From The Directors
Tick, Tock Goes The PSC Clock…
Some new systems augment previous research capabilities through improved performance, although they are very similar to previous systems. They are the ticks. But it is the TOCKS, the systems that break new ground in technology and architecture, that can, in addition to improving performance for existing application areas, have a disruptive effect. These are the systems that open advanced IT capabilities for a wide range of new fields of scientific research.
The Bridges system, now entering production and whose initial scientific output is described in this 30th anniversary issue of People. Science. Collaboration. (p. 6), is just such a system. With its new hardware, architecture, software and management capabilities it is already supporting multiple new, exciting areas of scientific research.
Using heterogeneous hardware components flexibly configured with powerful system management software, Bridges can fill a far wider range of research needs than would be possible with a statically configured system of identical components. Using software to reconfigure a tightly integrated (thanks to the first application of Intel’s new OmniPath architecture) system of very different components (GPU nodes, extreme memory nodes offering 12 TB of coherent shared memory, database nodes, etc.), Bridges can become a number of different supercomputers tailored for different users and computational tasks.
Included is a growing number of “new community” users that have never before needed HPC—and, therefore, have never before had a system quite tailored to their needs. PSC’s previous systems have been friendly to new communities; Bridges is, in a virtual sense, created for them, giving them unparalleled ability to port existing successful computational methodologies into HPC rather than having to rewrite their codes and procedures.
PSC also continues to help pave the way to super-specialized HPC systems that do a single task far more quickly than possible with an all-purpose supercomputer. Thanks to an agreement with D. E. Shaw Research (DESRES) and a new National Institutes of Health grant, we will continue hosting the Anton molecular dynamics simulation system until October . More importantly, at that point we will begin hosting a far-more powerful DESRES Anton 2, capable of modeling much larger sub-cellular structures than just proteins or small groups of proteins.
While we look forward to the future typified by these two very different but very powerful machines, our 30th anniversary also provides an opportunity to look back at the progress we’ve made. In this volume we have paired most of the new stories with PSC “greatest hits” about earlier work in similar fields. The story about Anton 2 includes a “PSC Lookback” sidebar that harkens back to University of Pittsburgh researchers’ use of one of our earliest machines, the Cray Y-MP, in the 1990s in an early molecular dynamics study of the interaction of the EcoR1 protein with DNA. A new study of driver-injury modeling appears alongside Alcoa’s work with our Cray C90 to model aluminum auto body designs in the late 1990s .
A report of how key stocks can be used to better predict movement in the market pairs with 2009 Carnegie Mellon University work using PSC’s Pople system to show how easily social security numbers could be guessed just from publically available information on the Web . PSC’s collaboration with Texas A&M researchers on coolant flow through a promising new, inherently safer, nuclear reactor design recalls a 1993 model of solar turbulence using the C90 ( An exciting new 3D reconstruction of sub-cellular connections between excitatory neurons in the visual cortex, made possible with PSC hardware, software and expertise is reminiscent of early C90 work in reconstructing the 3D anatomy of the beating heart . And a University of Delaware study with Anton of how cholesterol affects biological membranes appears with a PSC Lookback on University of Illinois work using our LeMieux system (at the time, the world’s second-fastest computer) that showed how the aquaporin membrane channel excludes hydrogen ions—work cited in aquaporin discoverer Peter Agre’s 2003 Nobel Prize acceptance speech . See the table of contents for more of these stories and their associated Lookback pieces.
Not all of the stories in this celebratory issue can easily be related to early PSC work, though, and that’s part of the point of Bridges. A University of Illinois group used our now-retired Blacklight system, the transitional Greenfield system and soon Bridges to glean clues about the life experiences of black women in historical periods when they could not share their lives in print (p. 8).
This work is exemplary of the new users whom, along with those in “traditional” supercomputing fields, we expect Bridges to serve by design. We look forward to helping and working with researchers in fields never before employing high performance computing to use Bridges to make transformational contributions. And with the help of the XSEDE network’s programs, sites and personnel, Bridges will reach an ever-expanding community of new users.
Education and workforce development continue to accompany research and technological development as major foci for PSC’s efforts. This year we are proud to note that the leaders of PSC’s bioinformatics education group have received a prestigious Carnegie Science Center Award for STEM Education . Accompanying this article is a pointer to a few of the many successful bioinformatics researchers and educators who have benefitted from our educational programs. In addition, our Director of Communications and Industrial Relations, Cheryl Begandy, has been recognized with the 2016 YWCA Greater Pittsburgh Tribute to Women Leadership Award for Science & Technology.
We at PSC enter our fourth decade (!) excited to see where new systems, new science and above all the continued ingenuity of our people will take us next.
As always we would like to acknowledge all our funders, especially the NSF, the NIH, and the Commonwealth of Pennsylvania. We would also like to hear any feedback you have on our work or this publication. You can send any comments or suggestions via our feedback page at psc.edu/index.php/feedback. You can also contribute to PSC’s nonprofit, academic mission at psc.edu/donate.