One thing was clear. HPC is no longer just about the big boys. Vendors are waking up to the vast potential digital simulation holds for smaller businesses. There’s a broad emphasis on delivering simple, production-ready cluster solutions that run popular applications out of the box. Given that emphasis, it’s not surprising that Intel Cluster Ready was visible almost everywhere.

We had two Intel® Cluster Ready partners, Altair and Fujitsu, presenting their solutions in the booth. Ten Intel Cluster Ready partners gave theater presentations, 25 were promoting Intel Cluster Ready with signage and collateral, and more than 150 participated in the Intel Cluster Ready Partner Celebration. Their enthusiasm says a lot. Their videos and presentations say even more. I’ve posted a few highlights below. Enjoy!

• transtec explains how Intel Cluster Ready can help you achieve higher productivity with lower cost and risk.
• E4 Computer Engineering discussed how Intel Cluster Ready helps to shrink deployment timeframes to just days and simplifies management and use.
• Fujitsu talks about its new line of ready-to-go HPC solutions and how Intel Cluster Ready helps them deliver higher value across the full range of cluster requirements.
• Intel Cluster Ready team members and partners discuss their solutions, Intel Cluster Ready, and the market inroads their making at the Intel Cluster Ready Partner Celebration.

Want a broader overview of ISC ‘12? Check out these cool simulation examples from Altair and these 6-minute “hot seat” vendor videos—including this one from Dr. Stephen Wheat, General Manager of High Performance Computing at Intel. He talks about Intel innovation in HPC, including an upcoming Intel processor that can deliver up to a teraflop of performance—all by itself.

That’s something new in the HPC world. It took a lot of behind the scenes effort for Intel and independent hardware and software vendors to synchronize their development cycles and launch dates. Dozens of systems and applications were available on day one. Customers had other options, too. They could upload their application to a test bed cluster from a leading cluster vendor. Or they could spin up a Intel Xeon processor E5 family cluster in a public cloud in a matter of minutes.

The upshot of all this is that new clusters based on the Intel Xeon processor E5-2600 are now available from a number of leading hardware vendors, along with cluster management software and a variety of optimized applications. Many of these products are Intel® Cluster Ready certified (or registered, in the case of applications). That means you can mix and match them with confidence. Any registered application will run reliably on any certified cluster right out of the box, which makes implementation a lot simpler and less risky. Pick an application. Pick an appropriately sized cluster. Load one on the other and you’re off and running—at significantly faster speeds than you would have experienced just a few weeks ago.

If you’re not familiar with the new Intel Xeon processor E5 family, you’ll find that it delivers more resources across all key parameters in comparison with the previous-generation Intel Xeon processor 5600 series. Whether your workloads are CPU-, memory-, or I/O-bound, you should see significant performance benefits. Highlights of the new architecture include:

• 2X more I/O bandwidth, with a 32% reduction in latency
• 2X more socket-to-socket bandwidth
• 2 more cores and 2.5 MB more cache per core
• Up to 2X faster floating point performance
• More and faster memory (up to 768GB/1600Mhz in a dual-socket system)

If you’re an HPC user looking for cluster solutions based on this new processor, any of the following Intel Cluster Ready members would be a good place to start: AMAX, Atipa, ClusterVision, Colfax, E4, Eurotech, Fujitsu, Silicon Mechanics, Viglen, and transtec. Each of these vendors had Intel Cluster Ready certified systems or registered software available at launch. The solutions they offer range from small clusters that could fit on your desk to multi-rack supercomputers that push the limits of HPC capability. (You can find additional lists of certified clusters and registered applications on the Intel Cluster Ready website.

