insideHPC » Computing Research

Drug Discovery through Molecular Matchmaking at TACC

Rich Brueckner — Thu, 24 May 2012 18:28:50 +0000

Researchers are using TACC supercomputers to target a small group of molecules for therapeutic use, saving significant time and money. As described in the February 2012 edition of the Journal of Structural Biology, the process uses information from electron microscopy and knowledge about drug binding to create 3D models and to simulate the interactions of drug molecules with the target site.

More and more, they’re moving into the computational drug screening arena, and more and more it’s teams of people working together,” said Chandrajit Bajaj, professor of computer science at The University of Texas at Austin. “The biophysicist, the biochemist and the synthetic chemist are sitting together with the computational expert, and they say it’s giving them clues as to what they should be doing next.”

Read the Full Story.

3D Supernova Simulations on Jaguar

Rich Brueckner — Thu, 24 May 2012 15:43:11 +0000

In an effort to measure the size of the universe, researchers are using the Jaguar supercomputer at ORNL to simulate the explosions of white dwarf star supernova.

The physics of supernova explosions is something astrophysicists have been trying to figure out for about 50 years now,” said Stan Woosley, principal investigator of an Innovative and Novel Computational Impact on Theory and Experiment (INCITE) project and professor of astrophysics at the University of California–Santa Cruz. “It is an interesting physics problem in turbulent combustion, but it is also important because—as the 2011 Nobel Prize attested—Type Ia supernovae can be used to show that the expansion of the universe is accelerating.” This is because the supernovas function as “standard candles,” brilliant lights of known properties usable as measuring tools because their distance can be inferred by how bright they appear.

Because of the size of the computational runs, the studies are decoupled and done in three successive stages—ignition, explosion, and supernova. The multiyear project was allocated 50 million computing hours on Jaguar in 2011 and 47 million in 2012 through the INCITE program, which is jointly managed by the US Department of Energy’s Leadership Computing Facilities at Argonne and Oak Ridge National Laboratories. Read the Full Story.

Massively Parallel Simulation Counters Antibiotic Resistance

Staff — Thu, 17 May 2012 17:34:42 +0000

A solution to the problem of antibiotic-resistant bacteria may be closer, as a result of a computer science project to investigate ways of making legacy software run efficiently on heterogeneous systems, the Nvidia GPU Technology Conference was told on 16 May.

Simon McIntosh-Smith, of the University of Bristol in the UK, presented results from a project that ported a very large molecular modelling program to systems with a mix of conventional CPUs and also GPUs. One early result has been the identification of around ten small molecules that could block the biological pathway that creates antibiotic resistance in bacteria. So confident are the researchers in their predictions that the candidate drug molecules are now being synthesised in the laboratory.

Typically, biomolecules such as proteins can be made up from between a thousand and two thousand atoms, whereas the docking molecules, the ligands, will be an order of magnitude smaller – about 50 to 100 atoms. The software is BUDE – the Bristol University Docking Engine – is used to predict the structure of small molecules that can bind tightly to the active sites in large biological molecules. It processes tens of millions of candidate ligands and uses a genetic algorithm-like methodology to select those that bind most tightly, using energy minimisation calculations. It is, said McIntosh-Smith, a very large piece of code, in the region of hundreds of thousands of lines of Fortran. However, only a few thousand lines needed to be ported across to GPU processors.

Because the ten million or so ligands all come in slightly different flavours – they have flexible side chains, for example – the project represents “an embarrassment of parallelism,” he continued. “When we get down to one molecule, we want to test it in many different positions and rotations, so we have even more parallelism.”

Only two results are outside the desirable bounds, so ‘we are now getting very accurate simulation results,’ McIntosh-Smith said. By porting the code across to a heterogeneous CPU+GPU system, he reported a factor of 20 increase in the speed of computation and a factor of 10 improvement in energy efficiency.

