2011 NCSA Private Sector Program Annual Meeting

What does music have to do with HPC, and what can it teach us? We'll present an unusual approach to highly-parallel, continuous, real-time, fine-timescale user interactions by using a violin to drive a simulation running on a high performance computing system.

What is "virtual realism" and why should we strive for it? We'll discuss how exponential improvements in resolution, scope, uncertainty, validation and verification, and other aspects of modeling and simulation are critical to driving industrial competitiveness in the 21st century.

The mathematical underpinnings of current numerical methods date to the era of scalar machines and codes. But now, massively parallel HPC platforms are challenging mathematical thinking to leap forward. We'll discuss this unique challenge, and the issue of scale in multiphysics environments.

Sequencing a patient's genome to help clinicians make informed therapeutic decisions is a reality today. It is still a slow and cumbersome process, however, including from a computational viewpoint. We'll discuss where the bottlenecks are and how they may be resolved in the near future.

How can we model and optimize every aspect of system and application performance? How can we use these models to guide design decisions and re-engineering efforts to adopt applications to changing platforms (e.g. GPU or multicore computing) and better estimate the costs of problem solving?

The promise of temporal modeling is slowly but surely coming to bear. But how can we progress from today's modal superposition methods to tomorrow's full transient analyses with rich, virtual terrains? We'll discuss temporal modeling's current bottlenecks and what is needed to overcome them.

The science, data, and computing challenges of cancer research are immense. What will it take to prepare the HPC community to aggregate talent, allocate computational resources, and collaboratively educate users? Which HPC strategies can deliver life-saving breakthroughs?

A better MRI exists that offers order-of-magnitude improvements in image resolution, signal to noise, and scan time. Image reconstruction demands, however, prevent us from using these techniques. We'll discuss how GPUs and new software can meet this challenge and help enable better MRI diagnostics.

The rapid advancement of HPC brings the opportunity to transform the process of engineering. How can we accelerate product development and system integration through HPC, large data transfer, and real-time, interactive visualization? What does the transformation look like?

Manufacturers must balance complex and competing requirements from multiple engineering disciplines to deliver products that meet quality, schedule, and cost expectations from customers. We'll discuss the simulation and HPC challenges of this process, and the opportunities for breakthroughs.

A keen understanding of variability and uncertainty propagation has become vital to manufacturers in their product development and virtual realism efforts. What are the hurdles to better incorporating these concepts into modeling and simulation solutions, and what does the ideal solution look like?

The Dark Energy Survey aims to deliver new insights about the expansion of the universe. But the task of quickly transporting and analyzing enormous amounts of small data files at various locations is significant for DES and other communities. We'll discuss this distinct challenge and plans for success.

Often, conventional cybersecurity tactics don't translate well to HPC environments. So, in that setting, how should we think about "data security"? We'll discuss how security experts can best protect industrial assets through collaboration with key personnel and understanding HPC data and users.

Today, industrial users can chose from a variety of technical computing options. But how can companies know which modeling and simulation resources to use, and when? How must the HPC community adapt to meet the end-to-end needs of the modern HPC user?

Improvements in question-answering technology will increasingly help a variety of professions with critical and timely decision making. IBM created "Watson" for that purpose, and to even play a little "Jeopardy!" We'll discuss how these grand challenges help advance computer science and society.

The movement this decade from analog to digital capabilities will become even more disruptive for a variety of industrial sectors. But what are the challenges and implications associated with this shift? We'll discuss these issues, with an emphasis on hardware and software capabilities needed by industry.

What are the unique and most significant software challenges facing today's HPC "power users"? What are the common boundaries of existing codes, and how do we push past them?

NCSA will grow from hosting fewer than 10 petabytes of storage today to more than 500 petabytes in the next five years. How can we intelligently scale, store, and share such an extraordinary collection of information? How will the Blue Waters project inform the leap to exascale data in the decade ahead?

What will it take for the industrial engineering community to fully embrace virtual reality? We'll discuss the past, present, and potential future of virtual realism, including the challenges that lie ahead. We'll also touch on what academia can do to help us change the way we do business in the 21st century.

How do we prepare software to make effective use of petascale systems such as Blue Waters? How can we identify performance and scaling issues for systems with 300,000 to over one million cores? What alternatives exist to using the Message-Passing Interface (MPI) in building massively parallel codes?

What grand challenges lie ahead with regard to data, science, and computing? How can academia, industry, and government collaborate effectively to overcome these hurdles? This year's Blue Ribbon panel will discuss these topics, with a focus on the next decade of petascale and even exascale HPC.