If you’re looking to build your own cluster solutions, you might want to check out the following resources:

• Intel® Cluster Ready Reference Designs
• Intel® Cluster Checker 1.8, which includes full support for the Intel® Xeon® Processor E5 2600 processor family (and the associated Romley platform).
• Intel® Xeon® Processor E5-2600 Family Cluster Solutions Guide, which provides an overview of supported components for the new Intel Xeon processors. This guide provides information and recommendations about Intel server hardware, new processor features, management software, storage, interconnects, and other HPC cluster components. It can help you get an optimized cluster up and running faster. You can download the guide here.

Regardless of whether you’re a user or a vendor, I hope you’ll look into the new resources that are now available. Cluster computing is getting faster, simpler, and a lot more cost-effective, which is opening up new opportunities for everyone. The Intel Xeon processor E5-2600 family is just another step forward on the journey. Well, maybe a leap.

Like many of you, I attended SC11 in Seattle this year. I have to say, the world’s largest HPC convention has become downright addictive—especially if you happen to be an ex-gamer. I had to extract myself from the “Gaming through the Decades” exhibition before I could even get started with more serious pursuits. Fortunately my willpower was well rewarded. In case you weren’t there, the focus was on big data and visualization, and the strides being made are truly amazing.

The advances made by the Intel Cluster Ready partner community are also very impressive. We held our third annual Partner Appreciation Award ceremony at SC11, and it was gratifying to see the progress we’ve all made in just the past year. The Intel Cluster Ready architecture is now represented across an enormous range of HPC solutions, from SMB clusters to some of the world’s Top500 supercomputers. Hardware and software support has also grown substantially, as has market penetration. For a look at a few of the most impressive developments, check out the award-winning partners below, and follow the links for more information. I think you’ll be intrigued.

We gave awards in three categories:

• The Pathfinder Award honors the vendor who has done the most to forge new paths into a previously unexplored area.
• Explorer Awards distinguish those who have continued to explore and implement programs that significantly advance the portfolio of solutions based on the Intel Cluster Ready architecture.
• Voyager Awards are for those who have contributed to expanding the global reach and overall market growth of Intel Cluster Ready solutions.

Here’s the list of the 2011 award winners. Enjoy!

Pathfinder Award

• Altair continues to drive innovation by making HPC cluster computing simpler and more productive for its customers. Over the past year, Altair integrated Intel Cluster Checker into PBS Professional, its HPC workload management and job scheduling solution, and registered three key applications for design and analysis: RADIOSS, AcuSolve, and OptiStruct.

Explorer Awards

• science+computing took Intel Cluster Ready into the upper reaches of HPC, installing a 292 TFLOP, 1700 node supercomputer at RWTH Aachen University. It’s the largest Intel Cluster Ready certified system to date, and one of the 30 largest supercomputers in the world. A joint success story will be coming soon to highlight the design win and promote Intel Cluster Ready.
• NEC delivered a huge Intel Cluster Ready win, with a certified 500-node cluster for the University of Erlangen. NEC then helped drive visibility and credibility for Intel Cluster Ready solutions with a success story and Web promotions.
• E4 Computer Engineering launched its E-Cluster 7225 platform, certified Intel Cluster Ready, which scales from four to hundreds of nodes! E4 has also integrated Intel Cluster Ready throughout its marketing communications and updated its website based on each new sales quote.

• Fujitsu has demonstrated a strong commitment to Intel Cluster Ready, integrating support throughout its PRIMERGY server product lines, and have achieved the greatest number of certifications in 2011 for certified Intel Cluster Ready clusters shipped in Europe. Starting at Fujitsu Technology Solutions it became a well-integrated part of the technical production workflow—beginning to end. Intel Cluster Ready is now an integral part of the Fujitsu standard cluster software stack for Intel architecture, and allows them to address a multi-geo market with HPC systems that are 'ready to go'.

• Bright Computing has played an integral role in enabling the Intel Cluster Ready ecosystem by fully incorporating support into its workflows and products, including integration of Intel Cluster Checker into Bright Cluster Manager 5.2. The company was instrumental in helping ensure that the complete HPC solution chain will be ready to support the Intel Xeon Processor E5 family at launch.

