Welcome!

Virtualization Authors: Deborah Strickland, JP Morgenthal, John Cowan, Kevin Benedict, Pat Romanski

Related Topics: Cloud Expo, Virtualization

Cloud Expo: Article

Got HPC Network Congestion? Take Voltaire

Voltaire says it can reduce collective operations' runtime by up to 100x

Voltaire says its 40 Gbs InfiniBand switches can speed up High Performance Computing (HPC) applications, stuff whose business value is in their speed such as reservoir modeling, fluid dynamics and crash analysis, by as much as 10x.

All it takes is a painless software upgrade to the switch; no changes to the applications software. If the applications are MPI-based, the scheme works out-of-the-box.

See, Voltaire has figured out a way to kill off a lot of the needless chatter that goes on in scale-out data center fabrics - there's a lot of redundant messages that it can short circuit and accelerate group communications or MPI collective operations by removing bottlenecks at both the node and interconnect levels and generating only a single message over each physical wire.

Voltaire says it can reduce collective operations' runtime by up to 100x.

Less noise translates into scalability too.

In the next iteration of its newfangled Fabric Collective Accelerator (FCA) software, Voltaire will move on to MapReduce and Hadoop widgetry.

Marketing VP Asaf Somekh, who fancies the solution is "groundbreaking," said Voltaire figured the widgetry would work on big HPC installations and was surprised to see it accelerate even tiny four-server clusters.

Previous solutions have been based on host-based offload, which Voltaire says addresses only a small part of the challenge at the server level. They don't deal with network congestion or computation and message performance and they're difficult to manage and orchestrate.

Voltaire expects to deliver FCA in early Q2. It should list for $240 a server.

Voltaire Wednesday also added a mid-size 40 Gbs InfiniBand director switch to its 40 Gbs portfolio.

The new Voltaire Grid Director 4200 is a high-performance ultra-low latency InfiniBand switch targeted at high-performance clusters running scientific, commercial HPC and enterprise applications. With configurations of up to 162 ports of 40 Gbs per port of InfiniBand connectivity, the device delivers 11.52 Tbs of bandwidth and between 100 and 300 nanoseconds of port-to-port latency.

It'll be available at the end of March.

More Stories By Maureen O'Gara

Maureen O'Gara the most read technology reporter for the past 20 years, is the Cloud Computing and Virtualization News Desk editor of SYS-CON Media. She is the publisher of famous "Billygrams" and the editor-in-chief of "Client/Server News" for more than a decade. One of the most respected technology reporters in the business, Maureen can be reached by email at maureen(at)sys-con.com or paperboy(at)g2news.com, and by phone at 516 759-7025. Twitter: @MaureenOGara

Comments (0)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.