The simulation to try to find ligands that would block recently emerged antibiotic-resistance in a bacterium found in Asia took four days and surveyed more than 8 million candidate molecules. Once, he pointed out, it would have taken the biggest and fastest supercomputers in the world to do the calculation, which can now be done by a university research group.

This story originally appeared on HPC Projects. It appears here as part of a cross-publishing agreement with Scientific Computing World.

Supercomputing Materials to Harvest the Sun

Rich Brueckner — Sun, 13 May 2012 11:00:37 +0000

Researchers are using the TACC Ranger supercomputer to investigate photosynthetic materials with their sights on creating a light-harvesting material that can turn sunlight into chemical energy.

By using computation, we can understand the properties and the behavior of this molecule and gain insight into improving it,” said Margaret Cheung, Assistant Professor of Physics at the University of Houston. “If we can capture the mechanism that converts solar energy into chemical fuel, it opens the door to many opportunities.”

To enable her research, Cheung uses Ranger to explore the role that confinement, temperature, and solvents play in the stability and energy efficiency of the light-harvesting triad. Her results provide a way to test, tailor, and engineer nano-capsules with embedded triads that, when combined in large numbers, could greatly increase the ability to produce clean energy. Read the Full Story.

Slidecast: Bina – Accelerating Genomics & Data-Driven Healthcare

Rich Brueckner — Tue, 01 May 2012 17:22:37 +0000

www.youtube.com/watch?v=nPF2ZJDrS-c

In this slidecast, Narges Bani Asadi from a new Startup called Bina presents: Bina - Accelerating Data-Driven Healthcare.

In 2008, a renowned team of researchers at Stanford and UC Berkeley came together to solve a critical problem for cancer researchers who were severely constrained by their existing computing capabilities for large-scale data analysis. Through this work they created a new approach to the integrated co-development of statistical algorithms, software and hardware for high performance data analysis. The result was a dramatic improvement in the accuracy, computational efficiency and cost of data analysis. The team founded Bina Technologies in 2010 to bring this approach to market.”

Download the MP3 * Subscribe on iTunes * If Dropbox is blocked, download from this Google page.

Simulating the Next Generation of Fuel-Efficient Engines

Rich Brueckner — Tue, 01 May 2012 15:51:29 +0000

Marianne Lavelle over at National Geographic News writes that supercomputers like Jaguar at ORNL are helping researchers design the next generation of fuel-efficient engines.

We don’t understand the coupling of turbulent mixing and ignition chemistry in fine enough detail to help us impact the design,” said Jacqueline Chen of Sandia National Laboratories. “You need to get the correct burn rate, or you get a very noisy engine.” In other words, either the fuel mixture needs to be adjusted, or the temperatures in different portions of the mixture need to be stratified, or layered, to control the speed at which pressure rises in the engine.”

Read the Full Story.

ANI Testbed a 100-Gigbit Highway for Science

Rich Brueckner — Tue, 01 May 2012 12:00:19 +0000

Linda Vu from Lawrence Berkeley Labs writes that the Advanced Networking Initiative (ANI) has created a 100 Gbps national prototype network and a wide-area network testbed that is changing how researchers think about moving Big Data.

It took us approximately 30 minutes to move 35 terabytes of climate data over a wide-area 100 Gbps network. This is a great accomplishment,” said Mehmet Balman of Berkeley Lab’s Scientific Data Management group. “On a 10 Gbps network, it would have taken five hours to move this much data across the country.”

Read the Full Story.

Research HPC Podcast Examines Near Threshold Voltage for Exascale

Rich Brueckner — Mon, 30 Apr 2012 21:59:02 +0000

Mike Bernhardt from The Exascale Report has released the second installment of his Research HPC podcast series featuring key voices from the Intel Labs think tank.

In this podcast, Mike discusses Near Threshold Voltage processing with Vivek De, an Intel Fellow and Director of Circuit Technology Research at Intel Labs. NTV processing is a research area that holds tremendous promise for more efficient power management, and is applicable to numerous future computing applications ranging from mobile applications to HPC to exascale systems.