• The HPC Advisory Council made broad contributions to Intel Cluster Ready by testing, benchmarking and documenting implementation best practices for open-source applications, co-authoring and publishing a technical brief, and partnering in workshops to extend the awareness and value of Intel Cluster Ready solutions to new audiences.

Voyager Awards

• Flow Science created a cluster solutions guide for Silicon Mechanics and other OEMs to simplify the deployment of Flow-3D, a leading computational fluid dynamics (CFD) solution, on Intel Cluster Ready certified systems. Through this and related efforts, Flow Science is helping to bring HPC—and Intel Cluster Ready—into the digital manufacturing market.
• Silicon Mechanics has integrated Intel Cluster Ready into its technical workflows and sales processes to simplify and accelerate cluster implementation. The company also played a pivotal role in supporting the launch of Sandy Bridge EP and partnered with SIMULIA and Flow Science to deliver optimized, turnkey solutions based on the Intel Cluster Ready architecture.
• SIASA (Soluciones Informaticas Argentinas S.A.) was the first Intel Cluster Ready member in Latin America and has delivered more certified clusters than any other vendor in the region. As one example, SIASA helped the University of Cordoba implement the largest cluster in Argentina.
• Platform Computing is a leading Intel Cluster Ready partner and made a major contribution this year by launching Platform Cluster Manager 3.0, which makes it easier for non-technical customers to build, deploy, and manage Intel Cluster Ready certified systems. The company also provided valuable development support for upcoming Intel server platforms.
• ClusterVision – their commitment to Intel Cluster Ready is so strong that they have edited all their marketing materials and website to promote Intel Cluster Ready. They also implemented Intel Cluster Ready training into their annual sales meeting and launched a sales incentive program. The strategy has delivered new cluster solutions, including a cluster deployment for Ceres Power that was documented in a success story.
• Atipa has shipped eight "Rapid Launch" Linux Clusters in the last two years, all Intel Cluster Ready certified. Atipa has played a valuable role in enabling the HPC ecosystem for the launch of Sandy Bridge EP.
• Eurotech launched its Intel Cluster Ready certified Aurora product line. Then, after collaborating with AuroraScience on a dense, green cluster that’s certified Intel Cluster Ready, Eurotech helped produce a success story about the unique 10 Teraflop Aurora cluster that serves a variety of Italian research organizations and provides a basis for scaling to petaflops of performance.
• Univa has been a long-time supporter of Intel Cluster Ready and recently launched its UniCloud middleware, which makes it easy to deploy Intel Cluster Ready certified HPC clusters in private, hybrid, and public clouds (including RightScale and Amazon EC2).

That’s it for this year’s winners. I’d like to add my personal thanks to all of them.

Stay tuned for more information. I’ll be focusing in on some of these technologies and solutions in more detail in future blogs.

Intel Cluster Checker can run a variety of tests depending on what the user is trying to accomplish. For general wellness of a cluster, Intel Cluster Checker offers five levels of thoroughness. For automation, we'll focus on levels one, three, and five.

Wellness Level One is a very short run of tests that check basic connectivity throughout the cluster as well as basic uniformity checks amongst the nodes.  This test focuses on BIOS settings, and processor, memory and system configurations. The level one tests are quick and show that the cluster is online and ready for use.

Wellness Level Three is the default run level for Intel Cluster Checker. It builds on level one and includes more rigorous modules that test parameters like disk and memory bandwidth, MFLOPS, and network performance. It also performs an in-depth hardware uniformity test along with an Intel MPI Collectives and message Integrity test. It takes a bit longer to run this check that assures the performance of the hardware is up to par.

Wellness Level Five adds a packages test, comparing currently installed packages with a generated list of expected packages at a given time. It also runs the HPCC module (a performance benchmark test). The level five tests can take a bit longer but assures the users that all the pieces of the system are working in harmony. This test is useful for the admin to make sure a user hasn’t installed or uninstalled anything that may affect the cluster.

