NVIDIA GRID Boosts Building SSC Remote Visualization Platform
Shanghai Supercomputer Center (hereinafter referred to as SSC), established in December 2000, is one of the main project of Shanghai Harbor - one of Shanghai 2000 No.1 Project. Invested by the Shanghai Municipal Government, it is located in Zhangjiang Hi-Tech Development Zone, covering an area of about 10,000 square meters. SSC is the first high-performance computing platform for public service that is open to the society with shared resources, first-class facilities and complete functions. It is an important information technology infrastructure of Shanghai. Currently boasting “Cube” (ranking 10 in the world and the first in Asia in 2008) “hummingbird” and other supercomputers, and equipped with a wealth of scientific and engineering computing software, it is committed to providing high-end computing services for the national scientific and technological progress and innovation.
Over the years, based in Shanghai and facing China, SSC has been providing on-demand high performance computing resources, technical support and advanced technical consulting services for more than 400 users from engineering research institutes, and many well-known universities. It has supported a large number of major scientific research, engineering and new product development of national and local governments, and has made a large number of significant results in a number of fields like automotive, aerospace, steel, nuclear energy, municipal engineering, new materials, bio-pharmaceuticals, astronomy, physics, and chemistry.
At present, the Shanghai Supercomputer Center is launching remote visualization platform. Users in the use of engineering calculation software will be involved in processing before and after or intermediate results view, with all these processes requiring graphics support. Yet existing graphics workstations cannot meet the multi-user’s concurrent demand or demand for the smooth operation and display of remote graphics.
Deploy GRID, Drive graphics Acceleration to a Virtualized Environment
By using NVIDIA GRID technology built on NVIDIA Tesla® M60 platform, Shanghai Supercomputer Center brings graphics acceleration to the enterprise virtualization. At the same time, GRID can provide comprehensive virtual application compatibility, which means that any application, which can run in physical desktop, can run in a virtual desktop, greatly enhancing user experience.
In addition, with GRID, IT managers can also easily allocate video memory number, provide custom graphic profiles to meet the specific needs of each user. Each virtual desktop has a dedicated video memory, just like on the desk. When you run the application, these virtual desktops all have their own resources. GRID allows up to sixteen users to share a single physical GPU, which can evenly allocated the available GPU graphics resources to virtual machines.
At the same time, GRID supports the leading Citrix, Microsoft, NICE and VMware enterprise virtualization solutions, which also facilitates and pushes SSC’s application deployment in the data center. The solution also allows GPU allocation and sharing among multiple users, allowing managers to focus on the application requirements and overall throughput of the data center.
Shanghai Supercomputer expresses satisfaction with the NVIDIA GRID technology and Tesla products, and hopes to do further research based on remote visualization environment under Linux virtual machine, in order to get familiar with building a Visualization Platform combining NVIDIA GRID new technologies and Maxwell new products.
There is no doubt accelerated computing is changing data center. Under the same power budget and footprint, accelerated data center will provide unprecedented throughput, which can bring new discoveries and services to end users.
First class HPC applications are GPU-accelerated through the deployment of GPU accelerators in the data center, in order to bring faster discoveries and scientific insight to users. Through broad support for HPC developers tools (such as MPI, scientific databases and OpenACC), most applications can now improve performance by means of GPU.
Applications have been quickly deployed in different infrastructures supported by Tesla platform in any infrastructure, including applications running on bare computer, operating as isolated container, or running in the virtual platform (on a user operating system). In the data center, these deployment options allow administrators to be very flexible to meet user needs and application requirements.
Business services provide comprehensive support solutions for the NVIDIA software on Tesla acceleration system. With the release of quarterly maintenance versions, process of priority critical problem solving, and communication with NVIDIA technical experts and on-site consulting and training services, the uptime and efficiency of GPU acceleration data center have been improved to some extent.