Entries filed under “HPTC”

News specifically related to the application of HPC in science, engineering, and other areas not directly related to a commercial purpose.

Computational biology at TACC

By John West | Published April 23, 2009

TACC is reporting on work being done on its system related to computational biology and understanding how the molecules that advanced laboratory techniques have identified

“The largest unsolved problem in the computational biology of proteins is how to take a protein sequence and accurately predict the three-dimensional structure it folds into,” said Nick Grishin, professor of biophysics at The University of Texas Southwestern Medical Center and a Howard Hughes Medical Institute investigator. “These structures are functional as three-dimensional bodies — they bind some molecules, they catalyze reactions — and knowing that structure is extremely important for the functional understanding of proteins.”

Grishin is pursuing this “unsolved problem” of protein structure prediction using Ranger— one of the world’s most powerful supercomputers — at the Texas Advanced Computing Center (TACC). His computations have led to new insights on protein activity, potential disease cures, and a deeper understanding of how humans evolved into such complex creatures.

Link courtesy HPCwire.

Also posted in Applied HPC | Leave a comment

Matlab announces Argonne, Tesla using their tools

By John West | Published April 23, 2009

The Mathworks announced two customers this week using their tools in science and engineering. The first at Argonne

The MathWorks today announced that Argonne National Laboratory significantly reduced the simulation time of models built with the Powertrain System Analysis Toolkit (PSAT) by using Parallel Computing Toolbox and MATLAB Distributed Computing Server from The MathWorks.

PSAT is an industry-standard model that enables automotive engineers to assess design tradeoffs by simulating a large number of advanced powertrain configurations. Using two other MathWorks products, Simulink and Stateflow, Argonne, funded by the US Department of Energy, developed PSAT in conjunction with leading automotive manufacturers to help automotive engineers evaluate fuel economy and performance of advanced vehicle designs, including hybrid electric vehicles (HEVs) and plug-in HEVs. Argonne used Parallel Computing Toolbox with PSAT to execute simulations on a 16-node cluster—reducing the overall run time from two weeks to one day. Using MATLAB Distributed Computing Server also enabled Argonne to move its simulations with ease from a desktop computer to a cluster in one hour.

And the second announcement is that Tesla is using Matlab to evaluate design tradeoffs for the 2008 Tesla Roadster

The MathWorks today announced that Silicon Valley electric vehicle maker Tesla Motors employed Model-Based Design tools from The MathWorks to develop the 2008 Tesla Roadster, the world’s first electric production sports car.

To meet aggressive technology goals on a strict budget and timeline, the Tesla Motors design team relied on Simulink and MATLAB to model the entire vehicle and its major subsystems. Engineers combined individual models for each major system of the car, including the motor, battery, transmission, brakes, tires, power electronics, and control systems, into one full-system model, which they used to simulate and predict overall vehicle performance before building prototype vehicles. The simulated results were then compared against road-test results from prototype vehicles to refine the full-system model. This approach shortened overall design and test processes and helped Tesla Motors to deliver the 2008 Tesla Roadster for a fraction of the typical vehicle program cost.


Also posted in Applied HPC, Enterprise HPC, Green HPC | 1 Comment

NSF awards support science with cloud infrastructure

By John West | Published April 20, 2009

Found at HPCwire, news that NSF has awarded a series of grants to UW to support the cause of scientists using clouds for research

The University of Washington has won three recent awards from the National Science Foundation related to cloud computing.

…Howe’s project aims to provide that interactivity for tens of thousands of gigabytes of simulation results. He created a tool, GridFields, to visualize the polygonal mesh of climate simulation output, and is now working to redesign GridFields to be efficient in a cloud computing environment.

…[A second grant]  will prepare astronomers to deal with data coming from telescopes scheduled to come online in coming years, such as the Large Synoptic Survey Telescope, of which the UW is a founding institution. The telescope’s 27-foot mirror is connected to a 3.2 billion-pixel camera that takes pictures every 15 seconds. It is expected to record more than 30,000 gigabytes of data and detect more than 100 million astronomical sources every night.

