XSEDE IMPACT DECEMBER 2017
A better approach: Bringing supercomputing to psychology
Studying the psychology of decision making can be messy. In a typical decision making study, subjects make choices and psychologists compare their responses to the predictions of a particular theory. However, changing the parameters of a theory even slightly can spawn thousands of variations. In consequence, testing just one of those theories, or a handful, is unlikely to lead to the best possible model -- and choosing the "best" of a few theories often means massaging the data. With support from XSEDE, specifically Bridges at PSC and ECSS support, University of Illinois' Michel Regenwetter is bringing a new level of mathematical and statistical rigor to studies of decision making and behavior -- and showing other scientists why they should do the same. Read more >>
Finding the Balance
Pitt Researchers Use Bridges to Optimize Doctor Visits for Kidney Patients
The biggest causes of death in developed countries—heart disease, cancer, chronic organ failure—are complex conditions that may be best treated by long-term management rather than attempting to "cure" them. Researchers at the University of Pittsburgh have used the XSEDE-allocated supercomputer Bridges at the Pittsburgh Supercomputing Center to mine a massive database of medical records to optimize preventive doctor visits for individual patients with chronic kidney disease.
The Impact of XSEDE Training
"One usually thinks of the impact of a supercomputing center as being a new cutting edge simulation or a scientific breakthrough only possible with the center's resources. There is little emphasis on the impact of training, including training on the access and use of HPC resources and training on HPC methods.
I was fortunate to be able to attend the last two monthly XSEDE HPC training sessions, one on OpenACC and one on Big Data. According to XSEDE User Support and Outreach Coordinator Tom Maiden, who also doubles as PSC's Manager of User Services, the training series tries to teach ‘the nuts and bolts, or building blocks of high performance computing' in a friendly computing environment with immediate student support. I have been using supercomputers since ‘Terascale' referred to a Supernova challenge and not to the size of a 50-dollar external hard drive from Best Buy; these courses do not simply teach the next generation of students how to use HPC, but they give us legacy FORTRAN programmers a wonderful chance to modernize our skill sets. The platform independent and user-friendly setup allows beginners to use the resources immediately and relieves seasoned HPC'ers from sitting through hours of ramp-up slides.
Hosted at Pittsburgh Supercomputing Center with 25 remote satellite locations, the closest venue to me was not in the usual tech hub of Research Triangle Park, NC. It was hosted at North Carolina Agricultural and Technical State University, NC A&T, the largest public HBCU in the nation. I'm sure this was a boon to local students and professors, but it was also eye-opening for me as I read more about the school's accolades and certifications when checking out their website. I realized that there is enormous potential out there that is being lost due to the lack of networking. XSEDE realizes the importance of capturing that potential, making resources available to institutions where it is not feasible (or desirable) to support an on-campus HPC center.
The courses are not available as webinars. This is important. Although it was slightly painful to drive to the venue and take vacation from work, the content covered in these sessions require full concentration and they are not something you can follow by donning headphones while doing a day job. The content is artful in the way that it covers everything completely in an understandable manner, but at lightning speed with absolutely no filler. Besides the need to concentrate fully and take advantage of the live question and answer sessions, the ability to be on-site to network is invaluable. I met a statistics professor (who drove 3.5 hours from her location to attend), and talking about our work and how we could use the information in the course for our current research was both enlightening and energizing.
My current job is as an HPC scientific consultant for EMVL (Environmental Modeling and Visualization Lab) at the EPA, which is basically the EPA's equivalent of XSEDE's ECSS. With ocean and air quality modeling, there is no shortage of legacy FORTRAN codes for me to parallelize with MPI; however, there are also plenty of projects that can and will benefit from my learning new HPC tools and techniques. Today was my first day back at work after the Big Data workshop, and I've already talked to four interested colleagues about Spark and TensorFlow. Being knowledgeable about XSEDE's platforms and resources also helps me present computing options to customers that are, for security reasons, unable to fully collaborate with outside agencies, and to customers who leave EPA and are subsequently unable to continue to use our systems. One final note about the training from the perspective of an outsider: a friend of mine joined me for the trip, so he could have some down time to study for a certification exam and so I could avoid driving myself to Greensboro. He used the word ‘amazing' twice: 'It is amazing how much there is to the topic of Big Data. I learned a lot from just sitting in the room' and 'It is amazing how much knowledge that guy John Urbanic has in his head.' Kudos to John. It is a rare gift to have someone who is both an expert in the field and a skilled presenter and educator. Thanks again to XSEDE for this amazing experience!"
