A key tenant of design thinking is rapid prototyping — the idea that testing a new idea early and often in an environment that supports and learns from failure is crucial to driving innovation forward.
Now, the University of Minnesota is bringing that same concept to high-performance computing. As part of its goal of advancing research, the U’s Minnesota Supercomputing Institute launched a new program July 1 that aims to give researchers and staff a chance to try a variety of new and experimental computing technologies, including visualization, data mining and big data systems. The program, called MSI Beta, allows MSI to investigate technologies in an environment that embraces failure as part of the discovery process and supports the development of novel applications that have the potential to accelerate research at the U.
“Universities are known for finding new ways to work with cutting-edge technology,” said Claudia Neuhauser, Ph.D., interim director of MSI. “It is important for the U of M to have exploratory, low-stakes environments like MSI Beta to let us try, fail and learn from the experiment. This program will ultimately outfit MSI with more tools and knowledge to better help university researchers push the boundaries in their fields.”
From time to time, MSI gets new technology in the form of software or hardware, often as part of a larger upgrade. For example, the institute’s newest supercomputer, Mesabi, includes cutting-edge graphics processing capabilities that staff believe could be particularly useful for high-resolution visualization — a technique that can, for example, help show models of the human body in greater depth to enhance medical researchers’ anatomical knowledge. Previously, the institute did not have an explicit space where it could experiment and learn from failure without hindering ongoing research projects. MSI Beta lets staff and faculty explore these new technologies and, if successful in gaining a more complete understanding of it, transfer the technology to MSI’s line of standard supercomputing services for broader use.
Neuhauser and MSI staff have set out to find projects around campus that look to be good matches for the experimental technologies they have now. For example, MSI recently received a piece of hardware called a Field Programmable Gate Array (FPGA). Unlike most FPGA systems, which are manufactured to perform a specific set of operations, this particular system lets users custom-program its functions using C, a widely known programming language.
The FPGA technology will be the basis for MSI Beta’s first project. MSI is partnering with Amy Kircher, director of the U’s National Center for Food Protection and Defense, to use the system for text mining — where intelligent computer programs automatically search through data from a variety of sources, organize it and analyze it. This system will help alert NCFPD to potential threats in the food supply and will assess the risk they pose and provide data to help mitigate the threat’s impact. Then, as part of its Focused Integration of Data for Early Signals program, the center will share this information with governments, non-governmental organizations and the private sector to help these groups react to food supply threats like listeria and salmonella.
Another technology within the MSI Beta program is a new Hadoop system, which is built specifically to address the challenges researchers face when analyzing very large data sets. MSI will explore how the new system can meet researchers’ big data needs and will test its ability to run other related computing frameworks that support machine learning — where the system can “learn” from the data it has already processed to discover new insights without being told by the researcher where to look for this information. Machine learning can help researchers draw new or unexpected results from their data without the need for any extra human input.
Researchers interested in exploring opportunities in MSI Beta may contact Claudia Neuhauser, MSI interim director, at email@example.com.