Building Canada s Future Research and Innovation Culture

Transcription

1

2

3

4

5 December, 2007 Building Canada s Future Research and Innovation Culture Executive Summary Canada s current high performance computing (HPC) infrastructure is a success story. Seven regional HPC consortia represent over 50 institutions across the country and more than a thousand faculty who, along with research associates, post-doctoral fellows and graduate students, represent a community of roughly six thousand researchers performing leading-edge computationally-based research. Working together, resources have been shared and collaboration has been fostered between researchers, between institutions, and between regions. With the creation of the Canada Foundation of Innovation and matching provincial funding programs, federal and provincial funding of research has been combined with industry and university contributions with the result that Canada s deteriorating international research competitiveness has been turned around. In 2005, C3.ca Association Inc. published its Long Range Plan for High Performance Computing (HPC) in Canada: Engines of Discovery: The 21 st Century Revolution [1]. In 2006, the Canada Foundation for Innovation created the National Platform Fund program in part as a response to the Long Range Plan. In 2007, the Long Range Plan for HPC in Canada was re-examined in light of the significant accomplishments of the past decade and the promise of the future. In the following pages, the accomplishments of the HPC community are recognized, while also highlighting what still needs to be done. This renewed plan recasts the recommendations of the Long Range Plan as Eight Imperatives for HPC in Canada. Meeting these imperatives will create and sustain Canada s future research and innovation culture. The following table lists the Imperatives, along with a brief status Imperative Communicate a clear and compelling vision Sustain our nationally shared regional HPC consortia Invest in the development of highly qualified personnel (HQP) Found a national body for HPC Compute Canada Collaborate with CANARIE to build Canada s emerging cyberinfrastructure Establish an internationally competitive Tier One Facility Establish sustained and robust funding for HPC in Canada Current Status Original 1997 vision still valid today, advocacy for HPC in Canada must continue. The pre-eminent success story for HPC in Canada is based on CFI, provincial, industry and university support of research/innovation. NSERC has responded with increased funding to address underfunding of HQP. A significant milestone was reached in 2007, more work needed. Grid computing and major collaborations build on fast national and regional networks No funding is available. A new policy context is needed. National leadership is required. Facility and operations costs only now being addressed. A second round of CFI NPF funding is crucial to reinforce current gains. 8. Demonstrate benefits for Canada Economic benefits are clear, HPC seen as key to global competitiveness.

6

7 December, 2007 Building Canada s Future Research and Innovation Culture 1 Introduction 21 st Century computing involves a richly-connected digital world where high performance computing, or HPC, enables research across all the sciences (including the health, environmental, mathematical, physical, engineering, life, and social sciences) and increasingly across all the arts (including the humanities, languages, commercial and performing arts). Canada needs sustained funding of computational infrastructure to build on the enormously successful investments made to date, to capitalize on Canadian researchers proven ability to produce world-class research, and to build on Canada s world-leading networking and telecommunications infrastructure. What is HPC? High performance computing has transformed research in Canada. Globally, HPC has become pervasive, growing from a handful of HPC sites worldwide in 1984 to numbers in the thousands of sites today. Computer simulations and models now supplement or even supplant traditional field or laboratory experiments in many disciplines. The use of high performance computers has created new means of enquiry in the computational arts, sciences, engineering and medicine. HPC s true power transcends the mere provision of cycles by serving more and more purposes, as innovators seize the opportunity to challenge fundamental assumptions regarding what can be done. With HPC, individual innovators can have a global impact on our daily lives. For example, Google began with the mission of organizing the world s information and making it universally accessible and useful [2]. Google is powered by massive server farms executing sophisticated search algorithms. Google typically responds to millions and millions of search queries, each in less than a second thanks to HPC. If Google is a search engine for all that is known then HPC is a search engine for all that is unknown. Three case summaries from the Long Range Plan document the importance of HPC: If Google is a search engine for all that is known then HPC is a search engine for all that is unknown. Over the next decade, the strategic intent of the Meteorological Service of Canada is to provide the science capacity in support of risk-based decision making involving atmospheric and related environmental change and variability which affect Canadians security and health, economy and the environment on the scale of a few hours to centuries. Significant and sustained investments in human and HPC infrastructure will be pivotal to attain our strategic objective. [Dr. Michel Béland, Director General, Atmospheric and Climate Science Directorate, Environment Canada] The human body is undoubtedly the most complex machine ever created. Ge-

