Intel® Cluster Ready Partner Newsletter Q2 2013 - Feature Stories

Attacking the Biggest HPC Challenges Through Cost-Effective Supercomputing

The potential for growth in HPC lies along many and diverse pathways. At one extreme, growth depends on delivering ever-increasing supercomputing capability. At the other, it depends on providing pre-packaged hardware and software solutions that are simple and affordable to deploy, manage, and use.

In some scenarios, all these pathways come into play. A prime example is personalized medicine. The ability to evaluate individual health and deliver targeted treatments based on a deep, quantitative understanding of each individual—from DNA and cellular chemistry to tissue interactions and disease mechanisms—presents computational challenges that go far beyond today’s capabilities. Yet, ultimately, personalized medicine must be delivered with the speed, simplicity, and affordability needed to support high-volume clinical settings.

The potential benefits of personalized medicine are vast, both economically and from the more fundamental perspective of reducing suffering and saving lives. Working to deliver personalized medicine may also provide many ancillary benefits. By pushing existing knowledge and technologies to the limit and beyond, this kind of work can provide a foundation for understanding and addressing the needs of next-generation HPC across many disciplines.

Intel and Oregon Health & Science University (OHSU) are joining forces to take on this challenge. They are combining their expertise in medical technology and extreme computing through a multi-year collaboration that may extend into decades. The team will focus initially on cancer. One of their first tasks will be to generate more data at higher speed, by accelerating the speed of genomic profiling. Intel will jumpstart the research process by donating a supercomputing cluster. Although the specific design will be decided by collaborative research and development with OHSU, one thing is certain. The supercomputer will be based on the Intel®Cluster Ready architecture.

Why Intel Cluster Ready? Because long-term success depends on being able to dramatically increase overall compute capability while driving down cost and complexity. Intel Cluster Ready is well aligned with these goals. By automating many common tasks, Intel® Cluster Checker software will help to reduce time and effort when building, testing, troubleshooting, and managing the cluster. It will also help to improve cluster quality and software interoperability by ensuring hardware and software consistency, completeness, and performance, not only during initial deployment, but as the cluster environment continues to expand over time.

The cluster will almost certainly incorporate both multi-core Intel® Xeon® processors and many-core Intel® Xeon Phi™ coprocessors so that hardware resources can be closely matched to the workload characteristics of the many different software functions. The industry-leading performance and energy-efficiency of the Intel Xeon processor E5 family provides the foundation for maximizing the performance and energy-efficiency of most of today’s existing applications. Intel Xeon Phi coprocessors offer the massive parallelism needed to accelerate performance and improve energy-efficiency for highly-parallel applications and code segments.

With up to a teraflop of double-precision performance per coprocessor and the ability to support up to eight coprocessors per two-socket server node, the design team will have flexibility for delivering levels of parallelism that closely match workload requirements. The ability to use the same code, development tools, and optimization strategies for both processors and coprocessors will help to simplify development and enable software engineers to be more focused and productive.

More than 12 million people are diagnosed with cancer annually and it is one of the most complex and idiosyncratic of diseases. The team does not expect to eradicate this killer disease quickly, if at all. However, they do expect to fundamentally improve our understanding of cancer disease mechanisms, and to help develop and deliver more effective treatments faster and more economically. Hopefully, their efforts will lay the foundation for delivering comprehensive and fully-personalized diagnostics and treatment plans for all ailments in the future.

This collaboration may also provide information and experience that will help Intel engineers define the path forward for delivering HPC products, technologies and solutions that are far more powerful, energy-efficient, and easy to use. Importantly, the supercomputing environment will be based on a standards-based computing architecture and a standards-based cluster architecture. As a result, Intel and the Intel Cluster Ready community will be well positioned to drive these advances efficiently and affordably into the broader HPC marketplace.

For more information about the Intel and OHSU collaboration, see:


See You at ISC’13

Come meet us in Booth # 541! We continue to expand the Intel® Cluster Ready program and promote the values of standardized cluster architectures for you to deliver cluster solutions to your customers. Technical computing clusters must be balanced, efficient, and reliable, and you may want to optimize your applications for massively parallel execution on Intel® Xeon® processors and Intel® Xeon Phi™ coprocessors. The Intel Cluster Ready architecture is able to meet these new challenges. See what is new with Intel® Cluster Checker 2, Intel Xeon Phi support and other developments at ISC!

An Intel Cluster Ready Partner Self-Guided Tour promoting partner High Performance Computing booths and solutions with Intel-provided attractions to gain customer interest is featured this year.

In addition to supporting partner booth and floor activities, we will also host our ISC’13 Partner Appreciation celebration on Tuesday, June 18, from 6:00-8:00pm at The Hacienda. Dr. Stephen Wheat, Intel’s General Manager of High Performance Computing, will welcome and provide opening comments. Click here to R.S.V.P.

Look forward to seeing you there. If you are not able to attend, look for updates on our website.  

Learn more about ISC’13.

 

For more complete information about compiler optimizations, see our Optimization Notice.