Huawei’s HPC platform supports GPU accelerators from the Nvidia Kepler to Pascal series, meeting enterprises’ diverse requirements on heterogeneous computing. In addition, Huawei’s 45ºC warm water cooling solution effectively resolves the heat dissipation and energy consumption problems caused by high-density computing. It can decrease the power usage effectiveness (PUE) of data center cooling to 1.1 or less, and reduce the total cost of ownership (TCO) by up to 40% for customers.
As technologies such as deep learning and virtual reality develop and widely spread, the GPU technology is accelerating computing intelligence. General-purpose GPU (GPGPU) computing with higher density demands more efficient HPC solutions. Huawei uses its FusionServer X6800 and E9000 servers to build the GPGPU computing platform and deploys a flexible manifold liquid cooling design. The servers can be installed in both standard and third-party customized cabinets. Each cabinet supports up to four Huawei FusionServer E9000 liquid-cooling servers with 128 compute nodes in total.
Lin Jun, Chief Architect of Huawei Server Product Line, was invited to the partner forum at GTC China 2016. He delivered the keynote speech, ‘Innovation Unleashes HPC Potential,’ which introduced Huawei’s core value of building superior HPC platforms for customers through chip and architecture innovation.
Santa Clara-based Nvidia, a leading vendor in the global graphics technology and digital media processor industry, hosted GTC China 2016 for Chinese GPU developers from enterprises, education and research organizations to share GPU technology information and demonstrate innovative GPU applications.
Nvidia is an important partner of Huawei in the HPC field. The two parties have been communicating and cooperating closely in HPC solution development and joint application commissioning.