Share |

Content about Feature

October 6, 2010

Feature - Achilles tendon a blessing, not a curse

Sprinters lining up for the start of the Women’s 100-meter at the Beijing Olympics. Image courtesy LIM CK, under Creative Commons license.

Compared with other runners on this planet, humans are feeble.
If Olympic sprinters competed against mammals of comparable size, they would never even qualify for the finals. The top speed for an in-shape male human is normally between 15 and 18 miles per hour (24 to 29 kilometers per hour). The world record is 27 mph (43kmh), and that was sustainable for only a few seconds.
Meanwhile, horses have been clocked at about 48 miles per hour, wolves about 42, and the speed champion — the cheetah — at 70 miles per hour. (That’s about 77 kmh, 68 kmh, and 113 kmh, respectively.)
Even warthogs are faster than us.
But in the field of endurance racing however, we leave everyone else in the dust. Over long distances, a well-trained human can outrun a horse.
What is the ke

October 6, 2010

Feature - HPC adds a spark to EDF’s computing capacities

Image courtesy Zsuzsanna Kilian, stock.exchng

Jean-Yves Berthou is responsible for IT in the research and development area of Electricite de France — a major energy company in Europe. EDF’s 2,000 researchers use computing to work on a number of different projects, including areas such as minimizing CO2 emissions, alternatives to fossil fuels, and ensuring the security of electricity grids. Here, he describes the use and impact of high performance computing (HPC) at the company.
Why do you use HPC?
Berthou: In many cases physical experiments and testing are not possible, for example in the simulation of fuel assemblies and crack propagation in nuclear reactors, or in optimizing electricity production and trading. Even when experimentation is possible, numerical simulation can go beyond what is physically possible. However, experimentation still remains an indispensable tool.
 
In what application

September 29, 2010

Feature - New ISO standard for special mathematical functions

Image courtesy of webdesign-guru.co.uk.

A new standard for mathematical special functions in C++ has been published by the International Organization for Standardization, or ISO. The functions included in the standard are frequently used in applications of high-energy physics and other mathematical, scientific, and engineering disciplines.
“Vendors will slowly choose to incorporate these functions in libraries they produce,” said Walter Brown, the project editor for the standard. In fact, this has already begun to happen, according to Brown, who is based at Fermilab in the United States. Vendor adoption is desirable to ensure code portability.
ISO ensures that standards are kept up-to-date by reviewing them periodically. As part of that process, the C++ working group submitted a technical report proposing additions to the existing standard. After the report was approved in 2005, the section on special mathema

September 29, 2010

Interview - Kostas Glinos peers into his crystal ball

Image courtesy Corentin Chevalier, eScienceTalk

Kostas Glinos is a member of the European Commission’s Directorate-General for the Information Society and Media — and he just presented the European Grid Initiative with a 25-million euro contract in a brief ceremony onstage at the start of the EGI Technical Forum on 14 September. iSGTW caught up with him afterward, during a coffee break held in the middle of the poster session floor. We asked him about the significance of the EC’s backing, his hopes for EGI, and his ideas as to what it all means for the future.
iSGTW: What was the significance of the contract presentation today?Glinos: EGI is the culmination of an effort over eight years to look for a sustainable, long-term commitment from European countries, supplemented by the European Commission (EC). Of a projected 73 million euro cost, 25 million is coming from the EC, with the rest from member states. We n

September 29, 2010

Feature - Prospecting with High Performance Computing

This photograph, taken by S.D. Ellen of the U.S. Geological Survey, shows damage from the Loma Prieta earthquake which took place 17 October 1989. Image courtesy NASA.

