Now in its 19th year, the High Performance Computing and Communications Conference addresses some major issues in supercomputing, grid computing, computer security, nanotechnology, information warfare, HPC applications, cluster technology, disaster recovery, and next generation networks. This high level government conference is known for its content and networking, not for any commercial glitz or hype. With increased controversy over federal budgets and a heightened awareness of supercomputing developments worldwide, the funding and implementation of HPC efforts is becoming increasingly more important.
Steve Pawlowski, Intel Senior Fellow and the Architecture & Planning director and CTO of Intel's Digital Enterprise Group (DEG), will kick-off the conference, which is to be held April 12-14 in Newport, RI. Pawlowski is a 23 year Intel veteran and has been involved in all aspects of Intel-based product design.
Pawlowski says he was chosen to keynote the conference because Intel's goal is to drive the market using commodity processing. He believes that in doing so, Intel can bring supercomputing “to the masses.”
Intel, according to Pawlowski, is putting more emphasis now on the HPC community by discovering the building blocks that will allow for blue-collar clusters to drive solutions in this area.
In addition to learning about Intel's advancement in the field, Pawlowski says conference attendees will have the opportunity to network with some of the key leaders and thinkers in the HPC space. Pawlowski plans to address the future path of supercomputing, processor performance trends, the ways software will take advantage of increasing cores, where technology will be going as Moore's Law increases and how Intel will take advantage of that.
“Within the next 7-10 years, we could potentially see systems on people's desks that have the processing horsepower of what some of the lower-end HPC systems have today,” stated Pawlowski. “If you look at the trend, like with BlueGene, we could see a petaflop machine in the 2009 time frame and a 10 petaflop machine in the 2012 timeframe. Having that level of performance capability, using common building blocks to configure the machines, may change the equation in terms of how we build high performance systems,” said Pawlowski.
Dr. Bob Lucas, Director of the Computational Sciences Division of the University of Southern California's Information Sciences Institute (ISI), will also be involved in highlighting HPC issues at the conference. Lucas is the former head of the High Performance Computing Research Department in the National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory. He is also one of the developers behind Project Urban Resolve for the Joint Forces Command. At HPCC, JFCOM plans to put on a real time demo via the Defense Research and Engineering Network (DREN) at the conference.
Lucas became involved with HPCC because he strongly believed that modeling and simulation of military forces (FMS) has been a field that needed to expand onto large scalable clusters of commodity processors (SPP). He says that the U.S. Joint Forces Command (JFCOM) has taken a pioneering role in using SPP systems for FMS.
“As such, it is a new actor on the HPC stage, and HPCC is a wonderful forum for introducing JFCOM's work and its unique properties to the leadership of the broader HPC community,” added Lucas.
So what is Project Urban Resolve?
Lucas, the principal investigator for the project, says that a major focus of work at JFCOM is on operations in urban environments, where people and infrastructure make warfare much more complicated. Urban Resolve, a three-phase, multi-year military simulation experiment, will help JFCOM better plan warfare in urban battles.
Urban Resolve uses the Joint Semi-Automated Forces (JSAF) entity-level, intelligent agent code to simulate military vehicles, ClutterSim to simulate civilians, and other FMS software packages to simulate a city, allowing war fighters to plan and train for potential battle operations. High- performance computing has allowed the team to simulate entire cities with realistic civilian populations.
When networks of Linux PCs failed to provide enough memory and processing needed for the growing complexity of these simulated spaces, Lucas' team at USC's ISI aided in migrating the code over to Linux clusters. The team will also help develop software to manage the massive amount of data that is now generated.
Yet Lucas' project has larger significance. The effort will span the entire country. There are Blue and Red teams in different parts of Virginia. The “civilians” in the simulation are controlled from California. Ohio and Hawaii provide HPC systems to run the program. The Defense Research and Engineering Network (DREN) is crucial to Lucas' projects, and he envisions no problems running the demonstration at the conference.
The High Performance Computing and Communications Conference is indeed a critical event that helps to propel governmental efforts in HPC. How important is it for the government to take the reigns in adopting HPC initiatives? Dr. Bob Lucas believes it is critical — and that change is happening, but slowly.
“Where the government has not invested recently, is in research into the long- term future of HPC,” says Lucas. “Even the DARPA HPCS program, which is the most forward-looking of such efforts, is aimed at deploying systems by the end of the decade. Many studies, including a recent one by the National Research Council (NRC), one that I participated on, have advocated an increase in research in this field.”
And though the U.S. currently boasts two of the fastest supercomputing systems in the world, its lead in the field is by no means insurmountable. Conferences like HPCC are held to bolster a renewed sense of ambition and innovation.
“I think it's in our national interest to develop a new generation of machines which allow US scientists and engineers to maintain preeminence for those problems which do not perform well on clusters,” says Lucas. “DARPA's HPCS program is investing in such systems. I think we need to start investing in the applications that will use the HPCS systems when they become available. I also think we need to start investing in long-term research so as to have a pipeline of new ideas that can follow beyond the arrival of the HPCS systems in the 2010 time frame.”
HPCC boasts some big names, big ideas and will attack even bigger issues. Other well-known speakers include Thomas Barnett, sole proprietor of Barnett Consulting and founding partner of the New Rule Sets Project LLC, who will discuss “The Pentagon's New Map: War and Peace in the 21st Century” and DARPA's HPC program manager Robert Graybill, who will offer a productivity update on high productivity computing systems.
Catch Steve Pawlowski's keynote address at HPCC at 9am on April 12th and the JFCOM Urban Resolve demo at 11:45am on April 13th. Full conference details can be found at http://www.hpcc-usa.org/.
More on HPCwire's coverage of the event can be found here: http://news.tgc.com/msgget.jsp?mid=350692&xsl=story.xsl