8 2 Building Canada s Future Research and Innovation Culture December, 2007 nome researchers are undertaking the challenging task of unravelling how it is organized and how it functions. Without extremely sophisticated computational infrastructure, genomics research would not be possible. [Dr. Cristoph Sensen, Director, Visual Genomics Institute, University of Calgary] Nanotechnology is research and development at the atomic or molecular level in medicine, biotechnology, genomics, manufacturing, computing, information technology, communications, and other fields. In the long term, nanotech is all about computing. It s the only tool we have to bridge the chasm between theory and experiment for hard problems that resist analytical treatment. Which, of course, means almost all problems [Dr. Mark Freeman, Canada Research Chair in Condensed Matter Physics, University of Alberta] The Technology Challenge is Unprecedented Research using high performance computing immerses users in a networked environment that is rich in powerful and interconnected technologies. The range of hardware used for HPC is staggering: from individual PCs on researcher s desktops to huge systems with tens or hundreds of thousands of connected processors, massive storage devices and advanced visualization. Moore s law 1 states that the number of transistors on a computer chip (one determinant of computer power) doubles every eighteen months (roughly), implying that computing times for specific research problems are rapidly being reduced, bringing more challenging problems continually within reach. Moore s law implies that physical and engineering limits are constantly being pushed back by the innovation process. Remarkably, this exponential growth has endured for the past four decades. The competitive challenge of maintaining the country s position in The Top 500 global rankings of HPC infrastructure [3] is unprecedented in Canada s history. 2 Roughly speaking, Canada s HPC research capability must double every year just to maintain the status quo. In 1993, the Atmospheric Environment Service of Environment Canada peaked at #6 on the Top 500 list with one third (33%) of the computing capability of the top position. Canada s best ranking as of November 2007 is #391 (Université de Sherbrooke) at 6.9 Teraflops (only 1.4% of the top ranking of 478 Teraflops). Roughly speaking, Canada s HPC research capacity must double every year just to maintain the status quo. It is important to note that, unlike other research investments, Canada s HPC capacity can remain competitive as long as the funding is sustained and the investment addresses all the imperatives of a comprehensive strategy. In the last decade, parallel computing has emerged as the foundation of HPC. The peak performance of parallel computers is limited only by budget, space and power. In contrast, the actual performance of an application is limited by its scalability, 1 Gordon Moore is the co-founder of Intel. Moore s observations were first made in Since 1993, the international standing of HPC facilities around the world have been compiled semi-annually in a list called the Top 500 Supercomputing Sites, available online at

9 December, 2007 Building Canada s Future Research and Innovation Culture 3 the ability to efficiently exploit the power of ever-larger parallel computing systems. Scalability deficiencies cause huge penalties in application performance which in turn limits the research that can be done. An application that is 90% efficient can effectively use a parallel computer with no more than 10 processors. Conversely, an application that scales effectively to 10,000 processors must be % efficient. The scalability of an application is split between the efficiency of the hardware on which it is executed and the efficiency of the software. Chip-makers have turned to multiple processors on a single chip (now called cores) to boost performance; in effect, transferring the innovation challenge from hardware developers to software developers. Software efficiency is almost entirely due to the development skills of the researcher and his/her team and, crucially, on the technical support they receive. Therefore, while scalability on the hardware side is a technology infrastructure issue, scalability on the software side is a human infrastructure (i.e. highly qualified personnel) issue. It is important to note that, unlike other research investments, Canada s HPC capacity can remain competitive as long as the funding is sustained and the investment addresses all the imperatives of a comprehensive strategy. The level of funding ultimately invested will then largely determine our international competitive standing. Eight Imperatives for HPC in Canada Our vision (originally crafted in 1997) continues to be the creation of a Canadian fabric of interconnected technologies, applications and skills based on advanced computation and communications systems applied to national needs and opportunities for research and innovation in the arts, sciences, engineering and medicine. Imperative 1: Communicating a Clear and Compelling Vision Recognizing the need for long-term vision and stable funding for HPC in Canada, C3.ca Association Inc. 3 convened an expert panel to develop a Long Range Plan. The panel began its work in December 2002, supported by the National Research Council of Canada, the Canadian Foundation for Innovation, the Canadian Institutes of Health Research, the Natural Sciences and Engineering Research Council, the Social Sciences and Humanities Research Council, and CANARIE Inc. C3.ca published in 2005 its Long Range Plan for High Performance Computing in Canada: Engines of Discovery: The 21 st Century Revolution [1]. The Canada Foundation for Innovation created the National Platform Fund program in part as a response to the Long Range Plan. The regional HPC consortia responded with a proposal to establish a new entity Compute/Calcul Canada [4]. An International HPC Experts Committee reviewed and supported the proposal and made many recommendations [5]. 3 C3.ca was incorporated in 1997 as a not-for-profit association advocating on behalf of the academic and industrial research community to promote high performance computing in Canada. This advocacy role will be assumed by Compute/Calcul Canada beginning in 2008.

