Ontario’s researchers increasingly depend on advanced computing to develop new treatments, understand diseases and much more.
For example, advanced computing helped researchers at the Sick Kids Hospital develop new treatments for a Parkinson’s-like disease, and the Brain-CODE project is making it easier than ever for researchers to share and analyze patients’ clinical data.
But does Ontario have adequate advanced computing infrastructure? And if not, what can we do to rectify that situation? This is a question ORION will be exploring in our “Advanced Computing: Transforming Research” workshop on February 13.
To kick-start that conversation, we wanted to share a few insights from one of the workshop’s key speakers. John Towns has gained significant experience in building national and international cyber-infrastructure as a board member of Compute Canada and Director of the Collaborative eScience Programs at the National Centre for Supercomputer Applications (NCSA) in the U.S.
We conducted a short Q&A session with Mr. Towns to get some of his initial insights into the future of advanced computing in Ontario:
Question 1: Why should institutions and/or governments make advanced computing a priority?
John Towns: Advanced computing, in the form of modeling and simulation, has become the backbone of research and development in the traditional “hard” sciences and engineering in both academia and industry. It is already critical to both academic and economic competitiveness and will only continue to be more so. Underscoring this, the use of advanced computing is flowering in many of the non-traditional fields – most obviously those that are heavily data-oriented.
For institutions, the availability of advanced computing resources will enable greater competitiveness of their researchers and as a result the institution will have another means for attracting good faculty and students. For the students engaged in research using advanced computing resources, they will become equipped with a necessary skill that is not common and can provide them with a competitive edge. Advanced computing is well established as a key enabler of research in many fields and this is only going to continue to expand to other disciplines. Before long, not having such resources will be a detriment to these institutions’ competitiveness. By extension, these same issues apply to Ontario. Further, there are many examples of the benefits of access to advanced computing resources in all of Ontario’s key economic sectors of agriculture, manufacturing, energy and mining.
Question 2: How does collaboration improve the provisioning and support of advanced computing?
John Towns: There are really two aspects here. Provisioning advanced computing resources in and of itself really does not require these collaborations, but those deploying, operating and supporting these resources benefit greatly from these collaborations since they allow them to be much more effective and efficient at what they do. This is very important, but probably of greater importance is that the purpose of provisioning these resources is to support the researchers using them. Those researchers are more and more engaged in multi-institutional, regional, national and international collaborations. Those distributed collaborations are best supported by a collaborating set of providers of advanced computing resources and services.
Question 3: What kinds of trends are you seeing with regards to advanced computing?
John Towns: I was recently asked this question as part of the interview by HPCWire associated with them naming me as one of their “People to Watch” in 2014. The short version of the things to keep an eye on are:
- Low-power processors: many of us have been anticipating the rise of these this for some time. We will finally begin to see this play out this year and spilling into next as more interesting – from a High Performance Computing (HPC) perspective – versions of these processors roll out.
- Accelerators become (HPC) mainstream: This is pretty much here already, but we will begin to think of these things as far less exotic by the end of this year. They will be assumed, particularly as they become much better integrated and/or interconnected.
- Data becomes the driver and not the raison d’etre: Of course, Big Data (or whatever term you want to use) is here to stay, but this has been looming on our horizon for a while. The change I anticipate is that we will begin to realize that it is not just about Big Data, but how it drives requirements in our computing systems and infrastructure.
- Integration of multiple resources is key: OK, so this one is more my hope than a technology to watch. Too many of us in the HPC community put too much emphasis on monolithic HPC systems. For real productivity on a day-to-day basis for researchers and engineers, they need comprehensive, integrated, distributed resource ecosystems to conduct their end-to-end workflows.
The most interesting part of all of this is how it will all fit together to provide fundamentally new capabilities to enable research. Low-power processors driving a bunch of accelerators with high bandwidth connectivity to support large/fast data flows integrated with distributed data sources, visualization resources, and analysis engines… now that could get interesting!
Question 4: Without revealing too much, can you give us a short preview or teaser about your presentation at the ACTION workshop?
John Towns: I will present a high level overview of XSEDE (Extreme Science and Engineering Discovery Environment – a virtual system for scientists to share computing resources, data and expertise) from the point of view of building a national/international scale cyberinfrastructure and the associated collaborations necessary. If nothing else, XSEDE has been an adventure in social engineering and I will comment on some of the aspects of these highly distributed and complex interactions. Along the way I will share some of the pitfalls, lessons learned and continuing challenges.
Want to hear more from Mr. Towns? Sign up to attend ORION’s Advanced Computing workshop.