Richard Hirsh, consultant to Science Foundation Ireland (SFI), recently spoke with HPCwire regarding Ireland's recent endeavors and how the “Celtic Tiger” is gearing up to become a driving force in HPC.
HPCwire: Describe who you are for our readers and the work that you do.
Richard Hirsh: Prior to leaving [the U.S.] in the summer of 2003, I spent 16 years at NSF, 13 of them in the Division that has been variously called Advanced Scientific Computing, Advanced Computational Infrastructure, and currently is called Shared Cyberinfrastructure. For the last 10 of those years, I was in the Division that ran the NSF High End Computing facilities, and was Division Direction from September 2001 until I retired from NSF in August 2003. My role was to oversee NSF's funding of its “Supercomputer Centers”, coordinate NSF internal activities in computational science and act as a liaison to other government agencies engaging in similar scientific pursuits.
While at NSF, I also had responsibility for other large, visible programs and initiatives other than supercomputing, such as the NSF Grand Challenge Application Groups, Multidisciplinary Challenges, and Science and Technology Centers. From December 1999 to January 2000, I served as the NSF Science representative at the South Pole.
I am currently a consultant to Science Foundation Ireland (SFI), an agency of the Irish government similar to a combination of both NSF and DARPA. My role here is to review proposals in all areas of IT (someone described my position here as being that of all the NSF Program Officers in CISE), and advise SFI on new initiatives.
HPCwire: When was the SFI formed and what was the impetus for its initial investment?
Hirsh: Science Foundation Ireland (SFI) was founded in 2001 with the largest investment in research ever undertaken in Ireland; the five year authorization for the agency was 650 million. SFI is the outcome of an intensive study commissioned by the Irish government in 1998 that assessed the state of the Irish economy and its future. Members of the study panel consisted of leaders in government, academia and industry.
For the past 15 years, Ireland has had unprecedented growth and development, what has been popularly called the “Celtic Tiger,” which has been achieved through focused government policy, partnership with the European Union, the skills of its well-educated population and strong historical ties to the United States. However, the basis for much of this growth was in manufacturing and localization. With the changing state of the world economy, the panel concluded that this approach was not sustainable in the future and recommended a new strategy to generate a knowledge based economy in Ireland.
They concluded that Biotechnology (BioT) and Information and Communications Technology (ICT) represent “the engines of future growth in the global economy… A world class research capability in selected niches of these two enabling technologies is an essential foundation for future growth.” SFI was created in response to this report with a mission that is a strategic one: to build and strengthen scientific and engineering research and its infrastructure in the areas of greatest value and benefit to Ireland's long term competitiveness and development, i.e. BioT and ICT, and to inspire and train tomorrow's science and engineering leaders in Ireland.
In the past couple of years, SFI has expanded its role by creating a Research Frontiers Program. This is an enhanced version of something that used to be called the Basic Research Grants Program. SFI was asked to assume responsibility for this program, which had formerly been administered by another agency within the Irish government, in 2003. These awards are of smaller scale and duration than our strategic grants, but cover the full spectrum of scientific research within Ireland.
HPCwire: Describe the rigorous international merit review for proposals at SFI? Why is this so important?
Hirsh: SFI has established a flexible grants portfolio for investing in research that occurs within Ireland with awards from large centers to individual investigator awards to short term visits of international researchers to Irish universities. Last year (2004) saw 290 grant approvals from SFI, the largest number since it was founded in 2000. SFI is now committed to 450 million in funding over the next five years.
In the IT sector alone, SFI has made over 70 grants, totaling more than 100 million, in subject areas covering Networking & Communications Systems, Information Systems, Software Engineering & Artificial Intelligence, Knowledge & Web-based Systems, Machine Translation, and Computer Modeling & Visualization. One of the newest of the awards within Computer Modeling and Visualization sector is for a National Center for High End Computing.
All proposals received are subjected to rigorous international merit review; it is only through this means that research within Ireland will be able to continue to be competitive at the world level.
