The 38^th edition of the Top 500 Supercomputer List provided some revealing insights into the networking layer of the most powerful supercomputers running today, an area where InfiniBand solutions, a notable portion of which is provided by QLogic, are moving forward. QLogic and a fresh Wikibon study  have pulled some interesting data, in addition to providing an observation of some emerging trends in this area.

Thirty-eight percent of the top 500 with a MPP architecture are powered by InfiniBand, while 98 percent of the clustered supercomputers that made it to the list ran either on InfiniBand or Gigabit Ethernet. This, along with even stronger statistics among the top 100 supercomputers, refute the notion that Gigabit is becoming a dominant interconnect fabric, Wikibon’s David Floyer explained.

“The fastest systems requiring the lowest latency and high bandwidth, InfiniBand and Customer/Proprietary interconnects dominate. Indeed, the only trend shown in Figures 1 and 2 is a slight increase towards Custom/Proprietary interconnects, as new ideas are tried out, particularly in MPP systems. Innovation is still alive and well.

In high performance systems and arrays, InfiniBand has become more prevalent.”

The InfiniBand solutions used in these deployments originated either from Mellanox or QLogic, for the most part. Three of the speediest 100 supercomputers are running on QLogic’s TrueScale solution as of November, and a total of 12 percent of the top 500 have incorporated QLogic networking, according to the company.

Supercomputers are another frontier where performance is ramping up the competition, and on the international level, only continues to increase. The area is also expanding on another front, thanks to innovations driven by providers such as Amazon Web Services, which made a new supercomputer-scale instance available not too long ago.