FRANKFURT, Germany, June 20, 2016 — The 31^st ISC High Performance conference(ISC16) will be held in Frankfurt. Inspur, one of the top leading manufacturers of IT infrastructure products for data centers worldwide, will be presenting the latest I9000 Blade System with Converged Architecture, and the Global User Plan – centering on the latest Intel KNL platform.

The 13^th HPC Connection Workshop jointly organized by Asia Supercomputer Community and Inspur Group will be held during ISC16.

Inspur-Intel KNL Joint Program

One focus of this conference will be the global launching of the second generation of the Xeon Phi chip “Knights Landing” (KNL) from Intel. The product is expected to revolutionarily improve high performance computing.

The official launching of KNL will inevitably grasp the attention of many manufactures and users. With regard to the latest KNL platform, Inspur and Intel will also jointly announce a global program with details on the way forward – expected to officially be disclosed to the public during ISC16.

Meanwhile, Inspur will also release the ppen source Deep Learning framework Caffe-MPI, developed for the KNL platform.

Converged Architecture Server

At ISC, Inspur will display the latest Converged Architecture Blade Server System I9000. It can support high-density blades, 4-socket and 8-socket mission-critical blades, software-defined storage, heterogeneous computing and various computing solutions in a single chassis.

With up to 16 2-socket blades in a 12U chassis, it can provide higher computing capability in the same space. The flexible scale to 4-socket and 8-socket system makes it ideal for commercial HPC applications, which have higher requirements for the computing performance of single nodes, reliability and scalability.

Additional to the I9000 Converged Architecture Blade Server System, Inspur will also be spotlighting rack scale systems: the SmartRack V4.0 for public cloud applications, and the InCloudRack Cloud Integrated System, for private cloud applications. Designed for large-scale data centers, SmartRack takes the rack as a unit and provides centralized power supply, cooling and management. Compared with traditional servers, the space utilization of the rack system increases from 81% to 90%, the deployment density can be improved by 2 times, and the deployment speed improved by more than 10 times.

The InCloudRack is designed for the enterprise private cloud, has outstanding enterprise-level RAS features, blade-level stability and hot-swappable design for power supply and management modules. It can support a variety of nodes: 2-socket, 4-socket, 8-socket compute blades, MIC/GPU co-processing acceleration nodes, storage expansion nodes, and others – supporting scale-out and scale-up scalability. With UNE (Unified Network Engine), the InCloudRack builds the next-generation 100G distributed network. With software-defined technologies, the InCloudRack can achieve centralizing and pooling of computing, storage and networking resources – ready for integrated dispatching and management.

Promoting the development and upgrade of Converged Architecture and related products is the core of Inspur’s “Computing+ strategy”, which is Inspur’s forward-looking blueprint for the future of the smart computing era. It focuses on integrating computing, storage, networking resources into the equipment with brand new computing architecture, display computing, networking and security features with software-defined technologies. In this way, it can support the demands of varying applications, including HPC, storage, big data, and human-computer interaction, amongst other areas.

Inspur will also display the ClusterEngine – the High-Performance Computing Cluster Management Platform, along with TEYE – the monitoring software of application characteristics of high-performance system that supports multi-threading application and elaborately analyzes system resources occupation of the application program. For the integrated solution to HPC, Inspur has the capacity to develop multiple heterogeneous platforms of MIC, GPU and FPGA, and have several hardware products supporting FPGA/MIC/GPU with professional optimization capacity in life sciences, earth sciences, deep learning and HPC application fields. All of the above have the capability to provide different types of acceleration schemes for HPC application.