“Cloud computing enables us to scale to the point where we can actually analyze that sort of data,” Connolly said.

There is a third grant to support the development of a cloud computing curriculum.

The advantage here is that, for problems where a standard scale out solution provides acceptable performance (not tightly coupled solutions), it will probably make sense for researchers to rent time than buy. No infrastructure problems, no admins, no technology obsolescence. But the existing infrastructures (EC2, etc.) are more complicated than scientists care to deal with. Projects like these, and the proteomics package discussed earlier last week, are bridging this gap.

Also posted in Cloud HPC | Leave a comment

Wild guess that noisy chips won’t lead to better weather simulations

By John West | Published April 20, 2009

Here’s an interesting idea from the NewScientist

While researchers are striving to make the models more realistic, they are limited by the processing power of the supercomputers that run climate models, Palmer says. “That determines how fine of a grid we can solve the equations on, because of the computing cost,” he says.

Adding a degree of randomness to a particular model and running it multiple times could provide a cheaper way to increase realism, Palmer and colleagues argue, as it could be a “poor man’s surrogate for high-resolution models”.

But, random number generation without special hardware can be expensive, and often not very random. The solution? Well, use special hardware…but not the way you may be thinking

A way around this could be to use cheap hardware – low-cost computer chips that generate output with some random noise due to the way electrons bounce through them. Essentially, those chips produce the necessary randomness for free.

“It’s very speculative,” Palmer says. “But if it can be made to work, it would make much more efficient use of power.” The idea of adding randomness into the models is “very interesting and might be helpful for some cases”, says Reto Knutti of the Swiss Federal Institute of Technology in Zurich, “but in my view it will not solve all problems.”

I tell you though, I’m not sure about this. Randomness in one, controlled, area of the simulation I guess could be a good thing. But if you have generally noisy chips, how do you trust any part of the calculation? How do you know your noise isn’t somewhere you don’t want it, for example? I guess I’m not alone; responses from the Twitterverse following HPCwire’s post of the original story aren’t postive

ianfoster: Sounds like nonsense to me — RT @HPCwire HPC News: Cheap and Noisy Chips Could Improve Climate Predictions

rplzzz:…I’m skeptical.If you use bit errors for “randomness”,how do you ensure that the errors are in the low bits instead of the exponent?


Also posted in Computing Research, HPC | 2 Comments

Compete for access to the world’s fastest supers

By John West | Published April 16, 2009

The DOE has recently announced its call for proposals for those seeking time on select supercomputers through the INCITE program. From the DOE HPC page

For the seventh consecutive year, the Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program invites proposals for large-scale, computationally intensive research projects to run at America’s premier leadership computing facility (LCF) centers, established and operated by the U.S. Department of Energy (DOE) Office of Science. The INCITE program awards sizeable allocations (typically, millions of processor-hours per project) on some of the world’s most powerful supercomputers to address grand challenges in science and engineering, such as developing new energy solutions and gaining a better understanding of climate change resulting from energy use.

This year the program will award 1.3 billion process hours, and is open to government, commercial, and academic participants. More info in the release, and proposals are due July 1.

Also posted in Computing Research, HPC | Leave a comment

Sun launches new HPC products

By John West | Published April 14, 2009

At Sun’s Partner Summit on Tuesday the company announced a slew of new products for HPC. The head of Sun’s HPC sales organization, Marc Hamilton, hits some performance points in a light post that you can read here, and Sun’s own press release is here.

Sun logoSun’s scalable new HPC systems, including the next-generation Sun Constellation System, are expected to power some of the world’s leading HPC facilities, and can address a broad range of HPC applications requiring high performance, high throughput, large memory and fast I/O.

Sun announced new flash-ready Nehalem-based HPC blades, a set of rack servers that upgraded existing lines with Nehalem, a cooling door, new IB networking options, and the Sun Lustre Storage System — “a complete Lustre and Open Storage hardware solution.”