10 4 Building Canada s Future Research and Innovation Culture December, 2007 In 2007, the Long Range Plan was re-examined in light of the Compute/Calcul Canada proposal and the International HPC Experts Committee recommendations. This document is the result and renews the Long Range Plan by updating the plan to include the proposed Compute/Calcul Canada structure and the international committee s recommendations. Any vision for HPC in Canada should not enshrine the status quo. This vision should include innovating HPC itself to meet the expanding needs of a knowledge society. New frontiers for HPC are available in the arts, social sciences and humanities. Deficiencies in capital resources should not limit research and innovation provided that we innovate in other ways, by investing in highly qualified personnel, for example. An open mind towards the future is needed. Current status: The Long Range Plan s vision and strategic priorities for HPC in Canada are clearly aligned with the strategic priorities of the Canadian Government s Science and Technology Framework [6]. A decade of community commitment to a clear and compelling vision is the foundation on which Compute/Calcul Canada can build Canada s future research and innovation culture. The National Platform Fund is an affirmation of the Long Range Plan and, in 2007, the Compute/ Calcul Canada proposal and the International HPC Experts Committee review have been used in turn to renew this Plan. Imperative 2: Sustaining the Nationally Shared Regional Consortia The Canada Foundation for Innovation s National Platform Fund has enabled the seven regional HPC consortia ACEnet, CLUMEQ, HPCVL, RQCHP, SciNet, SHARCNET and WestGrid to move one step closer to creating one national initiative called Compute/Calcul Canada. The regional HPC consortia represent the pre-eminent success story in the history of HPC in Canada. According to Prof. Martyn Guest of the UK s Central Laboratory of the Research Councils at the Daresbury Centre [1]: Canada has invested wisely in mid-range computing over the last five years and has created the best developed, most broadly accessible mid-range high performance computing facilities in the world. Since the regional consortia emerged, they have interacted both formally and informally. The consortia have historically shared up to 20% of their resources with HPC researchers from other consortia within Canada. Under the CFI National Platform Fund, the regional consortia have proposed that all the resources of Compute/Calcul Canada would be fully shared to serve the HPC needs of Canadian university researchers, regardless of affiliation. This represents a major (and evolutionary) leap forward. Canada has invested wisely in mid-range computing over the last five years and has created the best developed, most broadly accessible mid-range high performance computing facilities in the world. (Professor Martyn Guest, Daresbury Centre, UK) Current status: Public funding required for the regional consortia (excluding other sources of funding) is $44 million per year, rising to $54 million per year by 2012.