ISGTW likes to take a look now and then at the world outside of grid computing, and see things from the perspective of end-users. Here, we learn about high performance computing at one of Europe’s largest oil and gas companies, Repsol, from Jesus Garcia—who is responsible for the company’s information technology.
iSGTW: Why do you use HPC?
Garcia: HPC is the only way that large amounts of seismic data can be processed quickly. When prospecting for oil and gas, there is significant commercial advantage in knowing where the most promising fields are, given the highly competitive nature of the sector.
iSGTW: In which application areas do you use HPC?
Garcia: The primary use of HPC within Repsol is for the processing of seismic data. Repsol uses a

September 22, 2010

Feature - Surfing for earthquakes

Aftermath of Haiti earthquake. Image courtesy UN Development Program

A better understanding of the ground beneath our feet may come from research by seismologists and an organization called RAPID—a group of computer scientists at the University of Edinburgh.
The very structure of the Earth controls how earthquakes travel and the amount of damage they cause. Therefore, a clear picture of this structure would be extremely valuable to earthquake planners — but it requires the analysis of huge amounts of data.
To help, the RAPID team developed a system that performs the seismologists’ data-crunching, and have made it easy to use by relying on an interface familiar to all scientists: a web browser.
Seismologists measure vibrations in the Earth at hundreds of observatories across Europe, which allows them to study earthquakes as they travel across countries and continents. By measuring the speed and strength of the vibrations at d

September 22, 2010

Feature - GPU-based cheap supercomputing coming to an end

Intel’s Sandy Bridge architecture places the processor and GPU on the same chip.
Image courtesy Greg Pfister.

Nvidia’s CUDA has been hailed as “Supercomputing for the Masses,” and with good reason – amazing speedups ranging from 10x through hundreds have been reported on scientific / technical code. CUDA has become a darling of academic computing and a major player in DARPA’s Exascale program, but performance alone does not account for that popularity: price clinches the deal. For all that computing power, they’re incredibly cheap. As Sharon Glotzer of UMich noted, “Today you can get two gigaflops for $500. That is ridiculous.” It is indeed. And it’s only possible because CUDA is subsidized by sinking the fixed costs of its development into the high volumes of Nvidia’s mass market low-end GPUs.
Unfortunately, that subsidy won’t last forever; its end is n

September 22, 2010

Project profile: PL-Grid

Legacies of older architectures exist along side the new in Poland, and not just in research computing: Here contemporary architecture in Warsaw reflects the past. Image courtesy Jaime Silva, Flickr, under Creative Commons license

Beyond being at the geographic center of Europe, Poland plays a central, leading role in Europe’s grid community. In 2009, Poland became the first country to form an independent and autonomous National Grid Initiative — an operational unit, based in a single country, set up to run a national computing infrastructure to support the European Grid Infrastructure (EGI).
With the close of the Enabling Grids for E-sciencE project, the health of the EGI is dependent upon such NGIs — and the technical process developed by Poland to move away from the parent EGEE ROC is a model other countries will look to. PLGrid itself is one of a series of NGIs to form as legal entities over the last few years, particularly in South Eas

September 15, 2010

Feature - Deciphering the tree of life Image courtesy of Miriam Boon. What’s a bee without its honey, a butterfly without a flower’s nectar? It’s a pretty puzzle posed by the fossil record, which suggests that insects evolved long before flowering plants did. With the rise of genetics, a new window has opened onto evolution – one that could provide a fresh perspective on old problems such as the disparity between insect and angiosperm (flowering plant) evolution. Computational phylogenetics is the development of computational and mathematical techniques that aid in the estimation of evolutionary history, using molecular data such as protein and DNA sequences to construct a “tree of life.” To calibrate their molecular results, evolutionary biologists add the fossil record to the mix, and assume that a new species will evolve at the same rate as its ancestors. “Often they [other researchers] would force the date of angiosperms to correspond

September 15, 2010

Feature - Neighborly efficiency: Scaling kNN problems  (nearly) linearly using GPUs

This image depicts an example of a two-dimensional feature space. In this case, the unknown dot would be classified as “green,” because three of the five nearest neighbors are green.
Image courtesy of Cyrus Stoller and Libin Sun.