Download the MP3 * If Dropbox is blocked, you can download from this Google page.

New Paper: An Early Evaluation of the Scalability of Graph Algorithms on the Intel MIC Architecture

Rich Brueckner — Sun, 29 Apr 2012 12:00:52 +0000

Erik Saule and Umit V. Catalytirek from Ohio State University have published a new paper examining the scalability of the upcoming Intel MIC architecture on Graph algorithms.

Abstract
Graph algorithms are notorious for not getting good speedup on parallel architectures. These algorithms tend to suffer from irregular dependencies and a high synchronization cost that prevent an efﬁcient execution on distributed memory machines. Hence such algorithms are mostly parallelized on shared memory machines. However, current commodity shared memory machines do not typically offer enough parallelism to process these problems. In this paper, we are presenting an early investigation of the scalability of such algorithms on Intel’s upcoming Many Integrated Core (Intel MIC) architecture which, when it will be released in 2012, is expected to provide more than 50 physical cores with SMT capability. The Intel MIC architecture can be programmed through many programming models, here we investigate the three most popular of these models namely OpenMP, Cilk Plus and Intel’s TBB. We present scalability results of a parallel graph coloring algorithm, three variations of a breadth-ﬁrst search algorithm and a microbenchmark for irregular computations using these three programming models. Our results on a prototype board show that the multi-threaded architecture of Intel MIC can be effectively used for hiding latencies

Download the paper (PDF).

Better Batteries through Simulation

Rich Brueckner — Fri, 27 Apr 2012 15:57:02 +0000

With all the HPC conference I attend, the prospect of a better battery for my laptop and cameras is something I would really appreciate. Now, MIT researchers are using TACC’s Ranger supercomputer to investigate new material for high-density energy storage.

Most of our world is concerned with the organic chemistry of sustaining our lives. But most of the Earth – in fact most of the universe – is made up of inorganic materials formed by geological or cosmological processes. Despite their variety, all inorganic materials are composed of a not-so-terribly-large number of inorganic compounds: 50,000 to 200,000 depending on how you count. People have been studying these materials for millennium, but less than one percent of these have had their properties explored. Gerbrand Ceder, professor of materials science and engineering at the Massachusetts Institute of Technology (MIT), aims to change that. You don’t want to calculate the elastic quotient of 50,000 materials in your head, but it’s not impossible for the world’s most powerful supercomputers, Ceder says. He estimates that the Ranger supercomputer at the Texas Advanced Computing Center could calculate a given property for all known compounds in 80 hours, and Ranger is just one of a pantheon of powerful systems in the U.S.

Read the Full Story.

Cray to Sell Interconnect Hardware Assets to Intel

Rich Brueckner — Tue, 24 Apr 2012 20:52:20 +0000

Today Cray today announced it signed a definitive agreement to sell its interconnect hardware development program and related intellectual property to Intel Corporation for $140 million in cash.

This agreement is evidence of the leadership position we’ve established in high performance computing, and is an exciting win for our customers, our company and our shareholders,” said Peter Ungaro, president and CEO of Cray. “By broadening our relationship with Intel, we are positioned to further penetrate the HPC market and expand on our industry-leading technologies in support of our Adaptive Supercomputing vision. Our product roadmap remains intact as we continue to build the highly differentiated, tightly integrated supercomputers that our customers have come to expect from Cray. This agreement also dramatically strengthens our balance sheet and increases our options for further growth, profitability and creating shareholder value.”

Under the agreement, Cray will continue to develop, sell and support current product lines, as well as the Company’s next-generation supercomputer code-named “Cascade.” The transaction is expected to close relatively quickly, and as many as 74 Cray employees will join Intel. Read the Full Story.

