Narrator: The IBM Blue Gene program is dedicated to building a new family of supercomputers optimized for computational performance, scalability, and the efficient usage of power and floor space. The high performance of Blue Gene enables a tremendous increase in the scale of simulations beyond what is possible with other supercomputers. Blue Gene users today are dramatically accelerating their work and gaining new insight into their complex models. In this brief presentation, you will learn about Blue Gene and how it addresses the challenges in high performance computing. You will learn about Blue Gene’s strengths. And you will hear how a Blue Gene customer is benefiting from using this remarkable technology. Scientists and researchers face ever-growing challenges as they work to solve their most important and computationally intensive problems. Blue Gene was designed precisely to deal with these challenges in thoughtful, innovative, and efficient ways.
Dave Turek: High performance computing or supercomputing, as many people refer to it is a computational set of techniques that are pervasive in the economy today more so then ever before. From a historical perspective of course, this was always perceived to be something reasonably esoteric and limited in its scope and capability. The kinds of computational techniques that were associated with government laboratories or university kinds of research. But today courtesy of the advances over the last ten years or so with respect to evolution of hardware and software and more then anything price improvements of these kinds of technologies, the capabilities that we’re talking about today are accessible to a broad, broad array of users and institutes more so then ever before. So the pervasiveness of this technology is now something found in a variety of different domains within the economy. It ranges from what you see happening today in animation, industrial design, clearly scientific research, drug design et cetera. In all of these deployments of this kind of capability fundamentally result in strategic advantage accruing to those kinds of institutions and corporations that make use of this technology, faster time to solution more efficient solutions and better solutions all contribute to overall competitiveness of the firm, and this is something that as it gets deployed more broadly throughout a national economy contributes to international competitiveness as well.
Narrator: The Blue Gene program began in earnest in 2000 when IBM and the Department of Energy entered into a partnership to fund research and development to explore designs for highly cost-effective supercomputers. The IBM Research Division has led the way on innovating solutions to the hardware, software, packaging and application challenges associated with the development of a supercomputer with the capabilities of Blue Gene.
Dr. Tilak Agerwala: Over the last 25 years IBM Research has established an incredible record of innovation in computer systems ranging from high performance reduced instruction set computers to the chess champion Deep Blue and now Blue Gene. According to the president's information technology advisory committee computational science, which is really the use of advanced computing capabilities to understand and solve complex problems, has become critical to scientific leadership, economic competitiveness, and national security. IBM has long understood that ultrascale computing would be the path towards solving the grand challenge problems facing engineers, scientists, and researchers. Furthermore, we recognized that significant innovation would be required to overcome the barriers to building a supercomputer capable of achieving a quadrillion operations per second. Anticipating the need within the scientific community for new levels of computational capability, IBM designed and developed Blue Gene to deliver unsurpassed performance, and ultrascale computing in an environmentally efficient and affordable package. Blue Gene has already enabled our customers and I might add, our own researchers, to conduct breakthrough science. And I am confident that innovative products like Blue Gene will set the pace of leadership in high-performance computing for years to come.
Narrator: The Blue Gene system is built out of a very large number of compute nodes each with a modest clock speed, and optimized for low power consumption. Blue Gene utilizes IBM PowerPC embedded processors and system-on-a-chip techniques that allow for advanced integration of compute processors, memory and high-speed interconnection. The basic Blue Gene building block is a compute card that contains two dual processor nodes. Sixteen compute cards are mounted on a node board, and 32 node boards complete a one thousand twenty four node rack. The nodes are interconnected by several networks including a 3-dimensional torus network for point-to-point messaging between compute nodes and a network for accelerating collective operations over an entire application. Blue Gene software takes advantage of hardware features to deliver high performance without compromising stability and security. The programming and administration environments are based on familiar programming languages and tools, allowing users to benefit from the innovative design elements of Blue Gene without facing a step learning curve. Innovative design allows Blue Gene to be scaled up to more then 64,000 compute nodes yielding enormous levels of performance. Experiences with Blue Gene have shown it to be applicable to a number of parallel workloads found across a variety of disciplines. Today, Blue Gene is enabling breakthrough science in quantum chemistry, life sciences, financial modeling, hydrodynamics, molecular dynamics, astronomy and space research, materials science and climate modeling.
Dr. Kirk Jordan: Blue Gene is enabling breakthrough science. Customers are using Blue Gene to solve problems that were previously unsolvable. They are shortening the time to solution by orders of magnitude. And Blue Gene is allowing our customers to think about new ways to address scientific and engineering challenges. Scientists have been using Blue Gene to simulate earthquakes and assess the hazards they create. With its ultrascaleability, Blue Gene is well suited for this type of problem because scientists can increase the resolution of the earthquake waves to understand fine detail. Understanding earthquake dynamics will allow designers of roads, bridges and buildings to develop safer structures. The recent rash of hurricanes and other weather-related disasters have heightened peoples’ interest in the world’s climate and how it might be changing. To fully understand the dynamics, climate researchers must combine knowledge of the atmosphere with that of the oceans. Such complex models require the help of the most advanced supercomputers. Blue Gene has been used by atmospheric research centers to model atmospheric and ocean interactions to produce increasingly accurate portraits of the earth’s climate. Extracting oil and gas from underground reservoirs can be costly. Petroleum companies want to ensure they retrieve as much high-quality product as efficiently as possible. Supercomputers have been used for many years to model oil reservoirs. But with Blue Gene we are investigating for the first time combining reservoir-modeling techniques with those of seismic processing for better and more profitable reservoir management. In the future, Blue Gene promises to be an important tool for advancing the capabilities of researchers to achieve important breakthroughs in health care, life sciences, petroleum exploration and a host of other applications areas.
