This page contains a Flash digital edition of a book.
HPC predictions

Andrew Jones, vice-president of High-Performance Computing (HPC) Services and Consulting at The Numerical Algorithms Group (NAG)


or a number of years, the computing industry has been relatively static in terms of the processor technology

that has been deployed. However, due to underlying technology issues, times are changing and there’s a lot more diversity and willingness – among both users and vendors – to take risks on different technologies in search of greater performance. With that in mind, I anticipate a big battle between Intel and Nvidia on the accelerator front later this year when Intel’s Knights Corner becomes available. And if organisations are willing to head in a non-traditional direction one way (many-core processors), then why not go another and look at a greater use of lower- power ARM processors and other mobile processors? I definitely think a big story for this year is the potential for this technological diversity to make a difference. Te near-ubiquity of multi-core processors, coupled with no shortage of success stories,

is leading users to evaluate many-core and GPU systems to see whether they can get any advantage from that technology. We have an insatiable appetite for more performance, but individual processor cores are getting slower, not faster, and purchasing a GPU can be seen as the next best – and cheaper – option compared to clusters. Issues surrounding cost and particularly power consumption are forcing the industry down the many core or parallel processing route, and that in turn is creating a soſtware crisis. Te previous generation of soſtware just

won’t run efficiently on this new class of technologies and for several years now there have been a number of people in the industry shouting that something needs to be done, while the rest put their heads in the sand waiting for a magic bullet to come along and fix the problem. Tis isn’t going to happen and I think people are beginning to realise that and explore ways of evolving their soſtware.

Another issue

coming to a head is the shortage of people who are able to do the job. Tere needs to be this curious breed of person who is part computer technologist, part scientist, and there’s an international shortage of them. Whichever country solves this problem first will have a huge advantage because anyone can build a supercomputer, but training a generation of individuals to program parallel computing technology is a much longer process and a far more sustained investment. From a personal perspective, I think there’s too much focus on the question of how we program the biggest machines in the world, as opposed to how we get the breadth of HPC users able to program efficiently and correctly in parallel. Tis will need to be addressed.


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  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48