11 December, 2007 Building Canada s Future Research and Innovation Culture 5 Prior to the successful National Platform Fund proposal, CFI had invested a total of $108 million in HPC infrastructure since Since CFI funds are at most 40% of the cost, this investment has been leveraged with provincial and industry contributions to acquire $270 million of infrastructure. The successful Compute/Calcul Canada proposal represents $78 million of new CFI funding (including $18 million of infrastructure operating funds) in large part for four of the regional consortia over the next three to five years. NSERC is contributing an additional $10 million towards personnel. With leveraging and NSERC contributions, this represents new investments totalling $150 million for capital infrastructure and $28 million to operate the facilities and provide technical support for researchers. Imperative 3: Invest in the Development of Highly Qualified Personnel Highly qualified personnel literate in HPC are critical for leveraging Canadian investments in HPC to Canada s economic and social advantage a key requirement in the new global economy. Our global competitor s investments in the people who support academic research teams range from 20% to over 50% of their capital investments. The authors of the Long Range Plan chose a conservative target of 25% towards developing highly qualified personnel. Harnessing the power of parallel computing has highlighted a scalability gap in Canada s research and innovation infrastructure. In the United States, broadbased investments in human infrastructure over the past several decades has supported research into new software tools, applications and algorithms. Software innovations have provided performance gains that are, in some cases, on par with the astounding hardware performance gains over the past 40 years, largely by confronting this scalability challenge. In Canada, the authors of the Long Range Plan recognized the strategic importance of investing in the development of high quality personnel, but there has only been a weak response by government funding agencies to address this imperative. Developing human infrastructure is much more challenging than developing technology infrastructure. It can take years to train people with the necessary skill sets, and then they can be easily enticed elsewhere by the lure of better opportunities and higher salaries. If Canada is to invest in people and skills development, then it must also invest in creating the environment to attract and retain them. While the number and quality of researchers using HPC in Canada has enjoyed a renaissance; in other respects, Canada s human infrastructure in support of HPC remains inadequate (although recent investments are beginning to have a positive impact). Research Teams Enjoy a Renaissance The number of faculty, research associates, post-doctoral fellows and graduate students working on HPC-related projects has increased from a few-hundred in 2000 to roughly six thousand in 2006 (the last year user statistics were compiled nationally). This growth clearly demonstrates prior investments in HPC have played a critical role in increasing Canada s commercial, research and innovation competi-

12 6 Building Canada s Future Research and Innovation Culture December, 2007 tiveness through the development of highly qualified personnel. Leading this research and training renaissance in Canada is a growing pool of talented professors, ranging from senior Canada Research Chairs with outstanding international reputations to junior assistant professors with tremendous research potential. Computational based research, which is conducted by graduate students and post-doctoral fellows under the guidance of these research professors, is an important vehicle for the training of highly qualified personnel. It often takes a decade or more to build strong research teams and have their technology evolve from ideas into commercial products. It takes at least as long to train skilled workers with advanced research and development capabilities. The development of highly qualified personnel constitutes the most effective form of technology transfer. Many universities have created graduate programs in computational science and engineering to meet the demand for HPC skill-sets. Current status: Salary support for researchers is provided through their home institutions while research support is provided through alternate channels such as NSERC, SSHRC, and CIHR, for example. Almost 200 Canada Research Chairs currently benefit from the regional consortia HPC resources. World-class scientists are being attracted to live and work in Canada. As a result of the synergies between HPC and these complementary programs, the brain drain of our top scientists has largely been stopped [4]. Availability of HPC Support Personnel Still Insufficient The development of highly qualified personnel constitutes the most effective form of technology transfer. Effective HPC infrastructure is much more than hardware. The smooth operation of distributed facilities requires highly qualified personnel to manage, operate and maintain these facilities. The skills and experience required are extensive and increasingly specialized, requiring systems administration and operations personnel; programmer/analysts; applications programmers in multiple disciplines; data management and visualization; security; and management and administration. Current status: The availability of HPC support staff is still insufficient. The international HPC experts committee emphasized the paramount importance of the human resources strategy in the Compute/Calcul Canada proposal. Funding targets for people in the Long Range Plan was set at $13 million per year in 2006 rising to $17 million per year by Both CFI and NSERC have recently responded to this need. NSERC has responded with an award of $2 million per year for five years as part of a new Major Resources Support program, addressing roughly 15% of the levels called for in the Long Range Plan. A proposal to NSERC is now being prepared by Compute/Calcul Canada to increase this to $5 million per year, potentially raising support for people to 38% of the target. CFI has awarded $18 million in infrastructure operating funds for the operation of