The high end new blade and chassis create quite a dense solution

With the Sun Blade 6048 chassis, the Sun Blade X6275 server module also provides extreme density with 48 physical blades per rack — supporting 96 nodes of two-socket, quad-core processors per node, resulting in a total of 768 processor cores and nine teraFLOPS of peak performance in a single 42U rack. This represents up to 71 percent more cores/rack than IBM (as compared to the IBM BladeCenter H) and 50 percent more cores/rack than HP. With 2.25 teraFLOPS of peak compute capacity and Linpack efficiency of 89 percent per every shelf of 12 blades, customers can expect up to two teraFLOPS of actual computational power.

As both HPCwire and The Register point out, the 6275 is pretty innovative, cramming two nodes onto a single blade. Good stuff, although that’s basically the same total peak FLOPS you’d get out of Sun gear stuffed with Opterons, and less memory per node. Each half of the 6275 only supports 12 memory slots, not the full 18 that the Nehalems are capable of supporting. (Video of the X6275 here courtesy of Sun’s HPC Watercooler).

The 6275 also has a less dense little sister, the 6270, which supports half the sockets, and lacks the on board IB (dual GbE ports instead), but does support the full 18 DIMM slots. And while both blades support flash memory, the 6275 supports 2 24 GB modules while the 6270 supports a single 16 GB module.

The 6048 chassis also supports the Sun Blade 6048 Quad Data Rate InfiniBand Switched Network Express Module

Each node on a Sun Blade X6275 server module offers onboard QDR IB HCAs that interface directly with the integrated Sun Blade 6048 IB QDR Switched NEMs in the Sun Blade 6048 chassis, and the Sun Blade 6048 IB QDR Switched NEMs directly connect to Sun Datacenter IB switches in high-bandwidth fat-tree topologies or to other Sun Blade 6048 IB QDR Switched NEMs in low-cost 3D torus configurations.

With the QDR IB NEM you can connect all 96 nodes in a cabinet without external hardware, in case you need a really high performance but compact system.

Sun also has a Dual Port 4x QDR PCIe ExpressModule Host Channel Adaptor for clusters with multiple communications fabrics, as well as the Sun Blade 6000 Virtualized Multi Fabric 10GbE NEM which Sun claims will offer buyers a 20 to 1 cabling reduction. That’s a lot of high bandwidth networking options, which fits well with Sun’s high end target for this blade. The 6275 also supports flash (from HPCwire’s feature)

A SATA interface is also available to connect to an optional Sun flash module, which offers 24 GB of high performance storage per node. It’s designed for users interested in saving state, having a scratch data area, or booting an OS. Since the flash module is hooked up to a SATA controller, to the apps it looks like a hard drive.

In addition, the Sun Lustre Storage System “will enable customers to scale online capacity from 48 terabytes to multiple petabytes, and scale I/O performance from 1 GB per second to more than 100 GB per second.”

Interesting stuff, with a lot of focus on the high end, but despite the DoD Mod Program’s example, I think I’d wait for the dust to settle before I put tens of millions of dollars into a super from a company that wasn’t sure it wanted to be in business.

Also posted in HPC | 3 Comments

Software simplifies proteomics research on Amazon cloud

By John West | Published April 14, 2009

News posted at HPCwire late last week of a new software package that simplifies the task of using Amazon’s Elastic Compute Cloud for proteomics research

Researchers at the Medical College of Wisconsin Biotechnology and Bioengineering Center in Milwaukee …have developed a set of free tools called ViPDAC (virtual proteomics data analysis cluster), to be used in combination with Amazon’s inexpensive “cloud computing” service, which provides the option to rent processing time on its powerful servers; and free open-source software from the National Institutes of Health (NIH) and the University of Manitoba.

…One of the major challenges for many laboratories setting up proteomics programs has been obtaining and maintaining the very costly computational infrastructure required for analysis of the vast flow of proteomics data generated by mass spectrometry instruments used to determine the elemental composition as well as chemical structure of a molecule, according to senior investigator, Simon Twigger, Ph.D., assistant professor of physiology.

