We’re in a really crucial phase right now, which we call the early user phase – we have probably 15 research groups that are on the systems right now all doing science and reporting back results. We just got a notification the other day from one user saying the performance was spectacular, so now’s a time for debugging the software stack and identifying any nodes that are sort of straggler nodes, maybe performing slightly slower than expected.
Read the Full Story.
Based on the most recent release of the Top500 List, Slurm Workload Manager continues to be the most widely used on the fastest of the fast: 33 per cent of the top 15 supercomputers use the product.
Slurm, an open-source workload manager designed for the most demanding HPC environments, originated at Lawrence Livermore National Laboratory (LLNL) 10 years ago and has evolved over time with the contributions of more than 100 developers. It remains an important workload manager at LLNL, providing scheduling and other functionality to their Sequoia supercomputer, currently number two in the Top500 and ranked the fastest in the previous Top500 List.
The other supercomputers in the 15 fastest supercomputers using Slurm are Stampede at TACC; Tianhe-1A in China; Curie at the CEA in France; and Helios at Japan’s International Fusion Energy Research Centre. Beyond the top 15 systems, SchedMD, the organisation overseeing the code base for Slurm, estimates that as many as 30 per cent of the supercomputers in the Top500 list are using the open-source workload manager.
We built Slurm to schedule efficiently resources for the world’s biggest systems and, through simulation, have proven its scalability to an order of magnitude higher than the currently largest systems,” said Moe Jette, CTO of SchedMD. “It’s now one of the most widely used workload managers in the Top500. As we move to Exascale computing requirements, Slurm is the workload manager best positioned to schedule jobs at that scale.”
Outside of the large supercomputer centres, Slurm is gathering momentum. HPC computer manufacturers Bull and Cray frequently provide Slurm as part of their solutions, and Bright Computing now offers Slurm as the default workload manager in Bright Cluster Manager.
This story appears here as part of a cross-publishing agreement with Scientific Computing World.
In this video, KRQE reports that the controversial Encanto supercomputer may be split up among New Mexico’s three research universities. Purchased in 2008 for $11 Million, the supercomputer did not bring new business to the state as promised and racked up a stack of unpayed bills. The host datacenter for the system is owned by Intel and has been targeted for demolition.
Editor’s note: John West, former owner of insideHPC called the Encanto business model into question as far back as 2008.
In this video, Intel CTO Justin Rattner presents Connecting to the Future at the company’s 2012 IDF Developer Forum event in San Francisco.
To run on Intel Xeon Phi coprocessors, CUDA kernels need to be modified. At the moment, this needs to be done by hand. While it is technically possible to run CUDA on Phi coprocessors, products such as CUDA-86 do not currently generate code for these devices. An OpenCL compiler for the Intel Xeon Phi coprocessor is coming. This means that CUDA programmers can consider Wu Feng’s CU2CL CUDA-to-OpenCL source translator to port their code. In the future, an LLVM translation project might be able to create executable code for the Phi.
This is a deep-dive feature story that is well-worth a look. Read the Full Story.
TACC has released Most Pixels Ever: Cluster Edition, an open source software tool that allows researchers, especially those in the humanities, to create interactive, multimedia visualizations on high resolution, tiled displays.
The goal is to make visualization tools easier for humanities researchers to use,” said Rob Turknett, digital media, arts and humanities coordinator at TACC. “The proliferation of digitized textual, visual and aural resources is a great boon for the humanities, offering opportunities for new kinds of scholarship, but it also brings a new complexity.”
Supported by a startup grant from the National Endowment for the Humanities titled “A Thousand Words: Advanced Visualization for the Humanities,” the software is based on a language called Processing, a programming toolkit that makes it easier for people to create visualizations. The software is already in use at the University of Texas at El Paso and the University of Texas at San Antonio’s Center for Simulation Visualization and Real-time Prediction. Read the Full Story.
In this video from SC12, Lars Koesterke from the Texas Advanced Computing Center describes how hundreds of science applications are being ported to the the new Stampede supercomputer powered by the Intel Xeon Phi.
Using density functional theory, a quantum mechanical modeling method used in physics and chemistry to investigate the electronic structure of molecules, the researchers calculated the interactions of more than 200 atoms using Ranger. The simulations helped the group identify the presence of an intermediate chemical in the reaction and determined that it was in fact ketenylidene. The acetic acid-to-ketenylidene path combines dehydrogenation (oxidation) and the deoxygenation of the acetate, “which are crucial steps for biomass conversion into more valuable industrial chemicals,” the authors wrote.