13 December, 2007 Building Canada s Future Research and Innovation Culture 7 the proposed facilities (30% of the CFI capital funding of $60 million). However, these funds typically go to support operating costs (like electricity) rather than people. Additionally, all host institutions make a contribution to the operating costs of their local facilities, providing one or more of space, power, technical support personnel, management personnel, supplies and/or cash. Imperative 4: Founding a National Body for HPC The current academic HPC consortia are providing a highly effective model for regional coordination and sharing but, as HPC becomes more prevalent, the importance of effective national coordination assumes central significance. To take on this essential leadership and national coordination role, the establishment of a new national body for HPC Compute/Calcul Canada is now being implemented. Compute/Calcul Canada will provide national leadership, advocacy, national coordination of activities, represent Canadian HPC nationally and internationally, perform outreach to underserved communities, industry, and the public, advise Canadian research funding agencies on HPC issues, provide overall accountability, and conducting the annual High Performance Computing Systems conference. Current status: The regional consortia are establishing Compute/Calcul Canada as the national body for HPC in Canada. It is expected operations will begin early in Governance and national resource allocation policies have been drafted. At this time, staff have yet to be hired, initial goals and objectives are still to be defined, and programs have yet to be developed or delivered. The Board of C3.ca Association has firmly and enthusiastically committed its support for Compute/Calcul Canada, recognizing that HPC in Canada must continue to evolve to meet the needs of the Canadian research community. Imperative 5: Collaborating with CANARIE to Build Canada s Emerging Cyber-Infrastructure Over the past decade in Canada, a new infrastructure has been built to support a critical transformation of the research enterprise. Continuing investment in technological innovation is the critical first part of a comprehensive strategy to maintain Canadian competitiveness in HPC. Over time, such sustained investments will lead to an integrated intelligent or cyber infrastructure that will provide the research community with the essential tools (fast networks, fast computers, and massive data repositories, for example) needed for world-class work. The linkage between these two aspects of a comprehensive Canadian strategy for sustainable research competitiveness technological innovation andorganizational innovation is especially close, since building and using intelligent infrastructure requires new ways of collaborating. Also essential is the second part of the strategy: continuing investment in how Canadian and international research

14 8 Building Canada s Future Research and Innovation Culture December, 2007 communities organize themselves. In Canada, a consortium approach has been taken to address this challenge, resulting in a set of robust, multi-disciplinary multisectoral collaborations built around strong institutional building blocks. The linkage between these two aspects of a comprehensive Canadian strategy for sustainable research competitiveness technological innovation and organizational innovation is especially close, since building and using intelligent infrastructure requires new ways of collaborating. New arrangements must be created and new architectures designed to govern how this new research infrastructure is to be shared and jointly controlled. Grid Computing Transforming HPC HPC by itself can be the object of research and innovation. Within this area, grid computing has become an extremely active area of computer science research. Today, grids enable a number of large scale collaborative research projects that would not have been possible a decade ago. The major issues addressed by grid research include security mechanisms to enable controlled cooperation across institutional boundaries, creation of open standards for interoperability, and automation tools to address the scalability challenge. Broadband networks enable grids by supporting and integrating a rich array of services and hosting diverse distributed resources. Within this context, three important grid objectives must be addressed: 1) matching demand for resources to appropriate and available supplies of resources; 2) allowing data distributed anywhere to be accessed and processed by arbitrary computing agents located anywhere; and 3) enabling remote collaboration bringing researchers together via high-quality audio and video links, thereby building critical mass virtual communities of purpose from among a geographically dispersed research population. Current status: Canada s particle physics community has been in the forefront in driving the development of computing grids to become production facilities both domestically and internationally. WestGrid has been grid enabled since Currently, there is no national grid initiative. One of the barriers to implementing a national grid infrastructure is that the needs of computational scientists (traditional scientists who need high capability and/or high capacity computing to conduct their research) are distinct from the needs of computer scientists (who conduct R&D into computing itself; for example, grid computing). The needs of these different communities must be balanced as part of a comprehensive strategy in Canada. There is a continuing need for computational scientists and computer scientists to collaborate in the development of scalable algorithms and programming languages to support high performance computing. The International HPC Experts Committee recommends a strategy be developed for integrating all consortia systems into a national grid environment [5]. National leadership by Compute/Calcul Canada will be needed to address this recommendation.