…”The tools we have produced allow anyone with a credit card, anywhere in the world, to analyze proteomics data in the cloud and reap the benefits of having significant computing resources to speed up their data analysis,” says lead author Brian Halligan, Ph.D., research scientist in the Biotechnology and Bioengineering Center.

More in the release, or you can find detailed instructions for using the package on the web site.

4 Comments

In time of turmoil, Sun ramping up HPC efforts

By John West | Published April 6, 2009

Sun’s Marc Hamilton wrote on his blog last week about changes lined up for the HPC team there

Later this month, we will be announcing the newest upgrades to the Sun Constellation System line, providing significant new performance boosts not only in CPU power, but in storage and networking performance. We have already sold PetaFlops of Sun Constellation Systems attached to PetaBytes of storage. Now we’ve put together a global HPC Sales team to help bring Sun Constellation System to an event broader set of commercial, education, and research HPC customers, and I’m really excited to be leading that new team.

So a hardware refresh and a reinvigorated (possibly larger?) sales team. Marc’s post also has a video introducing the changes he has planned for the new organization.

Also posted in Business of HPC, HPC | 1 Comment

Computing research that changed the world symposium slides up

By John West | Published April 2, 2009

A quickie from the CCC blog

Slides from all speakers at the remarkable March 25th Library of Congress symposium “Computing Research that Changed the World:  Reflections and Perspectives” are now available:

http://www.cra.org/ccc/locsymposium_slides.php

Videos of all talks will be available soon.


Also posted in Computing Research | Leave a comment

10th Russian Top50 released

By John West | Published April 2, 2009

The 10th annual Russian Top50 list was released on Tuesday. You can find the list here, but unless you can read Russian it won’t do you much good. Happily, though, they emailed the announcement to me in English. Here are some highlights

The 10th edition of the rating has demonstrated further performance growth of the supercomputer systems in the CIS. Since the latest publication the total peak performance of such systems has grown by 14.95% and reached 510 TFlop/s (trillions floating-point operations per second). The total performance tested by Linpack has also grown by 15.33% and reached 382.6 TFlop/s within half-year. Thus, the average peak performance has for the first time exceeded 10 TFlop/s and amounts to 7.65 TFlop/s as tested by Linpack.

14 new systems were added this year, and 47 of the Top50 exceed a TFLOPS in performance. HP has the top slot, and the Russian-made SKIF MSU is in second

The supercomputer MVS-100K developed by Hewlett-Packard has maintained the leading position in the rating of Supercomputers.ru. The system is installed in the RAS Joint Supercomputer Center. The performance of the updated system amounts to 71.28 TFlop/s as tested by Linpack and 75% of the peak performance (95.04 TFlop/s).

The supercomputer SKIF MSU ‘Chebyshev’ has ranked second. The supercomputer was jointly developed by the Lomonosov MSU, the company ‘T-Platforms’ and the RAS Institute of Programming Systems. It is located in the MSU Research Computer Center. The real performance of the SKIF MSU ‘Chebyshev’ amounts to 47.3 TFlop/s as tested by Linpack which is 78.9% of the peak performance (60 TFlop/s).

HP actually has several systems in the top 10, including slots 6 and 7 along with number 1.

The amount of systems based on Intel chipsets in the 10th edition has decreased insignificantly, from 40 to 37, although they are still dominating. The rating contains 7 systems based on AMD chipsets (compared to 4 in the previous edition), 5 systems based on IBM chipsets (unchanged) and 1 system based on HP chipsets (unchanged). There is a continuous increase in the amount of CPU cores: in this edition it is 128 or higher, with 15 systems having over 1024 cores.