Read the Full Story.
What is the best way to manage an HPC cluster serving a multi-user tenant base? We asked David Gignac, Senior Systems Administrator at the Texas Advanced Computing Center (TACC). David is responsible for managing “Alamo,” a 96-node cluster that’s part of FutureGrid, a high-performance grid test bed for new approaches to distributed computing. Funded by the National Science Foundation, FutureGrid comprises 920 nodes distributed across eight clusters at sites in the U.S. and Germany, including TACC. Gignac has managed Alamo for three years as part of the five year FutureGrid study, giving him unique insight into the challenges of managing an advanced multi-user tenant HPC cluster.
insideHPC: What do you do for FutureGrid?
insideHPC: What do you do when you’re not managing FutureGrid?
David Gignac: In addition to FutureGrid, I am also responsible for managing more than 2,600 servers for a variety of other research projects at TACC. As with any network administrator, there are only a certain number of boxes I can realistically manage effectively. When you talk about clusters, the management requirement goes through the roof. I need to have a good solution to help me manage this complexity.
insideHPC: How do you keep up?
David Gignac: I depend on good cluster management applications. When I took on administration for Alamo, I reviewed a number of advanced management suites. With all of my other responsibilities, my top criterion was minimizing the amount of time I spend managing each cluster. I looked at cluster management software from all the major vendors including Bright Cluster Manager, Cobbler/LOSF, Platform Computing products, Rocks and xCat.”
insideHPC: How did you choose the cluster management solution for Alamo?
David Gignac: My decision was based on minimizing the time required to manage the cluster: automatic time-consuming tasks and reducing complexity— balanced with providing a high level of service to our users. Drilling down, I needed a solution that would minimize the number of custom scripts I was required to write and something that would provide maximum ‘at a glance’ visibility into the health and operations of each cluster. In addition, I looked for something that would integrate seamlessly with Alamo’s job schedulers: Moab, Torque, Slurm and SGE; yet is nimble to accommodate simultaneous requests from researchers. In the end, I selected Bright Cluster Manager.
insideHPC: Three years later, how’s it going?
David Gignac: It’s been a great run. Bright and Fedora EPEL distros have saved a tremendous amount of time for me. Bright’s image-based provisioning lets me reconfigure Alamo on the fly to meet the specific needs of each researcher’s compute jobs. I click on a check box and the cluster management suite installs a server, sets up a client and I’m done. Further, Bright’s ease of use and full integration with job schedulers have produced major time savings. I don’t need to spend hours writing and maintaining scripts because everything just works. I get dedicated product support, so I don’t waste time searching forums and message boards for answers. In addition, I can easily reproduce any testing environment in minutes and rapidly deploy a new environment.
insideHPC: What’s next?
David Gignac: Cloud bursting. I think there’s an opportunity to experiment with hybrid cluster solutions. Bright lets me manage on-premise and remote cloud-based clusters seamlessly. It all looks the same through the management suite portal. I want to work with FutureGrid participants to test it in the program’s next two years.
insideHPC: And in your free time?
David Gignac: I certainly have more of that now, in spite of all the clusters I manage. Because of the time savings, I am spending more time making improvements on the clusters.
None of us expected to win, but we didn’t expect to lose either…we were thrilled,” said Craig Yeh, a third year Computer Science major at The University of Texas at Austin. “The win validated all of the work we did since May leading up to the competition. I highly recommend this experience to other students at The University of Texas.”
The winning student team members include Andrew Wiley, Reid Douglas McKenzie, Michael Teng, Anant Rathi, Craig Yeh, and Julian Michael. The team and TACC partnered with Dell, Nvidia, and Intel to design and build a hybrid, power-saving system that integrated GPUs, Intel processors, and a new-generation Dell chassis. The company sponsors provide the hardware and travel funds, while the TACC mentors worked side by side with the students to teach them the fundamentals of cluster construction, systems administration, and program optimization. Chevron and Mellanox also served as sponsors of this year’s team.
It’s a real-world situation,” said John Lockman, the team’s lead mentor and a member of TACC’s High Performance Computing group. “For example, a data center in industry might need to expand, but can’t due to financial or space constraints, so they have a limited amount of power and a scientific workload that they have to accomplish in a reasonable amount of time.”
