high-performance computing
Phi. (Although the speed of the machine is expected to nearly double, target peak power consumption will rise only from 17.8 MW to around 18 MW.) Te hardware change that has been
forced upon them means that the machine’s designers also need to create new soſtware libraries: a GPDSP operating system; a compiler; a driver; a maths library and a communications library. All sit under an
EUROPE HAS LOTS OF
APPLICATION SOFTWARE EXPERTS AND WE WOULD LIKE TO CO-OPERATE
OpenMP4.0 complier system. At this point in time, it is not clear if the
Professor Lu, who is also deputy director of the software institute at China’s National University of Defence Technology, delivered a keynote lecture at the ISC High Performance Conference in Frankfurt in July
have just 18,000 as opposed to the current 16,000 nodes.) However, the embargo has interrupted the supply of the Xeon Phi co- processors and thus put back by about a year the point at which the new performance will be attained.
Developing the China Accelerator Professor Lu said that they had already upgraded the TH Express-2 interconnect network, concurrent with upgrading the nodes, so it would simply have been a matter
of changing the cards when the Phi co- processors were delivered. With the embargo in place, however, ‘the situation has changed’. Te technologists within the National
University of Defence Technology have therefore turned to a DSP chip that they developed themselves, and are adapting that to integrate it within the Tianhe as the Matrix 2000 GPDSP ‘China Accelerator’. Te chip is 64 bit, 2.4 Teraflops (in double precision) running at 1 GHz with a power consumption of around 200W – roughly comparable to, or even a bit less than, the Intel Xeon
Boeing beware? China’s aircraft prepare for take-off on the world’s biggest computer
Professor Yutong Lu, director of the system software laboratory at China’s National University of Defence Technology, was kept busy at the ISC High Performance Conference in Frankfurt earlier this month.
On the opening day, she collected the award
certifying that, as had been widely expected, Tianhe-2, the supercomputer designed by the university and of which she is deputy designer, had retained its position as the world’s No. 1 system for the fifth consecutive time. The following day, she delivered a keynote address on the Applications Leveraging Supercomputing Systems. On the final day of the conference, she was collecting a certificate again, this time testifying that Tianhe-2 had scored first place in an alternative metric for
www.scientific-computing.com l
measuring the speed of supercomputers: the High Performance Conjugate Gradients (HPCG) Benchmark. In her conference keynote, Professor Lu gave details of some of the large engineering and science problems that the Tianhe-2 was being used to solve. It was clear from the topics that the development of national supercomputers is seen as an essential part of China’s efforts to develop its own aircraft industry.
Computing for China’s aerospace industry Professor Lu illustrated the capabilities of Tianhe-2 with several examples of computational fluid dynamics. One example was a hybrid RANS/LES simulation of scramjet combustion which made use of the neo-heterogeneous architecture to deploy
@scwmagazine
‘China Accelerator’ will be developed further into a commercially available co-processor. All Professor Lu would say was that such development would take a long time and that, for now, ‘it is a solution’ to the problem with which the Chinese have been presented by the US embargo. Te signs are that the embargo has given
fresh impetus to Chinese efforts to develop their own processors. However, the pathway is not yet clear. Professor Lu said: ‘Te next step is hard to say. At NUDT we have the technologists that can do the research in processors. Aſter the research is done, we will choose the technologies that we think are stable enough for supercomputers and other servers. For now it’s just an accelerator, but the next generation system will change the CPU to a domestic one.’ l
a collaborating CPU and MIC (i.e. the Intel Phi co-processors) with offload model. She said that the large compute job was scalable, and simulated supersonic combustion using 998,400 cores on 26,880 million cells, with a parallel efficiency of 79 per cent. A great deal of work had been put into optimising the way the software ran, with grid-block based load balancing, vectorisation, and optimisation of data transfer and of memory. Other CFD examples run on Tianhe-2 involved simulation work for transport aircraft and large passenger aircraft. This time, the focus was not on combustion in the engines, but the airframe and aerodynamics of the exterior surface of the aircraft. However, aerospace engineering was not the only subject she discussed. She also discussed the use of the machine for: gene sequencing for clinical applications; drug design; climate modelling; and cosmology. ‘Our focus is on basic science and technology,’ she reiterated.
AUGUST/SEPTEMBER 2015 17
ISC Events
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