For my clusters, I have set up the following schedule that runs these wellness levels at various times.

• Level One - set to run each weekday
• Level Three - set to run once a week, on Saturday
• Level Five - set to run monthly, on the first Sunday

For our lab with multiple smaller clusters this automation schedule is perfect, allowing users to login to one of the clusters, see when each of the wellness levels passed and know that their jobs will run as expected.

Every cluster has a different purpose and a different load. Running tests at the intervals or frequency I have set may not be feasible or needed for every cluster. However, automating the test runs on a schedule that meets your needs will help assure that your cluster performance remains optimal, and frees up the cluster(s) for jobs during normal business hours.

References:

>> Review Part 1

>> About Intel Cluster Checker

>> Intel Cluster Checker Knowledge Base

Whether you're a cluster user that expects optimal performance and functionality every time you run a job, or a system administrator that needs to keep the cluster in perfect working order for your users, running checks on a regular basis is important!

A great way to ensure that your HPC system, certified Intel Cluster Ready, remains in the same great shape as when it was first built, is to run the Intel® Cluster Checker tool regularly.

While it’s easy enough to run Intel Cluster Checker on your cluster(s) once a week, I've found there are times when I was too busy, or simply forgot to run a check. For this reason, and because I have multiple clusters to check, I developed a method to automate Intel Cluster Checker runs. This automated solution runs Intel Cluster Checker and reports any errors directly to me and/or the system administrator. Not only does automating this process save the administrator a substantial amount of time each week, it will eliminate the chance of missing a run, and ensures that the cluster remains in optimal health.

A Cluster That Checks Itself

In order to automate the process, it is important that your Intel Cluster Checker passes a manual run first. To do this, make sure the configuration file is optimized to ensure the cluster is operating at its best. With the perfect configuration on hand, a script is then needed to setup the system, run Intel Cluster Checker, and report the results.

For the first phase of the process, I wrote a script that sets up the environment and runs the most in-depth wellness check in Intel Cluster Checker. I used cron to schedule my script to execute once a week. After the results are complete, the script updates the message of the day to show when the last check ran and the results. If the check fails, the log file created by Intel Cluster Checker will be copied to the specified directory where it can be accessed and analyzed.

The cluster is now setup to check itself, in the middle of the night, and report its status upon login!

Read Part 2 for more about scheduling test runs, and the different levels of wellness tests in Intel Cluster Checker.

References:

>> About Intel Cluster Checker

>> Intel Cluster Checker Knowledge Base

Check out this paper, Bringing Consistency to a Heterogeneous Research Cluster, related to upgrading the GreenPlanet cluster at UC Irvine to Intel Cluster Ready architecture.  GreenPlanet is an organically grown HPC cluster that went from seven nodes to over three hundred in just a couple years.  The system grew  as principal investigators pooled grants and added more nodes to the system in batches.  Adding nodes in this fashion means the system is comprised of many different types of compute nodes which can become a pretty tough administrative task.  In the paper, Dr. Nathan Crawford states, “We were trying to keep track of all the differences in spreadsheets or in our heads, and like any academic institution, we had the problem of graduate students moving on and taking their knowledge of system details with them.”

Enter Intel Cluster Ready.  UC Irvine combined an equipment move with an upgrade of their software stack using Clustercorp Rocks+ 5.3.  This version has built-in functionality to install the necessary components to meet the Intel Cluster Ready architecture specification including a Roll to install Intel® Cluster Checker.

The result: changes to over 100 nodes in the system.  “The process of getting certified cleaned up a lot of underlying issues, and immediately
made the system much more stable,” said Crawford. "We’ll have more jobs running to completion, and scientists can spend less time trying to
debug underlying hardware and software problems.”

Scientists doing more science.  Nice.