15 December, 2007 Building Canada s Future Research and Innovation Culture 9 CANARIE Provides Essential National Cyber-Infrastructure The success of the regional consortia model would be muted if it were not for the advanced national next-generation network provided by CANARIE. 4 CANARIE provides the national backbone network that is so essential for HPC based research collaborations. The CANARIE Network interconnects the optical regional advanced networks and, through them, the universities, research centres, and government laboratories, both Sustained investments will lead to an integrated intelligent or cyber infrastructure (of which HPC is a part) that will provide the research community with the essential tools needed for worldclass work. with each other and with international peer networks. The objective is to provide researchers access to remote facilities with the equivalent of local access, allow sharing of large volumes of data, and create virtual communities of practice amongst geographically dispersed experts. Current Status: In March 2007, Industry Canada funded CANARIE for an additional five years. A total of $30 million will fund CANARIE s Network-Enabled Platforms Program, which encourages the definition and implementation of collaborative projects that will accelerate the development of national and international cyberinfrastructure and e-research platforms. The regional HPC consortia are eligible to participate in this program. Imperative 6: Establish an Internationally Competitive Tier One Facility The most complex and computationally demanding applications require the greatest HPC capability: the largest and highest performance system available. Such a facility would address the most pressing group of applications facing society today the so-called grand challenges. (Examples include weather prediction and global climate modeling, decoding the human genome, designing pharmaceutical drugs, symbolic computations, and computing the fundamental nature of matter.) Due to the rapid evolution of technology, it is more convenient to classify a system that would place in the top thirty of HPC systems worldwide as Tier One. The threshold for a Tier One Facility in November 2007 is currently more than five times more powerful than the largest system in Canada [3]. Each year the performance of these systems approximately doubles (from November 2006 to November 2007, the threshold for a Tier One system increased from 15.2 Teraflops to 35.9 Teraflops). None of the regional consortia has been Sharing of resources between able to attain a global ranking in the top tier Compute/Calcul Canada and of HPC facilities in the world (Canada s top government agencies could increase facility, at the Université de Sherbrooke, is the efficiencies of investing in a currently ranked #391, delivering only 1.4% Tier One Facility. of the performance of the #1 ranking). Canada has never had a Tier One facility dedicated to research. A Tier One Facility is beyond the capability of any one university 4 CANARIE is the Canadian Network for the Advancement of Research, Industry and Education.

16 10 Building Canada s Future Research and Innovation Culture December, 2007 or regional consortia to sustain. As CFI funding for the regional consortia provides only 40% of the total funding, these funds must be leveraged using provincial funding channels and other sources. Therefore, the majority of funding supporting HPC is regional funding and decisions regarding HPC expenditures have (for the most part) addressed regional priorities and constraints. Establishing a Tier One Facility requires a new policy framework by the Government of Canada. Supporting a Tier One Facility requires national funding that allows 100% of the funds to be employed to support it. Instead of the 40/60 leveraging formula underlying CFI funding, a different leveraging mechanism could be to encourage sharing of a Tier One Facility between Compute/Calcul Canada (representing academia) and government agencies (representing the National Research Council, Environment Canada, Department of Defence, Natural Resources Canada, and Fisheries and Oceans, for example). In the same sense that resource sharing increases the efficiencies of investing in HPC resources within the regional consortia, the sharing of resources between Compute Canada and government agencies could increase the efficiencies of investing in a Tier One Facility. A new government policy context is needed to establish Canada among the top tier of international HPC facilities. This initiative could also provide a critical migration path for Compute/Calcul Canada researchers seeking to become internationally competitive by increasing the complexity (and societal impact) of the research questions that can be addressed. Current status: No funding is currently available for a Tier One Facility. A group of pre-eminent Canadian researchers needing a Tier One Facility has been identified as the champions for a Tier One Facility. Discussions between Compute/Calcul Canada and Environment Canada have been initiated to conduct pilot projects using upgraded facilities at Environment Canada. Benefits include providing Canadian academic researchers with a migration path to higher performance while forging a closer working relationship between academic and government researchers. Imperative 7: Establish Sustained and Robust Funding for HPC in Canada Overall funding requirements for a national HPC infrastructure are presented in detail in the Long Range Plan s budget and only summarized here [1]. Funding requirements are shown starting in 2006 (year one of a seven year plan). Total public funding required is $76 million per year in 2006, rising to $97 million per year in These funds do not include industry and university contributions. This budget has the following components, in roughly decreasing levels of priority: Capital investment for the regional HPC consortia requires $44 million per year in 2006, rising to $54 million per year by (This goal can only be met through a second round of the CFI s National Platforms Fund, and this must be sustained through ongoing competitions.) Facility costs (e.g. power) are estimated initially at $8 million per year in 2006, rising to $10 million per year by (This goal can only be met through a second round of the CFI s Infrastructure Operating Fund (30% of CFI s NPF

