Subscribe / Unsubscribe Enewsletters | Login | Register

Pencil Banner

Treat supercomputing like a national resource says ARM Research HPC lead

Tamlin Magee | Sept. 21, 2016
Supercomputing capability tends to be regarded as a kind of arms race, and Van Hensbergen believes that there is merit to this claim.

Distinguished engineer at ARM Research and high performance computing specialist Eric Van Hensbergen has warned that Britain, Europe, and even the USA are not taking supercomputing seriously enough - and that it should be considered a national resource.

ARM recently partnered with Fujitsu and Japanese research organisation RIKEN to power the 64-bit Post-K supercomputer. Van Hensbergen is working on the project that although delayed, will mean ARM architecture is behind exascale computing in the near future.

Supercomputing capability tends to be regarded as a kind of arms race, and Van Hensbergen believes that there is merit to this claim.

"There's a huge advantage to having a well-trained domestic workforce," he told Computerworld UK at the ARM Research summit in Cambridge. "To be honest, and Europe is a victim of this, a lot of computing has ended up dominated by US players.

"Europeans are essentially run out of the business from a manufacturing perspective - I think that comes at a disadvantage, because if you want to get to a state where you're building for economic or security reasons, you're starting from zero."

Earlier this year ARM was acquired by Softbank for £24.3 billion, to bolster the Japanese conglomerate's share of the internet of things and data centre markets. Softbank has publicly committed to, at the very least, leaving the funding for existing projects unchanged, and across the business there is a chance of reinvestment where profits previously had to be reported to shareholders.

The fact that Softbank acquired ARM in the first place, along with the Post-K supercomputer development currently underway in Japan, suggests there may be differing attitudes to HPC research compared to other regions.

Van Hensbergen says Japan's approach to supercomputing is one that's worth emulating.

"In Japan, with the previous generation K Computer, at one point they were going to cancel the project," he explained. "They rented out an hour of primetime TV and had 10 of their Nobel laureates get on and explain why it's important for Japan, and then they held a national referendum.

"Everybody knew what it was and everybody understood how important it was. You just don't see that kind of exposure and understanding elsewhere."

Van Hensbergen suggests that in the west, there's a tendency to dismiss supercomputing as wasteful spending that the government makes to prop up academics.

"That's really not the case," he said. "It's important that it be directed and focused, because there are real societal impacts. These national infrastructures are not stunts to say 'the US is better than China, or Europe, or Japan' or anything like that - it really is about the future of science, innovation and economies in these countries, as well as supporting a high-tech workforce, which is incredibly important.

 

1  2  Next Page 

Sign up for CIO Asia eNewsletters.