Ireland is a small country, smaller than most people realize. The population of the Republic of Ireland only just crossed four million people last year (2004). To give a comparison with the US, South Carolina is the state that is closest to Ireland in both population and geographical size. It is also an amazingly tightly connected country. You can forget about six degrees of separation; in Ireland you are lucky if you can find two degrees!
Getting unconflicted reviewers within the country is a major problem. In addition, and probably as if not more important, SFI is trying to create a world class research environment here. So, we need to compare any proposed work with the best that exists worldwide. Thus, for all proposals we receive, we only use non-Irish scientists as reviewers. They can give us assessments as to whether the research would pass muster in comparison to the best that is going on elsewhere.
HPCwire: Compare SFI to NSF. What are the major differences?
Hirsh: SFI was created using the best of NSF as a model: e.g., merit review, program officers with expertise in their scientific domains, many similar programs, etc. But there are major differences, and these are mostly related to the size of the organization. At the moment, SFI has 40 employees, and roughly a third of those are scientists. There are only 7 people who are at program officer level. This means that SFI's scientific enterprise is the size of a large NSF Division, but with a budget that, if you scale for the differences in populations of the two countries, comes to more than 11 billion dollars per year. We can really make an impact in Ireland.
In addition, since we **are** a small agency everyone working at SFI knows everyone else very well, and we all work together to fulfill the mission of SFI. What it also means is that there is a lot of work to go around. Our size also gives us a distinct advantage when it comes to making decisions. NSF prides itself on being an agile agency within the US government environment, and indeed it is. However, SFI is able to make decisions **very** quickly if there is a strategic need to do so, substantially more quickly than any US agency can. We are one of the policy arms of the Irish government, and as such, are expected to take advantage of opportunities when they arise, collaborating with other agencies within the government if necessary.
HPCwire: Let's go back to your mention of the HEC Center. What was Ireland's role in high-end computing before SFI and the HEC Center initiative? How important is the Center's development to Ireland's overall well-being?
Hirsh: While some activities in computational science existed within Ireland, the systems that were deployed to support them were at an individual, departmental, or university level, funded by individual research projects. Each deployment was viewed as a one time placement of capability with no long term plan for sustainability; there was no easily accessible national computing facility available within Ireland for all researchers.
Ireland simply had not kept up with advances in high-end computing. SFI queried the Irish research community and heard this comment many times from Irish scientists. This fact can be quantified from a more objective point of view by examining the Top500 list of placements of computer systems in use by academia, industry, and government. The last list from November 2004 did not show any systems in Ireland, although there are listings for Belarus, Hong Kong, Indonesia, Malaysia, New Zealand, Singapore and South Africa, nor did the previous list which had entries for Belgium, Colombia, and the American Museum of Natural History in New York City. In fact, Ireland has been listed in the Top500 during only one cycle, at number 485, and that for non-academic usage.
In keeping with its strategic focus on targeting investment that develops human capital, supports strong ideas and promotes partnerships, SFI views its HEC Center award as an infrastructural one of equipment and people. It is envisioned as a Center that underpins, and is expected to support and benefit, the ICT and BioTech communities, but also advances research in other academic disciplines and provides assistance to the Irish industrial sector needing advanced computational capability. The Center is initially foreseen as primarily a strategic boost to Irish research institutions, but included in that is a role as an important asset for Irish industry. And, with substantial capability available, the Center should leverage Irish collaborations in large EU projects.
HPCwire: So how did the Center come to fruition? What obstacles stood in the way of its formation?
Hirsh: To meet the current and future needs of Irish scientists and engineers, SFI issued an announcement for a National Center for High-End Computing in September 2004 to develop a capability available within Ireland. The purpose of this initiative was to enhance the high end computational infrastructure available to the overall Irish research community, push Ireland to the next level in scientific research, and provide a **national** high end computing solution for Ireland.
Two proposals were received by SFI in response to this call, and these were subjected to the scrutiny of international review. The resulting award was to a consortium of Irish universities. It was led by Prof. Andy Shearer of the National University of Ireland – Galway (NUIG).
