Energy efficiency in HPC ➤
high-performance computing
A bridge to the US market ‘We have plans to set up a US office early next year,’ Wang continued. At the present moment, Sugon is not really focused on selling HPC systems into the United States. Although he did not articulate the reasoning behind this, it is clear that the US market is already dominated by many domestic players and also clear, at least to anyone outside the USA, that the US Government is not above using defence and security export controls to influence trade policy in high-technology in favour of its own industries. But in the long run, ‘to expand globally
we can’t ignore the US market,’ he said, ‘so we could acquire or create a merger with US companies as a bridge to open the market in the USA.’ In the interim, he continued: ‘What we want to do here is to know the trends of technology development, connect with service providers, and to connect with the upper levels of the industry.’ A US presence would open the way, he suggested, to closer relationships with companies such as Intel, VMware, and Seagate. Such partnerships are already happening. In
October this year, Sugon opened a joint venture with VMware in China to use VMWare’s technology, albeit badged as Sugon, in products aimed at the Chinese market. It is not a small undertaking – the joint venture has more than 200 employees. Te news of Sugon’s cooperation with
VMware prompted a US$2 jump in the company’s share price. Wang remarked that when the company was first listed on the Shanghai stock exchange in 2014, its shares sold at US$1 each whereas now the price is around US$20, ‘so these statistics tell you all you need to know about our stockholders’ confidence in the company’. As far as Sugon is concerned, in Wang’s view, the recent turmoil on the Chinese stock exchanges represents only an ‘adjustment – a return to realistic valuations’.
Inspur and Intel cooperate on technology Inspur has already created close links with US companies, such as Intel. Te two companies took the opportunity of SC15 to launch, jointly, a new converged architecture blade server system, which they claimed was the only blade server system in the industry to flexibly support 2-, 4- and 8-way nodes. It is envisaged as a solution for enterprise-level cloud data centres and, at the same time, offering high density and low power consumption. All the Chinese supercomputer vendors are
broadly based technology companies, in the sense that they are interested in the business and enterprise markets as well as the high-end technologies of supercomputing. In this respect
22 SCIENTIFIC COMPUTING WORLD Sugon’s headquarters building in Beijing
their commercial strategies mirror those of US companies such as IBM, Dell and HP, rather than that of Cray, which focuses solely on the high-end sector. Inspur’s US links are more developed than
just technology cooperation, with companies such as Intel. At SC15, Inspur stressed that, in the United States, its servers have been operating for some time for internet customers. According to John Hu, vice president and chief technology officer, the company has already established local production lines in the USA to serve its customers there, and has offices on the west coast of the USA: a technology research and development centre in Silicon Valley, and a worldwide services office based in Seattle. Hu continued: ‘Te storage and server market
in the USA is the most important. We have to be here.’ Inspur’s global development strategy started last year, in 2014, he continued and was very important to the future of the company. It sees Huawei as one of its competitors. Smart cities are also on the Inspur roadmap.
In September, the company signed a strategic cooperation agreement with the US telecoms technology company Cisco to set up a joint venture in China with an initial investment of US$100 million. Part of its mission will be to provide advanced technology, products, solutions and services for cloud centres, smart cities and big data.
Inspur eases path to FPGAs On the technology side, the tide in high- performance computing generally seems to be running in favour of FPGAs, and Inspur is keen to ensure that it catches that tide. Te drive to incorporate FPGA technology in Inspur’s portfolio means that it is becoming an HPC systems vendor capable of offering heterogeneous computing architectures embracing GPUs (for it already has cooperation agreements with Nvidia), Intel Phi co- processors, and FPGAs.
Hu remarked that an FPGA could deliver
twice the compute performance of a CPU, with an energy consumption of only 30W compared to something like 190W. Te most significant barrier to the wider adoption of FPGAs is the difficulty in programming them, and Inspur is addressing this issue, he continued, by using an OpenCL app to accelerate the process. He cited one project in which it took only two engineers
THE STORAGE AND
SERVER MARKET IN THE USA IS THE MOST IMPORTANT
two months to port a neural network algorithm to FPGAs. ‘It is very interesting for us to lower the barriers to FPGAs,’ he said. One application where the company is
interested in using FPGAs is in ‘deep learning,’ he continued. At SC15, Inspur launched one example of such a project – a speech recognition system based on Altera’s Arria 10 FPGAs and a deep neural network (DNN) algorithm from iFLYTEK, an intelligent speech technology provider in China. It is a ‘deep learning’ application using Inspur’s FPGA- based heterogeneous design of architecture and soſtware developed in a high-level programming model in OpenCL to enable migration from CPU to FPGAs. Looking further to the future, Hu explained
that Inspur will develop an FPGA-based system, covering full rack computing, internet servers, and storage, with the aim of making these solutions widely available. He envisaged the possibility that the new heterogeneous computing model for HPC could be a CPU+FPGA system, with more and more HPC applications, data centre applications, internet, and deep learning applications being ported to a CPU+FPGA architecture. l
@scwmagazine l
www.scientific-computing.com
Sugon
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40