Also posted in HPC | Leave a comment

STFC Daresbury Laboratory’s new Tesla, Nehalem combo cluster

By John West | Published March 30, 2009

More news from the UK’s Streamline Computing, this time about a new system to be installed at the UK Science and Technology Facilities Council’s Daresbury Laboratory. This system is interesting in that it uses the new Xeon 5500′s (Nehalem-EP) and the QuickPath Interconnect along side NVIDIA’s Tesla graphics cards.

Technical specialists from NVIDIA and Streamline Computing will work together to ensure that the cluster runs to maximum efficiency but some early research from Streamline Computing has already demonstrated 10 fold speedups in CUBLAS and CUFFT libraries.  Further collaborations may also be possible with software tool and compiler companies to test new tools for GPU computing.

…The cluster has been designed to show how effective GPU Computing can be for many classes of application but linked with Nehalem they also allow Daresbury to ensure the minimum of bottlenecks when connecting GPUs through Infiniband in each node to high performance storage systems. This additional cluster enhances a facility installed in the previous year which utilised Harpertown and Seaburg based nodes with the then fastest IB solution with DDR and PCI-E Gen 2 connect X cards from Mellanox.

The release isn’t up at the company’s site just yet, but you can find it at HPCwire.

Also posted in HPC, New Installations | Leave a comment

Oxford installs supercomputer from Streamline Computing

By John West | Published March 29, 2009

In mid-March I had an email show up in the insideHPC news hopper with a pointer to news that the University of Oxford had just installed a new cluster from Streamline Computing (a division of Concurrent Thinking, an HPC vendor in the UK).

The High Performance Computing (HPC) cluster was implemented by Streamline Computing and provides over 4TFlops of peak performance across the Dell SC1435 & R905 computing nodes based on AMD quad-core processor technology.

Fairly standard news, though I wasn’t clear on the size of the system so I sent back an email asking about that (the answer:  there are 68 compute nodes (8 cores per node) plus a head node & redundant head node as well as 5 X 4 socket SMP nodes all distributed across 3 racks). I was also intrigued by another paragraph in the release

To date, just 6 months after commissioning, the cluster has executed nearly 2 million jobs, ranging from those that take just a few 10s of minutes to those that are still executing since the day the machine was made available to users.

I was curious to know if there had really been jobs executing since day 1 without interruption, or whether someone had simply misunderstood something along the way (writing a press release often looks like a big game of telephone), and so I also asked about that line.

Turns out that, yes, at the time of this release the system has indeed been running jobs since day one. I know this isn’t a huge machine, but that’s still pretty impressive. HPC that “just works” is a big part of penetrating the low end market. Good job.

Also posted in Datacenter operations, New Installations | Leave a comment

Academic HPC discounts

By John West | Published March 29, 2009

A couple weeks ago I invited vendors with academic discounts to send me info and I’d do a round up here in response to what looked like a growing trend. I only got a few responses, but they are good ones. So if you are in an academic setting and short on cash (doesn’t the former imply the latter?) then you may want to check out these deals.

Scalable Informatics logoScalable Informatics Think Smart! Academic & Research Stimulus Sale

Scalable Informatics announces Think Smart! our academic/research stimulus sale. Now through July 31, 2009 we are offering significant savings off  list price on JackRabbit high performance storage servers to university and other research groups. And don’t forget to check back as we’ll be adding Pegasus-GPU to the list shortly.

Discounts and details are listed at the Scalable Informatics site.

SiCortex logoSiCortex

SiCortex has two special offers through May 15, 2009 to help universities and colleges make parallel computing accessible to their students.

  • Classroom environment, 72-processor, desk-side teaching & development system for $14,999 (37% off)
  • Parallel Computing Lab-in-a-Box: multiple desk-side systems plus curriculum assistance

Equip your teaching lab with multiple SC072-PDS systems and take advantage of on-site, expert curriculum development assistance. Leverage the expertise of parallel processing computer scientists to jump start your course development, and tap into a network of peers who are advancing parallel processing curricula at their universities. Prices start at $65,000 for 5 systems. Contact us at [email protected] for further details.

More information on both offers here.