17 December, 2007 Building Canada s Future Research and Innovation Culture 11 funding), and sustained through ongoing competitions.) Funding for people is set at $13 million per year in 2006, rising to $17 million per year in (15% of this goal has been met by NSERC funding of $2 million per year for five years. A proposal to increase this to $5 million per year is being prepared by Compute/Calcul Canada.) A Tier One Facility requires $10 million in capital investments in 2006, rising to $14 million per year by (No funding is available.) Finally, the national body requires $1 million per year beginning in 2006, rising to $2 million per year by (33% of this goal has been proposed for Compute/ Calcul Canada s initial budget.) Under the umbrella of the C3.ca Association (and now Compute/ Calcul Canada), the HPC community has become sufficiently large and sufficiently focussed in its message to be heard by policy makers. Traditionally, sustainability of research funding follows from policies established by the Government of Canada towards funding of granting agencies and councils. Individual researchers traditionally had little voice or influence on the processes that directly impacted them. Under the umbrella of the C3.ca Association (and now Compute/Calcul Canada), the HPC community has become sufficiently large and sufficiently focussed in its message to be heard by policy makers. With this increased input into public policy comes added responsibility. The research community itself must also look beyond grant funding as the only strategy for sustainability. Building a sustainable research and innovation environment that responds to global economic and societal change also requires corresponding changes in how research and innovation is done in Canada. The Imperatives outlined in this document are key elements of a sustainable HPC initiative, but sustainability of Canada s HPC infrastructure also depends strategically upon: A focus on research excellence Enhanced training of highly qualified personnel as the key to economic competitive advantage Strengthened partnerships amongst over 50 institutions, plus numerous industrial and research institute partnerships Instituting a national governance model for Compute/Calcul Canada that empowers researchers, that addresses the numbers and diversity of the research community, that respects the geographical breadth of the regional consortia, is sensitive to balancing the interests of institutions, regional consortia and national views, while also being responsive to the political and economic realities of provincial and federal government priorities Maintaining a long range plan defining the research priorities of the HPC community and the strategy for achieving these priorities

18 12 Building Canada s Future Research and Innovation Culture December, 2007 Current status: The Canada Foundation for Innovation has provided sustained funding support for the regional consortia in each of the competitions it has held since NSERC has contributed support for operations and personnel. Under the National Platform Fund in 2007, four regional consortia will obtain a badly needed technology refresh. The needs of the remaining consortia have been deferred to a future round of CFI funding. There is no assurance there will be a second round of national platform funding. A second round is crucial to fully integrate the remaining consortia within this national infrastructure and thereby meet the Imperatives of this plan. Imperative 8: Demonstrating Benefits for Canada Given the substantial mandate of Compute/Calcul Canada and the investment needed to establish and operate this national initiative, a regular review of its operations and role as part of a full review of Canadian HPC investment is recommended. This long-term funding initiative will need to be re-evaluated every five years, perhaps as part of a national review of Canada s international research competitiveness. The following performance evaluation criteria should be considered: a) Academic excellence (publications, awards, retention and recruitment of faculty); b) Qualified personnel produced (graduate students, postdoctoral fellows, research associates, technical analysts and support personnel); c) Societal/economic impacts (patents, spin-offs, industrial partnerships, technology transfers, improved infrastructure, health outcomes, outreach activities); and d) Effect of the investments in HPC on Canada s international competitiveness. The Economic Impact of HPC is Huge At Bombardier, we firmly believe our competitive edge is critically dependent upon use of, and access to, HPC. [Fassi Kafyeke, Manager, Advanced Aerodynamics, Bombardier Aerospace, Montréal] The past decade has seen HPC technology embraced by industry. The number of companies on the Top 500 List has grown from 3% in 1994 to over 50% today. HPC has created many new industries, including search engines (Google), internet service provision, environmental modeling (weather prediction and global climate models), coupled structural and computational fluid dynamics modeling in aerospace, oil and gas seismic processing and reservoir modeling, and computer graphics rendering and animation in both film and commercial arts. The economic impact of HPC is huge. The following examples are a small fraction