-Lisa L. Lowe, High Performance Computing Scientific Programmer at Leidos Innovations Inc., and CSRA for the United States Environmental Protection Agency (EPA) in Research Triangle Park, North Carolina.
To learn more about XSEDE trainings and register for courses, please click the following link.
XSEDE's booth at SC17 was quite a success! We were able to share XSEDE's value and opportunities with several conference attendees, and hope to see increased use from those interactions. We look forward to seeing you at SC18 in Dallas, Texas!
Satellite Site Invitation for the XSEDE HPC Monthly Workshop Series
XSEDE is pleased to announce a regular series of remote workshops on High Performance Computing topics. These hands-on workshops provide a convenient way for researchers to learn about the latest techniques and technologies of current interest in HPC. We are currently accepting satellite sites for:
· January 9 - OpenMP
· February 7-8 - Big Data
You, or another representative of your institution, should contact us if you are interested in hosting a remote site. The number of sites will be limited for any given event. Please note that this is not individual student enrollment; we will post registration pages for student enrollment once the remote sites have been determined.
Unless otherwise stated, this series will run from 11AM-5PM Eastern Time each day so as to encourage participation in all time zones.
Interest in this series has grown dramatically and as such we routinely have more sites apply than we can handle. We will announce the list of accepted sites about 1 month prior to each event. Please do not assume your site has been accepted until you are notified as such.
These workshops are very hands-on, and use the foremost available platforms for their subject matter. They are a continuation of the previously oversubscribed XSEDE remote workshops and an effort to reduce the backlog of demand while maintaining the quality.
Satellite sites MUST be able to provide a facility capable of two-way video for the duration of the workshop. There should also be a full time TA available to assist with local technical issues. An AV system test will be scheduled sometime before each workshop. Failure to complete a successful AV test prior to an event and subsequent AV issues during the workshop my affect the priority level of your site for future events. Address any questions regarding course content to John Urbanic (firstname.lastname@example.org) and questions regarding registration to Tom Maiden (email@example.com).
The schedule for the remainder of 2018 will be as follows. We will put out a call for interested satellite sites at an appropriate time in the future for these events
· March 6 - GPU Programming Using OpenACC
· April 3-4 - MPI · May 1-2 - Big Data
· June 4-7 - Summer Boot Camp
· August 7 - OpenMP
· September 5-6 - Big Data
· October 2-3 - MPI
· November 6 - GPU Programming Using OpenACC
· December 4-5 - Big Data
February 2018 Deadline Approaches for the Blue Waters Graduate Fellowship
The Blue Waters Graduate Fellowship Program is a unique program funded by the National Science Foundation. As part of this program, several graduate students from across the country will be immersed in a year of focused high-performance computing (HPC) and data-intensive research. The fellowships will empower the selected PhD students to advance their HPC knowledge while also providing them with access to Blue Waters and support to accelerate their research. Applications, including all supporting materials except reference letters, must be submitted no later than 11:59 pm PST on Feb. 2, 2018. Reference letters are due by 11:59 pm PST on Feb. 9, 2017. Awards will be announced in spring 2018, with the tuition allowance applied to the 2018-2019 academic year. For complete information on the fellowships, visit https://bluewaters.ncsa.illinois.edu/fellowships. Questions? Contact firstname.lastname@example.org.
SC18-Nov. 11-16, 2018 | Dallas, TX