Read the Full Story.
We try to keep this page up to date, but you can always find our latest works on our RichReport YouTube Channel.
This year at SC12, we shot over 50 video interviews. It was a lot of work, but we want to bring you the very best of what this amazing conference had to offer.
SC12 Videos (Alphabetical by Vendor Name):
Adaptive Computing
Adaptive Computing SC12 Booth Theater
Aeon
Allinea
Altair
AMD
Asetek
Bull
CAPS-Enterprise
Colfax International
Cycle Computing
DDN
Dell
Gnodal
HPC Advisory Council
IBM
IDC
Inktank
Intel
Intersect360 Research
NAG
Nirvana
Numascale
Nvidia
Mellanox
OpenSFS & EOFS
Panasas
Penguin Computing
Rogue Wave Software
Samplify
SC12 Committee
Scalable Informatics
Seneca
SGI
Solarflare
Spectra Logic
Supermicro
Sugon
Texas Instruments
The Portland Group
VMware
Xyratex
The Texas Advanced Computing Center (TACC) expects its new Stampede supercomputer to pursue complex science problems with more detail and complexity than before. Debuting at #7 on the TOP500, the machine will deliver 20 times the peak performance of its predecessor, Ranger, which is scheduled to end its five-year career in early February 2013. Stampede, which costs $27.5 million, will move into formal production in January as one of the world’s most powerful computers devoted to “open science” research.
New technologies enable you to see the world in new ways,” TACC Director Jay Boisseau said, “enabling scientists to build more complex models of everything from tiny cells to global climate patterns. Sometimes you pass some resolution thresholds and you see things in an entirely different way.”
Boisseau also said his team’s tight partnership with Dell and Intel was essential for the project. The machine features more than 6,000 Dell Zeus servers powered by Intel’s Xeon E5 processor chips and the new Intel Xeon Phi coprocesssor. Read the Full Story.
The new TOP500 list is out with a new world leader and more than a few surprises. Coming in at #1 is the newly built Titan Cray XK7 supercomputer at Oak Ridge with 17.59 Petaflops on the Linpack benchmark. As a hybrid system, Titan is powered by 299,008 AMD Opteron cores and 18,688 Tesla K20X GPUs.
The Sequoia supercomputer at Lawrence Livermore National Labs falls into second place with 16.32 Petaflops and Fujitsu’s K computer round out the Top 3. IBM BlueGene/Q powers the next two entries with the Argonne’s Mira coming in at #4 and the newly upgraded JUQUEEN at Germany’s Juelich becomes #5 as the most powerful system in Europe.
Surprise #1: The Intel Xeon Phi co-processor makes its debut in the Top10 with TACC’s Stampede supercomputer from Dell with 2.6 Petaflop/s.
Surprise #2: The highly anticipated Blue Waters supercomputer is not on the list. NCSA did not submit.
Here are some highlights from the list.
- 23 Petaflop systems total
- The U.S. is the leader with 251 system on the list with Europe coming in with 105 systems and China with 72
- 62 systems are accelerated by Nvidia GPUs
- 7 systems are accelerated by Intel’s Xeon Phi processors
- 226 systems use Infiniband and 188 use Gig Ethernet
- Intel processors are used in 76 percent of the Top500 systems, with AMD at 12 percent and IBM’s Power processors at 10.6 percent
- IBM has six of the top 10 systems and 192 total entries, followed by HP with 149 and Cray with 30
We’ll be hearing a lot about the TOP500 this week as there are a million ways to spin this thing. Read the Full Story.
Our Video Sunday feature continues with this preliminary simulation of the cosmic web. Sometime next month, the Mira supercomputer will complete tests of its new upgraded software and begin running the largest cosmological simulations ever performed at Argonne National Laboratory.
These simulations are massive, taking in huge amounts of data from the latest generation of high-fidelity sky surveys and crunching it into models of the universe that are larger, higher-resolution, and more statistically accurate than any that have come before. When it’s done, scientists should have some amazing high-quality visualizations of the so-called “cosmic web” that connects the universe as we understand it. And they’ll have the best statistical models of the cosmos that cosmologists have ever seen.
Read the Full Story.
Advertisement
insideHPC.com is a production of insideHPC, LLC. © 2006-2013 Sitemap