What is k-Nearest Neighbor?
The class of machine learning algorithms known as kNN classifies objects based on the closest training examples in the feature space. What does that mean? Let us illustrate that with an example.
Imagine you have a large set of digital images of friends and family. Many of them have already been tagged according to the person in the image. You would like a way to automate the tagging of the remaining photographs.
First, you would need a program that can analyze images of faces, quantifying features such as eye color, the distance between the eyes, and so on. The program would take in each image – say, 1000 – and re

September 15, 2010

Feature - One person’s view, behind the scenes of middleware

EMI tries to pull together three different middlewares. Image courtesy EMI

John White, security team leader of EMI, discusses his work.iSGTW: Can you explain to me what EMI stands for?White: European Middleware Initiative. EMI unifies under one project the three middlewares that have been put together over the past six years: gLite, ARC and Unicore.
Each of these middleware do different things; for example, Unicore runs inside high-performance computing centers like a ‘monster’ supercomputer, while gLite typically runs over a distributed system, such as a farm of 1,000 or 2,000 batch nodes — which can be anything from a cluster of batch nodes or a group of white, ‘pizza box’ type home-PCs. We used to have these at CERN up until a few years ago.iSGTW: What is your role?White: I have a few roles; my most important is as security area leader of the security components of all three midd

September 8, 2010

Feature - BiG Grid’s big idea

Image courtesy BiG Grid

With the EGI Technical Forum coming up next week, iSGTW thought ths would be a good time to learn more about one of the Forum’s sponsors, BiG Grid.
Modern detectors, medical imaging instruments and micro-arrays produce huge volumes of data, far beyond the storage capacities of conventional computing — thus calling for ever-increasing enlargement of infrastructure.
To help solve this problem, the Netherlands-based BiG Grid project is turning to the grid as a place to combine data, analyze it and allow scientists to conduct research in a wide range of disciplines. BiG Grid is a collaborative effort between Nikhef (the National Institute for Sub-atomic Physics), NBIC (Netherlands Bioinformatics Center) and NCF (the National Computing Facilities foundation). “Our goal is to build and roll-out a nationwide, grid-based, e-science infrastructure,” said Arjen Van Rijn, chairman of the BiG Grid executiv

September 8, 2010

Feature - Inflated performance

Are some performance numbers full of hot air, like this balloon?
Image courtesy of Ben Brezina.

The more things change, the more they stay the same. Since the advent of scientific computing, researchers have sought out ways to get more research done in less time and for less money. Proponents of new technologies have, in turn, claimed improvements in technology that may seem too good to be true. How, then, is a researcher to tell the difference?
Back in 1991, Berkeley Lab scientist David Bailey published an entertaining tongue-in-cheek paper in the now-defunct Supercomputing Review entitled, “Twelve Ways to Fool the Masses When Giving Performance Results on Parallel Computers” (PDF). As with any paper discussing computer performance, there are some obvious aspects of the paper that are now out of date, such as the shift from megaflops to teraflops.
“Also, the focus of comparison at the time was vector supercomputers then being manuf

September 8, 2010

Profile - People behind the European Grid Initiative: Tiziana Ferrari

Image courtesy EGI

The European Grid Infrastructure’s role is to support research and collaboration across the continent by providing seamless, instant access to computing resources. But who has the job of making sure things actually work? Meet EGI’s Chief Operations Officer, Tiziana Ferrari. She spoke recently with iSGTW to tell us what is rewarding about her job, what is challenging and why it is important.
 
iSGTW: Describe what you do for EGI.Ferrari: I am responsible for coordinating the operations of the infrastructure across Europe. The user doesn’t care whether a resources is in Spain or in France, they just need it to work — that is my job. In EGI though, every country is responsible for its own operations.
But EGI needs to coordinate this and ensure that everyone is using the same protocols. That is my role. I need to make sure the production and accounting infrastruc

September 1, 2010

Feature - Joe Hellerstein on cloud programming

Image courtesy of UC-Berkeley.

Earlier this year, we discussed the possibilities raised by working with parallel programming languages in an interview with John Shalf. This week, iSGTW interviews Joe Hellerstein, the principle investigator for the Berkeley Orders Of Magnitude project, about cloud programming languages. Hellerstein is a professor of computer science at the University of California, Berkeley, where he focuses his work on data-centric systems and the ways in which they drive computing.
Why do we need a language for cloud programming, what makes a cloud programming language different from a parallel programming language, and how do cloud programming languages work? Read on for answers to these and other questions.
iSGTW: What is cloud programming?
Hellerstein: I think of “the cloud” as a new computing platform, the way that the PC or the mobile phone were new platforms when they were introduced. In general, a

