Welcome to english.eastday.com.Today is
Follow us @
Contribute to us!










Home >> auto >> Article
Towards Exascale computing, Chinese company achieves 1st OPA based Torus switch
From:Xinhua  |  2017-11-20 02:25

Video PlayerClose

LOS ANGELES, Nov. 19 (Xinhua) -- As the flops of a supercomputer arrived Petascale and will reach Exascale, the efficient communication among thousands of nodes becomes a million dollar question.

One pioneer solution has been given as the Silicon Switch, OPA based Torus topology switch, which was released by Chinese high-performance computing company Sugon at the Super Computer Conference (SC17) this week in Denver in the U.S. state of Colorado.

"After much discussion, we formed a joint team. And that team collaborated internationally for over a year to produce this product," Joe Yaworski, Intel OPA marketing director told Xinhua.

"This is just a great example how the two countries, the two companies and the two organizations can work together. This is really a good example of international cooperation to develop products," he said.

Large-scale supercomputers, especially those quasi-Exascale or Exascale systems, are facing severe challenges in terms of system scale, scalability, cost, energy consumption, reliability, etc.

The new released Silicon Switch, supports the cold-plate direct liquid cooling as well, adopts the Torus architecture and the state-of-art OPA technology, and then carries more competitive features, including advanced performance, almost infinite scalability, and excellent fault tolerance ability, according to Sugon.

"It shall be a wise choice for Exascale supercomputer," Dr. Li Bin, General Manager of Business Department for HPC Product of Sugon, said in a news release.

Compared with the traditional Fat-tree network topology, the Torus direct network, which emphasizes the neighboring interconnection, has obvious advantages in scalability and cost/performance, since it only holds a linear dependency between the network cost and the system scale.

In addition, the rich redundant data paths and the dynamic routing give inherent superiority in fault tolerance ability.

All these features well meet the requirements of Exascale supercomputer and pave a new trend of high-speed network technology, researchers say.