Argonne Supercomputer Run to Help Explain Antimatter Mass in the Universe

Rich Brueckner — Mon, 23 Apr 2012 19:10:26 +0000

An international team of scientists has, for the first time, simulated the decay process of a kaon into two pions with extreme precision. The calculation took 54 million processor hours on the IBM BlueGene/P supercomputer at the Argonne National Lab and will help researchers explain why there is more matter than antimatter in the universe.

In the simulation, a technique known as lattice quantum chromodyamics (QCD) is used to carry out the computation. The parameters of the decay are input into a computer as a finite grid or lattice of space–time points. “Using lattice QCD was tricky, as the lattice box has a finite size and this means that the quarks cannot separate infinitely,” says Sachrajda. He goes on to explain that the process the researchers considered involved the kaon decaying into two mesons with isospin 2 (a quantum number related to the strong interaction). “This isospin has a real and imaginary part – the real part has been predicted and experimentally verified, and our value was in good agreement with that. The imaginary part, on the other hand, is not known from experiment. This is the first time it has been experimentally determined,” explains Sachrajda.

Read the Full Story.

Cycle Computing Spins up 50K Cores on AWS for Viral Research

Rich Brueckner — Fri, 20 Apr 2012 15:30:38 +0000

Jason Stowe and the crew from Cycle Computing has upped the Ante once again this week with the announcement that company successfully provisioned a 50,000-core utility supercomputer in the AWS cloud for Schrödinger and Nimbus Discovery.

We ran Naga across each of the 7 regions that AWS currently supports, scaling-out all supporting systems of a cluster (scheduling, software configuration, etc.) so we could use idle capacity wherever we found it. However, the magic really happened when we layered CycleServer on top of Naga and allowed our revolutionary new job submission algorithm to intelligently dole out work to each region based on real-time measurements from that region. Using this architecture, we had built ourselves a secured, automated 50,000 core supercomputer in under two hours using AWS infrastructure.

Read about how they did it at the Cycle Computing Blog.

Human Brain to be Built Using Supercomputer

Ralph — Mon, 16 Apr 2012 15:53:44 +0000

The Hindu reports that scientists are trying to build human brain using the world’s most powerful computer. At least 12 years from completion, the machine will reportedly try to combine all the information so far uncovered about the brain’s mysterious workings — and replicate them on a screen, right down to the level of individual cells and molecules.

The complexity of the brain, with its billions of interconnected neurons, makes it hard for neuroscientists to truly understand how it works,” said project leader Professor Henry Markram. “Simulating it will make it much easier — allowing them to manipulate and measure any aspect of the brain,” he said. Housed at a facility in Dusseldorf in Germany, the ‘brain’ will feature thousands of three-dimensional images built around a semi-circular ‘cockpit’ so scientists can virtually ‘fly’ around different areas and watch how they communicate with each other.

The project has received some funding from the EU and has been shortlisted for a 1 billion euro EU grant, which will be decided next month. Read the Full Story.

CURIE Super Simulates the Full Observable Universe

Rich Brueckner — Sun, 15 Apr 2012 16:42:46 +0000

A team of researchers from the Laboratoire Univers et Théorie has performed the first-ever computer model simulation of the structuring of the entire observable universe, from the Big Bang to the present day. Using GENCI’s new supercomputer CURIE, this unprecedented simulation followed the evolution of 550 billion particles, which will shed light on the nature of dark energy and its effects on cosmic structure formation, and hence on the distribution of dark matter and galaxies in the universe.

The implementation of this exceptional project would not have been possible without the powerful resources made available to the researchers by the “Grand Equipement National de Calcul Intensif ” (GENCI), whose new supercomputer CURIE is equipped with more than 92,000 CPUs and can perform 2 Petaflops. The CURIE supercomputer is housed and operated by the CEA at the ” Très Grand Centre de Calcul, at Bruyères-le-Châtel (Essonne). Designed by Bull, it is one of the world’s five most powerful supercomputers.

Read the Full Story.