September 1, 2010

Feature - People behind EGI: Steve Brewer steps in as the voice of the user

Image courtesy Steve Brewer

With the EGI technical forum coming up in a few weeks, readers may want to know more about the people behind the scenes.
Much of the organization’s success will hinge upon its ability to foster strong communities between users and resource providers. Who will act as a communication point between these two groups? Steve Brewer, a long-time member of the European Grid Community, has recently been appointed chief community officer for EGI.eu, the new organization responsible for coordinating the European Grid Infrastructure.
iSGTW: Where does EGI.eu fit into the grand scheme of things?
Brewer: From physicists, to chemists, to geologists, many European researchers need distributed computing for their work. While these researchers previously used an infrastructure coordinating by EGEE (the project Enabling Grids for E-sciencE), the same infrastructure is now coordinate

September 1, 2010

Feature - The forecast before the storm
How supercomputers and hybrid workflows helped beat tornadoes to the chase

A Doppler On Wheels collects data in a tornado during VORTEX2, as PI Nolan Atkins stands nearby collecting photogrammetric data.
Image courtesy of VORTEX2.

Chasing tornadoes won’t get you very far, if your goal is to understand how tornadoes form. To get results, researchers need to get their instruments on the ground before the tornado touches down.
That’s the big catch 22 of VORTEX2 (Verification of the Origins of Rotation in Tornadoes Experiment), according to principal investigator Joshua Wurman. Current techniques predict tornadoes an average of only 13 minutes in advance, a fact that makes it difficult to evacuate or properly prepare for the impending disaster. To improve that lead time, or learn how to predict how destructive a tornado will be, scientists need data recorded as the tornadoes form.
“In order for us to collect good data we had t

August 25, 2010

Feature - OSG and TeraGrid join forces for ExTENCI

Image courtesy of Jayanta Behera.

Last week marked the kick-off for ExTENCI, the first major technical collaboration between TeraGrid and Open Science Grid.
“The idea is to have Open Science Grid and TeraGrid work together on a joint project as an experiment,” explained Paul Avery, the principle investigator for ExTENCI (which stands for Extending Science Through Enhanced National CyberInfrastructure). Although the two infrastructures have worked together to agree on principles and attend each others’ meetings, ExTENCI marks the first time that they will work together, sharing milestones and goals.
In order to make the partnership possible, the National Science Foundation’s Office of Cyberinfrastructure and Math and Physics Directorate each awarded ExTENCI approximately one million U.S. dollars over two years, for a grand total of just over two million U.S. dollars.
The limitations imposed by a small budget

August 25, 2010

Opinion - Scientists, meet the citizens

Screenshot from the Foldit online game for protein folding. Image courtesy Foldit

François Grey is the coordinator of the Citizen Cyberscience Center.
In a week’s time, an unusual meeting of minds will occur in London.
Billed as a Citizen Cyberscience Summit, it will bring together scientists from a range of distributed, volunteer computing and volunteer thinking projects, to mingle with some of the volunteers who participate in these online projects.
The upshot of the event, hosted by King’s College London on 2-3 September, should be a stimulating dialogue about how to make citizen cyberscience even more compelling for the public and even more useful to science.
The timing of the event could not be better. August saw a bumper crop of major scientific results from online science projects involving public participation. An article in Nature described progress made in protein folding using an online multiplayer game called

August 18, 2010

Back to Basics - How hardware virtualization works: Part 4

BY GREG PFISTER Since retiring from his position as an IBM Distinguished Engineer, Greg Pfister has worked as an independent consultant as well as serving as research faculty at Colorado State University. Pfister is the author of “In Search of Clusters,” and over his 30-year career, he has accrued over 30 patents in parallel computing and computer communications.

It is possible to find many explanations of hardware virtualization on the Internet and, of course, in computer science courses. Nonetheless, there remains a great deal of confusion regarding this increasingly popular technology.
This is the last part of a multi-part series that attempts to provide an approachable explanation of hardware virtualization. To catch up, see Part 1, Part 2, and Part 3.
Drown It in Silicon
In the previous discussion I might have lead you to believe that paravirtualization is widely used in mainframes (IBM zSeries and clon

August 18, 2010

Feature - OSG Summer School a success

The summer school in progress. Image courtesy of OSG.