[Ed. Note: Andy Shearer was named a top person to watch in 2005 by HPCwire. Please visit http://www.taborcommunications.com/hpcwire/features/people05/index.html#shearer for more info on Professor Shearer.]
The successful proposal for the Center outlined a fairly comprehensive inventory of Computational Science and Engineering in Ireland and identified scientists across Ireland who could use high-end computing in their research. It breaks down the needs for high-end computing into four fairly broad “areas”: Physical Sciences (which includes biomolecular modeling and environmental science), Engineering, Life Sciences and Earth Sciences. In addition, it included industrial impact by connecting to two partners in the communications and chip sectors.
The formation of the Center required that all of the major universities within Ireland come together and collaborate on both the scientific and management aspects of the proposal that was funded. After the expected initial traces of territorial behavior were overcome in a country where there never has been a site larger than any single university or department, the major obstacle was time. SFI had purposely set a very tight schedule to get any proposers to focus on the country's need. In addition, the proposed Center leveraged funding for individual pieces of equipment funded by another agency of the Irish government, HEA, which had not yet been acquired. Again, because of the small size of the government and the collegiality of the various agencies, the cognizant members of the HEA staff agreed to allow this transfer of funds across programs to happen.
HPCwire: Now what is ahead for the Center? What are its short term and long term goals?
Hirsh: The Center is in start-up mode right now. They have already hired about half a dozen people, including a very experienced COO. A tender for the systems to be hosted was issued in January, and the orders for two types of machines have already been placed. Given the expected mix of usage, the Center has opted to have both a shared memory machine and a cluster. Delivery of the SMP system is expected momentarily, and the cluster should be in place by late June or Early July. The summer will be taken up by acceptance tests and early users scrambling to do some new science, and the expectation is that the Center will formally open its (electronic) doors to the Irish science community by September 1st.
Longer term goals include participating to a much greater degree in some of the EU's High End Computing and Grid initiatives. Last year I chaired a meeting at SC2004 where the major players in several of these presented an overview of European activities. There were presentations on DEISA, the Distributed European Infrastructure for Supercomputing Applications; EGEE, Enabling Grids for E-science, the major Grid activity funded by the EU; and the UK e-Science Program. Certainly, the IRISH HEC Center can participate in DEISA as another node, and there has been substantial interest in Grid computing in Ireland for several years (we're even funding a project called WebCom-G to develop middleware to hide the grid from a user), so connections to EGEE seem likely also.
Scientifically, we certainly hope that one of the researchers who has access to the system will be able to come up with a breakthrough in whatever field he or she is investigating. And, with the large number of pharmaceutical and device companies within Ireland, we expect that the Center will be able to aid them in any area they may need from bioinformatics to drug design, from chip design to understanding of traffic on a wireless network.
HPCwire: More specifically, what is your role in Ireland's HEC endeavors?
Hirsh: I think that I could describe myself as a catalyst in all of this. There has been a pent-up need in Ireland for some time. There has also been an ongoing policy discussion within the EU on what is called eInfrastructure over here. Last April it just so happened that Ireland held the Presidency of the EU. I organized and chaired an eInfrastructure meeting held in Dublin under the auspices of the Irish Presidency. By that time I had been here for about eight months and had noticed a definite lack of “computation” within Ireland. But this was brought home by a presentation by one of the leading scientists in Ireland who succinctly summarized the situation in Ireland with the following bullet on a viewgraph “Ireland VERY lacking in compute part of eInfrastructure”.
With this as impetus, acting through SFI, I organized a series of small workshops of the leading computational scientists within Ireland to determine if there truly was a need for an Irish computational infrastructure. These meetings and other discussions convinced SFI that this was indeed the case, and the call for proposals followed in September. This is another example of the agility available to SFI. From concept to execution of the call took only four months; I just don't see this happening as quickly in the US.
So I was just in the right place at the right time.