Building a Human – The 200-terabyte Construction Manual in the Cloud

Rich Brueckner — Thu, 12 Apr 2012 13:38:51 +0000

Caleb Garling over at Wired Magazine writes that Amazon is taking genomics to the Cloud in a big way:

What do you do with a 200-terabyte instruction manual that tells you how to build a human? You put it on a cloud. That’s what Amazon and the National Institute of Health (NIH) have done with the 1000Genomes project, using Amazon’s S3 storage service to offer over 1,700 human genomes to genetics researchers across the globe. “This is what allows us to drive more complex maps of how genes interact with each other and their environment and zoom in on areas that may have a role to play in human health and disease,” says Matt Wood, who oversees Amazon’s side of the project and holds a PhD in bioinformatics. “This is the seed to create a tree of data.”

Read the Full Story.

The First Contract for Exascale? IBM to Build Super for SKA Telescope Array

Rich Brueckner — Mon, 02 Apr 2012 16:46:05 +0000

Today, IBM and the ASTRON Netherlands Institute for Radio Astronomy announced an initial 32.9 million EURO, five-year collaboration to research extremely fast, but low-power exascale computer systems targeted for the international Square Kilometre Array (SKA). The SKA is an international consortium to build the world’s largest and most sensitive radio telescope. Scientists estimate that the processing power required to operate the telescope will be equal to several millions of today’s fastest computers.

www.youtube.com/watch?v=dvSnPhxe-8U

ASTRON is one of the leading scientific partners in the international consortium that is developing the SKA. Upon completion in 2024, the telescope will be used to explore evolving galaxies, dark matter and even the very origins of the universe—dating back more than 13 billion years. Read the Full Story.

Nvidia Helps Power Bid For 2015 X-Prize Moon Mission

Rich Brueckner — Tue, 27 Mar 2012 16:47:08 +0000

Today Nvidia announced that the company’s Tesla GPUs are being used by a team of German scientists participating in a global competition to land a robotic rover on the moon by 2015.

NVIDIA GPUs will be instrumental in helping us land the Asimov rover safely and allowing us to calculate a wealth of detailed information to enhance our understanding about the lunar surface,” said Robert Böhme, team leader of the PTS team. “At the same time, we will demonstrate the amazing scientific accomplishments that are possible with modern, high-performance GPU technology.”

Developed to foster a new era of lunar exploration, the Google Lunar X PRIZE offers the largest international incentive prize in history. A total of $30 million will be awarded to the first privately funded teams that safely land a rover on the surface of the moon, drive the rover 500 meters over the lunar surface, and transmit detailed video, images and data back to Earth for further study. Read the Full Story.

TACC Supercomputers Map Tree of Life

Rich Brueckner — Sat, 24 Mar 2012 17:06:09 +0000

While the Tree of Life film bares the rare distinction of being the only movie to put me to sleep three times in one sitting, the study of evolution through Phylogenetics is a very lively topic these days thanks to the power of HPC.

Researchers are using now TACC supercomputers to develop and test smarter, faster, and more accurate genetic alignment and tree-building algorithms and apply them to some of the largest biological datasets ever created.

The most accurate trees are estimated using methods that try to solve hard optimization problems,” said Tandy Warnow, professor of computer science at The University of Texas at Austin and a Guggenheim Fellow. “While those solutions can be done on small datasets or moderate sized data sets, on large datasets, they can take a very long time — weeks to months to years of computational time. The Texas Advanced Computing Center ends up being essential for those problems.”

Collaborating with evolutionary biologists, the software they developed is leading to new insights into the Tree of Life and the relationships among species.

Read the Full Story.

Video: Intel Exascale labs: 4 Collaborative Competence Centers for Enabling Software Co-design

Rich Brueckner — Thu, 15 Mar 2012 10:46:17 +0000

www.youtube.com/watch?v=OW9rLx7EMoY

In this video, Marie-Christine Sawley presents: Intel Exascale labs: 4 collaborative competence centers for enabling software co-design. Download the slides (PDF).

Recorded at the HPC Advisory Council Switzerland Workshop on March 13, 2012.