Last month, Open Science Grid hosted its first summer school, and by all reports, it was a hit with both teachers and students.
“The OSG Summer School was a great experience,” said Vishagan Ratnaswamy, who will begin work on his doctorate in aeronautics this fall at Caltech. “I was able to learn more about the script files I was using as well as the systems I was running my simulations on.”
Before attending the summer school at the University of Wisconsin-Madison, Ratnaswamy was already an OSG user. But by attending, he said, he learned more about the available computational resources and the many ways in which grid computing can be applied to various types of research.
Ratnaswamy was one of 17 students chosen from a pool of 45 applicants to receive funding to attend the summer school, the TeraGrid ’10 conference, and next year’s OSG All-Hands Meeting. At

August 11, 2010

Feature - Education and the future: eLearning

Image courtesy GridTalk

Computers and the web have transformed homes and businesses, and could do the same for education and training. Known as “eLearning,” this can be as simple as accessing a school timetable online, or as complex as running virtual communities for sharing and creating knowledge. eLearning is defined by the European Commission (EC) as ‘the use of new multimedia technologies and the internet to improve the quality of learning by facilitating access to resources and services, as well as remote exchanges and collaboration.’ The EC sees eLearning as an integral part of education and calls for member states to include eLearning in national policies; its Lifelong Learning Program, running from 2007-2013, includes it in schools, higher education, vocational training, and adult education.Where do grids fit in?
Grid technologies help researchers worldwide collaborate, analyze data and carry out research.

August 11, 2010

Feature - NetLogo: A low threshold, no ceiling language

Two fifth grade students use NetLogo to learn about electrical current. Image courtesy Pratim Sengupta.
Front page image: Tiling with squares whose sides are successive Fibonacci numbers in length. Courtesy Wikipedia under Creative Commons license.

Elementary school students may not be able to decipher mathematical models such as Maxwell’s Equations. But given the right visualization and computational modeling tools, they can learn the underlying concepts.
Meet NetLogo, a multi-agent programmable modeling environment authored in 1999 by Uri Wilensky, a learning sciences and computer science professor at Northwestern University, and founder of the Center for Connected Learning and Computer-Based Modeling.
Remember the turtle?
A generation of adults were introduced to functions and programming through Logo and the “turtle” - an on-screen triangular cursor - that accompanied it. Logo was first created in 1967

August 11, 2010

Feature - The sky’s the limit

Image courtesy Simon Langton Grammar School

Becky Parker, head of physics at the Simon Langton Grammar School in Kent, UK, is introducing her students to outer space. In 2007, Becky organized a trip to CERN for her 16 to 18 year-old students. There, they were introduced to the Timepix computer chip, a sensitive light-detector used for medical imaging. Back in Britain, one of her students came up with the idea of using the chips to measure cosmic radiation. Parker’s response: “Brilliant!” A Timepix chip has 65,536 pixels over a 2 cm² area. An event occurs when a particle strikes a pixel and is converted into an electrical signal, which can be measured. Her students wanted to use Timepix chips to detect particle type, energy and possibly, the directionality.Consequently, her students entered and won a space experiment competition with their design made from adapting readouts of the chip. Their instrument, called LUCID (Langton

August 4, 2010

Back to Basics - How hardware virtualization works: Part 3

BY GREG PFISTER Since retiring from his position as an IBM Distinguished Engineer, Greg Pfister has worked as an independent consultant as well as serving as research faculty at Colorado State University. Pfister is the author of “In Search of Clusters,” and over his 30-year career, he has accrued over 30 patents in parallel computing and computer communications.

It is possible to find many explanations of hardware virtualization on the Internet and, of course, in computer science courses. Nonetheless, there remains a great deal of confusion regarding this increasingly popular technology.
This multi-part series attempts to provide an approachable explanation of hardware virtualization. You can see Part 1 here and Part 2 here.
Translate, Trap and Map The basic trap and map technique described previously depends crucially on a hardware feature: the hardware must be able to trap on every instruction that could a