Exceeding Expectations With the Intel Omni Path Architecture

18 months ago at the International Supercomputing Conference (ISC14) in Leipzig, Germany,  we started talking about Intel Omni Path Architecture (Intel OPA), the next generation fabric technology from Intel for High Performance Computing (HPC) and low-latency scale out systems. Intel OPA is a foundational element of the Intel Scalable Systems Framework (Intel SSF), a new holistic, system-level design approach for scalable, flexible HPC systems.

We had ambitious goals for Intel OPA, but as of February, 2016, we’ve exceeded those goals and are experiencing tremendous market momentum for this next-generation fabric technology. From ISC14 until today, we've seen a market response greater than we anticipated. Intel OPA has far exceeded our initial expectations, on many fronts.

Fast forward to SC15 late last year, I said we’d be shipping Intel OPA by the end of the year, and that is what we did.

Last year, Penguin Computing successfully benchmarked and demonstrated Intel OPA. Penguin also won the Commodity Technology Systems (CTS-1) procurement for the National Nuclear Security Administration’s three national labs—Lawrence Livermore, Sandia, and Los Alamos national laboratories.

In addition to Intel OPA, CTS-1 integrates the latest generation of Intel® Xeon® processors. The Tri-Labs CTS procurement runs for three years.

Intel OPA also went into Hewlett Packard Enterprise’s integration of the Bridges system for Pittsburgh Supercomputing Center (PSC), which was delivered early this year. The Bridges system is unique, in that it is designed for converged workloads, combining traditional HPC simulation projects with visualization, machine learning, and high performance data analytics. The industry has been talking about the converged data center for some time; Bridges is the first installation of a system that covers the gamut of these computing workloads, and we’re excited that it uses Intel OPA to support them.

We’re also very pleased that the first part of Bridges with its Intel OPA fabric was up and running for the PSC Hackathon that took place February 25 and 26 at PSC. Our congratulations go to the winners—Team Spew, Team Blitz and Chitz, and Team Rocket—who were successful in running their codes on the Bridges cluster.

The latest announcement on Intel OPA installations is University of Colorado gearing up for a cluster with both Intel OPA and Intel® Xeon Phi™ processors, also known as Knights Landing.

To drive future efforts for Intel OPA, in 2015 we engaged our worldwide OEM channel of some 75 partners and made tremendous progress, enabling them to launch many successful efforts with OPA in 2016. We worked with some partners to build benchmarking clusters to help them and their customers understand the value of Intel OPA. Through those clusters we helped to establish Intel OPA’s performance and price/performance value.

We’re excited about bringing this new architecture to the HPC marketplace to deliver the performance and price/performance that HPC shops desire. And, we’re incredibly pleased at how the momentum of Intel OPA market acceptance, both with OEMs and end users, has exceeded our initial expectations. We look forward to 2016 and delivering Intel OPA solutions for our partners and their customers throughout the HPC ecosystem.

Published on Categories High Performance ComputingTags ,
Barry Davis

About Barry Davis

Barry Davis has over 28 years of experience in the computing and telecommunications’ industries. While at Intel Corporation, Mr. Davis built multiple businesses from the ground up creating Intel’s I/O Storage and Wireless Networking groups. Mr. Davis was one of the original people behind the worldwide success of Wi-Fi, including the introduction of Intel CentrinoTM Mobile PCs. He has numerous industry awards and holds 11 U.S. and 2 international patents. Mr. Davis has been at the forefront of Intel’s recent fabric plans and was the lead on creating the strategies, plans, and products that have resulted in the Intel® Omni-Path Architecture. Barry is currently General Manager of the Accelerated Workload Group that is part of the Enterprise & Government Group (E&G) in Intel’s Data Center Group. He holds a B.S. Electrical Engineering from Lehigh University.