Since the PITAC report titled Computational Science: Ensuring America’s Competitiveness was released in June, much discussion has taken place regarding the conclusions and recommendations contained in it. On July 1st, HPCwire published a commentary on the report by Dr. Robert Panoff, the founder and Executive Director of The Shodor Education Foundation, Inc.. Included in Dr. Panoff’s analysis was the suggestion that the report authors

… missed the chance to make their case for a broader impact of computational science: the most compelling science challenges that face us -challenges that do, in fact, justify a national effort at the large end of the spectrum- were relegated to the appendices in the report. Their main recommendation is to sustain software centers, not science. As Stan Lee would say, “‘Nuf said,”

Well, it looks as if some of the contributors to the report took exception to Dr. Panoff’s opinion. On July 15th, Dan Reed, Jack Dongarra, Chris Johnson, and Ken Kennedy issued a response to Dr. Panoff in HPCwire. In their response, Reed et al. clarify their point by stating that the main recommendation of the report calls for a

… balanced, integrated, long-term program that addresses all aspects of computational science.

Regardless of the primary recommendation, the bottom line is that any enhanced role of computational science in the United States will depend on greater emphasis by the federal government as well as clearer vision and understanding by all of the players, academia included, of how computational science fits into the overall research agenda.

The tragedy that befell the shuttle Columbia in 2003 resulted in numerous changes within NASA’s shuttle program. One of these changes, as outlined at SiliconValley.com, is greater intra-agency collaboration within the shuttle program itself. Experts at NASA’s Ames Center in California are now involved in certain aspects of the shuttle program, once reserved for NASA’s Johnson Space Center in Texas and the Kennedy Space Center in Florida. Taking advantage of it’s institutional-wide expertise, NASA now involves all of its centers in shuttle processes. The Ames Center has specifically been called upon to run simulations using it’s supercomputer (the third fastest in the world based on June’s Top500 list), appropriately named Columbia, to test scenarios involving the thermal tiles and other components. Personnel at Ames will also be on call to use the Columbia supercomputer to run immediate simulations during missions to find the best solution to specific problems.

NASA leveraging its overall talent, expertise, and multiple resources for a common goal represents an approach that should be utilized in greater frequency, not only within other agencies, but between separate organizations as well (such as academia and government). Such a paradigm would, in many ways, accelerate the nation’s cyberinfrastructure efforts.

Thanks to SiliconValley.com for the article.

As a partial acknowledgment of and natural extension to the recent academic recommendations for more multidisciplinary collaboration in higher education research, John Marburger recently submitted a memo to the Heads of Executive Departments and Agencies calling for greater unification of funding efforts to address project duplication among other things. From the optimists perspective, the Office of Science and Technology also wants to place greater emphasis on high performance computing efforts. Quoting from the memo, Marburger says

While the importance of each of the Networking and Information Technology R&D (NITRD) program areas continues, investments in high-end computing and cyber infrastructure R&D should be given higher relative priority due to their potential for broad impact.

The memo goes on to state

Advanced networking research (including test-beds) on hardware and software for secure, reliable, distributed computing environments and tools that provide the communication, analysis and sharing of very large amounts of information will accelerate discovery and enable new technological advances.

This is certainly a step in the right direction for making high-end computing a more salient issue on the federal research agenda. Thanks to Government Computing News for their coverage of the memo.

An extensive, ongoing discussion on Slashdot regarding the merits of Itanium. You’ll have to turn on your pure drivel filter, and the discussion often focuses on the consumer aspects. But there’s a lot there, and it was provoked by the status of Itanium on the Top 500 list, so it might be worth a look.

Interested in scientific visualization and wonder what some of the more salient research issues are in computational visualization? Chris Johnson, director of the Scientific Computing and Imaging Institute at the University of Utah, recently published his choice of the top 15 research problems facing this relatively new area of computational science. Although rather brief, Johnson’s paper provides a good summary of all 15 issues/problems.

An html version of the paper is available as well as a pdf version.

It’s no surprise that scientific visualization of the computational variety represents a critical area for high performance computing. The old adage that a picture is worth a thousand words could be no more accurate here. With the current ability to quickly perform astronomical size equations that result in tera- and petabytes of data, producing high resolution visualizations of data has tremendous value to “real-world” applications such as medical imaging and simulation and modeling of weather forecasts (such as mentioned in the previous post).

It seems only fitting as hurricane season is well underway that some news about weather forecasting get the spotlight. The Pittsburgh Supercomputing Center, in a multiple partnership led by NOAA, successfully demonstrated never before achieved storm forecasting by producing higher resolution results than currently used forecasting models are capable of. Over a three month period from April to June, PSC utilized a new forecasting model on its Terascale Computing System to generate three forecasts a day over an area of the Great Plains in the midwest. According to Kelvin Droegemeier, director of the Center for Analysis and Prediction of Storms at the University of Oklahoma and one of the partners of the effort,

Results from the spring experiment suggest that the atmosphere may be fundamentally more predictable at the scale of individual storms and especially organized storm systems than previously thought. Real time daily forecasts over such a large area and with such high spatial resolution have never been attempted before.

It’s good to see a parallel software success story as is the case with this new weather modeling system called the Weather Research and Forecasting (WRF) Model, especially given the fact that high performance software development lags well behind the much publicized advances in computing power.

The full story can be found on PSC’s website here.