Narrator: Customers of Blue Gene have derived significant value from Blue Gene for solving their most challenging problems, and for providing insight into the underlying science. Let's hear from Boston University.
Dr. Claudio Rebbi: My own field of research is Particle Therory. The therory of the constituents of the nucleus, of the atomic nucleus. Most specifically I started a way in which quarks are bound into protons and neutrons which are the constituents of the Atomic Nucleus. Here at Boson University I also direct a Center for Computational Science. Now the Center for the Computation of Science has the mission of overseeing research and education that involves computers. And so beyond being involved personally in research I oversee a broad spectrum of research. I am certain as I was while considering the aquisiton of Blue Gene, I was Certain that this Machine would have a wide range of applications. With Blue Gene a way in which has been able to pack, you know the design of the machine, has been able to pack so much electronics in a small space using a very wise approach, ok, of not pushing the cycle but rather working on a good balance between CPU and Memory we are getting back to the implemention of the paradigm that has produced wonderful results. So for us it has been quite enthusing to see that once again we at Boston University are at the forefront of massive parallel computing. Again, there are calculations we are doing today that would just not have been possible to do a few years ago. First of all the moment at which we made the aquisition of the Blue Gene was by far the most powerful machine available. The most Powerful, the most promising, the most innovative .. its architecture. Second during the last few years we have strengthened our relation with IBM. It has become a true partnership which has been very satisfactory for us and you treasure this relation and this was another deciding factor for how we would go. So, you know, leveraging our good relations with IBM we were able to position ourselves to be one of the very first institutions to have the Blue Gene. And this is so far how the story goes.
Narrator: Let's hear from ASTRON in the Netherlands.
Dr. Eugene de Geus: The key challenge that we have with LOFAR is to be able to calculate very massive amounts of data at a very high speed. We're looking at a compute challenge of about - of the order of - thirty teraflops while at the same time we stream at the order of terabits per second through the system and that is the combined challenge that we face for LOFAR. LOFAR is an astronomical telescope. It’s a radio telescope. It is going to be one of the largest radio telescopes in the world. It’s a system combining the signals from about twenty thousand antennas spread over an area of about 350 kilometers in diameter. and porting all that data into the central processor. What brought us in contact with IBM was a broader search trying to find companies that were interested in this challenge. And fortunately IBM responded and we got in touch with the Research Centers at IBM and that's when things really started to happen. The value with Blue Gene I think comes from three points. The first one was when we really started to discuss the possibility of using Blue Gene as the central processor for the radio telescope was that we got access to some of the very impressive intellectual powers at IBM through the research labs that we got in touch with. And then, of course, the second point was that the Blue Gene was able to offer - in effect in a very small system and very energy efficient system - very large amounts of compute power which are needed just to operate LOFAR. And, the third point really is that in combination with that compute power we were able to get the streaming capabilities of Blue Gene which means that we can actually do very high data intensive computing and that combination for us really was very powerful and allowed us to get this central processing system for the LOFAR system in a very short time in fact. We are now discussing with IBM the next steps that should lead us, from our perspective, into the peta scale level of computing, and having access to the intellectual powers at IBM really makes this a very interesting challenge for us and a very interesting road forward.
Narrator: Let's hear from Forschungszentrum Julich:
Dr. Bernd Mohr: The John Von Neumann Institute for Computing is the computing center for Research Center Juelich, Germany’s largest national lab for that we have to provide computing power for all the researchers in the lab, but also the academic users in Germany and more and more in Europe. All the users who have access to this machine so far are totally excited about the machine. We can do now simulations that we couldn’t do before. We can go to a much finer resolutions; we see effects we couldn’t see before because resolution was not high enough; and we just are getting computational power to finish calculations much much faster. We were surprised that it was possible to get a wide range of applications running on Blue Gene. So we have meanwhile applications from material science, elementary particle physics,computational biology, computational astrophysics, computational geophysics, computational chemistry, medical applications. Almost everything is running nicely on Blue Gene. Blue Gene was a solution ...it was the only machine that really scales to a level which is needed for our purposes. But On the other hand it is also space efficient and power efficient in the same way. The price-performance is just unbeatable.
Narrator: IBM Blue Gene Enables Breakthrough Science. It provides massive, near-linear scalability and offers leading performance and price-performance. Blue Gene is environmentally efficient, yielding significant savings in power and floor space utilization. And Blue Gene is applicable to a growing set of computationally intensive workloads. This is what Our Customers have to say
- Blue Gene "provides insight"
- They are addressing "previously intractable problems"
- Blue Gene allows problems to be studied at a "quantum leap in the level of detail"
- Blue Gene "dramatically accelerates our work"
- And, Blue Gene has encouraged customers to "think even bigger" in their research
Thank you for your time.
| 