19 December, 2007 Building Canada s Future Research and Innovation Culture 13 of the examples and cases provided by the Long Range Plan [1] and Compute/Calcul Canada proposal [4]: Genomics is an area where technology transfer is accelerated by the use of HPC, as is the case with the Montréal company Genizon. In association with CLUMEQ, Genizon has analyzed data from five Whole Genome Association Studies and pinpointed up to 12 of the genes that cause Crohn s disease, an affliction of the bowel compared with two genes that were previously known. In collaboration with SHARCNET researcher Dr. Michael Bauer (The University of Western Ontario), the development of computational models for use in predicting the impact of hazardous incidents has allowed Keigan Systems to create the world s first ultra high-speed incident mapping and management application designed specifically for First Responders and Command teams at the municipal and commercial site level. One critical discovery used calculations within an HPC environment to enable the design of novel molecules for binding to beta-amyloid, a peptide involved in Alzheimer s disease. According to Dr. Donald Weaver, Canada Research Chair in Clinical Neuroscience at Dalhousie University, an effective drug for Alzheimer s disease or a single new antibiotic drug with widespread usefulness will be a billion dollar molecule. The Montréal-based biotechnology company Neurochem Inc. grew out of research using HPC conducted at Queen s University. The company is publicly traded (TSX: NRM) and now has a market capitalization of over $1 billion and more than 200 employees. Conclusion HPC funding in Canada must accommodate the need for both nationally shared mid-range and internationally competitive Tier One HPC infrastructure, investments in people and operations, and the establishment of a national body for HPC. Stable sustained funding is needed to allow for long-term planning, responding to evolving needs, accommodating demands from industry and encouraging the evolution of a coordinated approach. Canada must invest strategically and effectively in order to maintain our HPC based research strengths, and maximize the benefits of the investments already made. The development of Canadian computing platforms to permit collaboration across organizational boundaries and among distributed members of communities of practice will create a virtual organization that will: 1) enhance Canada s ability to develop unique solutions to complex problems; 2) allow researchers to capitalize on our existing expertise and network of infrastructure; and

20 14 Building Canada s Future Research and Innovation Culture December, ) maximize the benefits, both social and economic, that will be derived from the innovation and advanced research thus supported. Canada itself will not only benefit from the research outcomes enabled through this investment, but from the jobs and expertise that this investment produces. Canada must invest strategically and effectively in order to maintain our HPC based research strengths, and maximize the benefits of the investments already made. References 1. R.K. Rowe, J. Borwein, R. Boyd, G. Brunet, H. Couchman, A. Evans, M. Guest, I. Lancashire, J. Schaeffer, HPC*CHP Engines of Discovery: The 21 st Century Revolution, C3.ca Association Inc., August For more information, see 3. For more information, see 4. H. Couchman, R. Deupree, K. Edgecombe, W. Habashi, R. Peltier, J. Schaeffer, D. Sénéchal, Compute Canada Calcul Canada: A Proposal to the Canada Foundation for Innovation National Platforms Fund, Canada Foundation for Innovation, HPC expert committee review of project #12866, National Platform Fund, High Performance Computing, Industry Canada, Mobilizing Science and Technology to Canada s Advantage, Policy Branch, Science and Innovation Sector, Industry Canada, (see Acknowledgements This document is based on previously published materials the 2005 Long Range Plan (Engines of Discovery: the 21 st Century Revolution) and the 2006 Compute Canada Proposal. Acknowledgement is due to the original authors for the compelling content of these documents and its elegant expression. Dr. Andrew Bjerring s comments from the Foreword of the LRP have been incorporated in the body of this document and are gratefully acknowledged. There are also the anonymous contributions of researchers across the country who contributed case examples and research summaries, some of which are included here. Finally, the editor would like to thank all the reviewers of this document: Andrew Bjerring, Jon Borwein, Hugh Couchman, Peter Graham, Ian Lancashire, Normand Mousseau, Andrew Pollard, Jonathan Schaeffer, Randy Sobie, James Wadsley, and Andrew Woodsworth. (A.J. Stacey, December 2007)