Sun logoSun 2009 Matching Grant Program For Education (from Marc Hamilton’s blog post)

Just in time to help school’s balance their budgets, Sun’s 2009 Matching Grant Program for Education is here. So no matter if you are trying to build a world class supercomputer to simulate unpredictable tornadoes like was done recently on Ranger, TACC’s Sun Constellation System supercomputer, or just need to buy a few servers storage, qualified educational institutions can save big on Sun servers, storage, workstations, and software between February 18th and June 19, 2009.

More information here.

Also posted in Business of HPC, Computing Research, HPC | 2 Comments

CERN and Argonne use science clouds for computing

By John West | Published March 26, 2009

Their phrase, not mine…here, I’ll prove it. From the release

A novel system is enabling high energy physicists at CERN in Switzerland, to make production runs that integrate their existing pool of distributed computers with dynamic resources in “science clouds.” …The integration was achieved by leveraging two mechanisms: the Nimbus Context Broker, developed by computer scientists at the U.S. Department of Energy’s Argonne National Laboratory and the University of Chicago, and a portable software environment developed at CERN.

Researchers working on A Large Ion Collider Experiment (ALICE) needed additional resources to complement their fixed deployment, so the research wanted to answer the question of how dynamically provisioned resources could be integrated into that pool to support surge requirements without requiring users to leave their existing (and familiar) compute environment.

The CernVM technology was originally started with the intent of supplying portable development environments that scientists could run on their laptops and desktops. A variety of virtual image formats are now supported, including the Xen images used by the Amazon EC2 as well as Science Clouds. The challenge for Harutyunyan was to find a way to deploy these images so that they would dynamically and securely register with the AliEn scheduler and thus join the ALICE resource pool.

Here the Nimbus Context Broker came into play.  The broker allows a user to securely provide context-specific information to a virtual machine deployed on remote resources. It places minimal compatibility requirements on the cloud provider and can orchestrate information exchange across many providers.
“Commercial cloud providers such as EC2 allow users to deploy groups of unconnected virtual machines, whereas scientists typically need a ready-to-use cluster whose nodes share a common configuration and security context. The Nimbus Context Broker bridges that gap,” said Kate Keahey, a computer scientist at Argonne and head of the Nimbus project.

This is interesting. What I like about it is that it combines the two models of commercial and scientific computing for the user’s benefit without requiring the user to leave the environment with which she is already familiar.

The new system dynamically deploys a virtual machine on the Nimbus cloud at the University of Chicago, which then joins the ALICE computer pool so that jobs can be scheduled on it.

…According to Keahey, one of the most exciting achievements of the project was the fact that the work was accomplished by integrating cloud computing into the existing mechanisms. “We didn’t need to change the users’ perception of the system,” Keahey said.


Also posted in Cloud HPC, Enterprise HPC, HPC | 3 Comments

Ideas for improving computer science education

By John West | Published March 23, 2009

A post last week at the new (I think) ACM blog highlights how some of the ideas from a 2007 PCAST report on NITRD are being used by policy makers today to improve computer science education in the US

We outline four recommendations (and specific legislative language for the wonks out there):

  • Promote computing education, particularly at the K-12 level, and increased exposure to computing education and research opportunities for women and minorities as core elements of the NITRD program;
  • Require the NITRD program to address education and diversity programs in its strategic planning and road-mapping process;
  • Expand efforts at the National Science Foundation (NSF) to focus on computer science education, particularly at the K-12 level through broadening the Math Science Partnership program; and,
  • Enlist the Department of Education and its resources and reach in addressing computer science education issues.

…We must do more to expose kids to a quality computer science education program at the K-12 level, support teachers and bring innovative new curricula into the schools. Opening a serious education front in the NITRD program would be a good start to this ambitious goal.


Also posted in Computing Research, HPC | Leave a comment

Advertisement

ISC’13 Ad

Video Archive

insideHPC.com is a production of insideHPC, LLC. © 2006-2013 Sitemap