi Laboratory Launches AI Innovative Products, Ping An Technology (PAT) and Inspur Jointly Develop the Strongest AI Host in the Financial Cloud

2019-04-03 22:08 870

SHANGHAI, April 3, 2019 /PRNewswire/ -- With the publication of the Thirteenth Five-Year Plan for the Development of Information Technology in China's Financial Industry, it has become a general trend for banks to employ the cloud. Meanwhile, its demand for improving the ability of servers to integrate resources has greatly increased. As the high-tech core of Ping An Group (PAG), PAT is committed to applying advanced technologies such as AI, cloud computing and block chain to meet the growing demand for financial cloud servers, thus enabling the group and ecological partners to promote the intelligent upgrading of financial and other industries. Founded by PAT and Inspur Group, Fuxi Laboratory, combines the advantages of both sides and launches the world's leading Ping An Cloud B1 ultra-high performance AI financial cloud host (hereinafter referred to as PAC B1), which can be widely applied to massive data computation, analysis and mining in AI applications. Besides it has become one of the most powerful AI compute hosts in the world.

Based on computational empowerment and scenario integration, Fuxi Laboratory, is committed to creating world-leading innovative cloud products and promoting the integration of scenario-based frontier technologies in financial field, thus enabling the five ecospheres of financial services, medical health, automobile services, real estate finance and smart cities to achieve ecological win-win situation. Ultra-high performance AI financial cloud host is one of the three research directions of Fuxi Laboratory in the future. Deeply integrated PAG's world leading AI application R&D capabilities and the Inspur Group's global leading AI software, hardware products and its R&D capabilities, the latest launched PAC B1 is known as the most powerful financial AI solution in the world and leads the development trend of AI technology in the financial industry.

The Most Powerful AI Compute Host in the World

In terms of computational performance, PAC B1 has sixteen NVIDIA Tesla V100 (32GB) GPU accelerators with the strongest performance at present, which can support the next generation of stronger GPU accelerators and provides a very powerful AI computing performance of up to 2 thousand trillion per second. At the same time, PAC B1 also has two powerful CPUs with 24 cores, which can provide the most top-notch general computing performance for users.

In terms of data throughput performance, based on NVIDIA's latest HGX-2™ platform, PAC B1 adopts the most advanced NVSwitch™ interconnection structure in the industry. It can achieve 48 channels, 2.4TB/s high-speed interconnection of the whole chip group, and 512GBHBM2 global sharing high-speed graphics cache. It also provides nearly linear AI computing performance ratio, and achieves a comprehensive improvement in data throughput speed.

High cost performance gains the initiative of AI innovation competition

PAC B1 adopts a dual redundant hardware architecture, which not only ensures the hardware environment of mainstream commercial networks, storage and encryption devices to be more secure and reliable, but also provides a more powerful AI computing platform for massive and fast-growing data and complex deep neural networks, and gains the initiative in the fierce competition for AI innovation.

According to FFT Benchmark (Fast Fourier Transform Algorithm), compared with the previous generation products, PAC B1 can support four times the super large-scale deep neural network model and the training speed is 10 times higher. The single and double precision operation performance of the whole GPU is also improved by 2.5 times. It has great application value in high-performance computing scenarios, such as AI scenarios, life sciences, molecular simulation, physical modeling, environmental sciences, resource exploration, medical imaging, internet of things, quantitative finance, 3D design, engineering simulation and etc.

PAC B1 also implements a more flexible pricing model, in which both monthly payment and yearly payment are available, including RMB 49,183 per month and RMB 590,196 per year. With high cost performance ratio, it enables the majority of users to enjoy a better AI innovation acceleration experience at a lower price.

As a financial AI solution jointly created by PAT and the Inspur Group, PAC B1 has greatly improved its computing performance, chip interconnection, data throughput and other aspects. It will strongly promote the pace of bank cloud, and realize the cross-border innovation of financial service scenarios and the continuous upgrading of financial service experience. With the deepening research on computing empowerment and scene fusion, the Fuxi Laboratory supported by PAT and the Inspur will create more cloud innovative products, thus enabling more ecological scenes.

PAC B1 Function Configuration

CPU: 2*24 Core
Memory: 1536 GB
Storage: 8 *1966G Local SSD Storage
GPU: 16 * NVIDIA Tesla® V100 (32GB) GPU
Network: Support Virtual Private Cloud (VPC) network, maintain interconnection with common ECS instances and other cloud products. At the same time, it also provides performance and stability comparable to physical machine network.
Operating system: Provide common operating system, which can customize the work order individually.
Security: Maintain the same security policy and flexibility as existing ECS virtual machines.

About Ping An Technology

Ping An Technology is the core technology arm of Ping An Group, using AI, cloud and other cutting-edge technologies to develop and operate mission-critical platforms and services that support financial services, medical health, automotive services, real estate services, and smart cities. With "Cloud Infinity" as its value proposition, Ping An Technology is actively practicing the corporate philosophy of "Professionalism makes life simpler", and is determined to become a global leading "AI+Cloud" company.

Ping An Technology, as the high-tech kernel of the Ping An Group, has applied its technology solutions to more than 400 scenarios, incubating new technology-driven businesses and helping build ecological closed-loop. Ping An Cloud has formed full-stack service capabilities from IDC infrastructure to SaaS. Through Ping An Cloud, the ecological platform has been built to serve 500 million users and enabled a better future with technology globally.

Cision View original content:

Source: Ping An Technology