HPCwire: Do you or did you work with Tony Hey's e-Science Program? If so, how has it helped or influenced your work?
Hirsh: Tony and I are old friends, and we are aware of each other's programs, but we have never had the opportunity to work directly with Tony on any of the UK initiatives.
HPCwire: What prompted your move to Ireland?
Hirsh: I had been working with the NSF Supercomputing/HPC/High-End Computing program for 10 years when I was contacted by an old friend from NSF, Bill Harris, who is currently the Director General of SFI. He called me in late 2002 to ask whether I could chair a site visit review panel for one of SFI's Center awards in January 2003. I had never been to Ireland, my calendar was free, and figured it would probably be my only chance to see the country, so I accepted. When I got here it was obvious that he had other fish to fry besides my chairing the panel, and he and I had a long conversation about me coming to work here as a consultant.
After listening to what he had to say about the impact that SFI could have on the country I decided to give the idea serious thought. My wife and I had always wanted to live abroad, after 10 years in supercomputing at NSF I needed a change, and there was nothing really holding us in the US (there were no grandchildren then; there are now), so I brought my wife to Dublin to have a look around and, to make a long story short, we moved here in August of 2003.
HPCwire: How is your life different than in America? How are HEC endeavors different?
Hirsh: Dublin is a lovely little city. The greater Dublin area has probably about a million and a half people, so it provides all the amenities that big city lovers like me can enjoy. Probably the biggest difference between living here and back home is that we don't own a car, **and don't need one**. I live about a mile from work, and walk every day. I'll be in better shape when I get home than I have been in probably ten to fifteen years.
The vaunted Irish weather is not a problem. But it is, shall we say, changeable. I think the old saw about “if you don't like the weather, wait half an hour” must have its origins in Ireland. It doesn't rain like we're used to in the States – two days of unrelenting downpours. The rains here are “soft” and usually only last for and hour and a half or so, and are followed by periods of mild sunshine. It's a very temperate climate; in the winter the temperatures range from 40 to 55 (compare that to New Hampshire), and in the summer the range is from 50 to 65 (and compare that to DC). There is hardly any day above or below these temperatures, and you can very easily get used to not being cold or hot during the year.
In answer to the question about HEC differences, what we've done here for High End Computing very closely mirrors what happened at NSF in the mid-80s. The needs of the research community are the same the world over. The only differences are the scale of the country, and SFI's ability to move quickly to meet these needs.
HPCwire: Looking ahead, where do you see Ireland on the HEC map in one year? How about 5 years?
Hirsh: In one year I would like to see the Ireland HEC Center as an established part of the research infrastructure of the country. It will be a ramp up period, and the Center needs to meet the requirements of its users as its principal start up goal. In five years, I think it will be an integral part of a greater European infrastructure, while still playing a vital role for Ireland's academic and industrial researchers.
We also have some interesting possibilities on the shorter term horizon that I can't discuss right now, but, hopefully, watch this space!
HPCwire: I'm sure we will be! How do perceptions of U.S. HEC endeavors play out in Europe?
Hirsh: The Europeans believe that with projects like EGEE and programs like e- Science, Europe is ahead of the US in Grid-related efforts. They see their eInfrastructure initiative as being equivalent to, and ahead of cyberinfrastructure. I think they look longingly at the US's ability to continue to push the envelope at the very top end; there is nothing equivalent over here.
HPCwire: What should the world look for as Ireland carves out its high-end computing and research plans? What's next?
Hirsh: Some possibly interesting applications of high end computing, possibly in data intensive computing. Research, in some cases, on a par with any in the world. And again, maybe a surprise or two in the next year.
HPCwire: Thanks for informing us about the developments there, Rich. I'm sure the community will be watching even more closely now!
Mr. Hirsh's official title is Consultant – ICT Research Advisor. He spent his active research career, from 1970 to about 1988-9, doing computational fluid dynamics and research into numerical techniques and algorithms. In 1983, some of this work was higlighted in a lecture series at the von Karman Institute for Fluid Dyanmics in Belgium.