>> Download the full case study

Your hardware vendor may be able to provide the information you need to install and run open source HPC applications. Another great resource is the HPC Advisory Council. The Advisory Council provides best practice recommendations and installation guides for a variety of HPC applications, as well as training opportunities, an online forum for interacting with HPC experts, and cluster resources you can use for testing, optimizing, and benchmarking your applications.

Intel has been working with the HPC Advisory Council to develop best practice recommendations and installation guides for popular open source HPC applications running on Intel Cluster Ready certified systems. Installation guides are available now for Adaptive Mesh Refinement, BQCD, CP2K, CPMD, Espresso, Himeno, LAMMPS, NAMD, NWChem, OpenFOAM, and WRF. If you don’t see the application you’re interested in, stay tuned. More are coming. Or contact the HPC Advisory Council. They’re always looking for input from HPC users.

Update - these installation guides are now available: Amber 11, CESM, GADGET-2, GROMACS, MrBayes, NEMO, Octopus, and OpenMX.

>> View all Intel Cluster Ready Installation Best Practices

For more about the HPC Advisory Council, go to hpcadvisorycouncil.com

For more Intel Cluster Ready information, visit intel.com/go/cluster

Read the HPC Advisory Council and Intel Cluster Ready Technology Brief

As an industry, we need to continue to push HPC technologies into mainstream and make it easier to access for people like Marshall.  Computational simulations can drastically reduce research cost and speed time to results - not just in academia but in commercial environments as well.  HPC is no longer a niche technology but is viable for broad academic and commercial use.  There are a lot more discoveries and advancements out there that we can get to faster with bigger and better simulations.

In case you missed it, read Teen Discovers Promising Cystic Fibrosis Treatment

Seamless Transitions...

I'd add that the Intel Cluster Ready architecture applies in the cloud as well.  In fact, using cloud images based on Intel Cluster Ready architecture allows more seamless transitions between clusters running in the cloud to physical clusters running locally.  As the video points out, cloud provides an entry point into using HPC or even a try-before-buy approach to HPC clusters.  Cloud bursting benefits as well, since the application that runs on the architecture locally would see the same architecture in the cloud.

The common architecture enables the applications to better understand and execute on top of the computing environment without being tied to the exact details of the solution.  In other words, you can get better mileage with a common architecture in both the cloud and local clusters.

The results are then recombined to deliver whatever information design team is looking for, whether it's how the latest chassis protects the driver in a 60 MPH collision, or how a potato chip holds up between factory and bowl.

Intel® Cluster Ready provides an optimized platform for parallel computing, and helps to ensure that registered applications run seamlessly on certified systems. But that's only half the battle. Unless applications make efficient use of all that parallel computing capacity, they aren't going to deliver the kind of value that justifies the cost of a high performance cluster.

In other words, if you build a great cluster they will come, but only if you - or someone else - also builds great, compatible applications. Of course, writing good, parallel software code isn’t easy. It's tough enough writing applications that can efficiently use 12 processor cores in a high-end workstation. Writing code that can fully utilize 120 cores in a 10-node cluster, or 480 cores in a 40-node cluster, takes another level of expertise.

Optimized development tools can help - a lot. I encourage you to take a quick look at this video interview from SC10, or forward it on to one of your software-developing colleagues. Doctor Clay Breshears and Werner Krotz-Vogel, talk about Intel Cluster Studio 2011, a comprehensive bundle of tools for software developers who want to write highly optimized applications for Intel processor-based clusters (compilers, libraries, analyzers, and a lot more). Especially noteworthy in this package is a variety of new threading mechanisms for task, data, and vector parallelism. Clay and Werner provide an overview and discuss some of the features. They also give a shout out to Intel Cluster Ready and the more than 150 vendors who are delivering compatible systems and applications.

Enjoy the video!

Check out the Intel® Cluster Studio 2011 - a tool suite for MPI developers on HPC clusters.

Learn more about Intel® Cluster Ready