Public ICT policy

Home » Public ICT policy

What are Supercomputers used for?

Posted on July 6, 2010 in HPC, Innovation, Public ICT policy by ronvanholst

So if I expect people to vote for my proposal of a National Canadian Supercomputer centre, I should provide some examples of what we would do with such a machine.  Certainly not a comprehensive list, but a sampling of some interesting articles that I have read recently.

Health Sciences

Ivanov_cycleFig2We all hope technology will bring us breakthroughs in the prevention and treatment of disease.  Supercomputing lends a hand in many ways.  Here is an example of some basic research in the human proteome enabled by a supercomputer shared by universities in Ottawa and Kingston.  This type of basic research also known as bioinformatics, will accelerate the discovery on new cures.  There’s a small Canadian company, that is developing GPU based solutions for bioinformatics.  Genome research, which has similarities to proteome research, is being used to find cures for cancer.  Canada is participating in a global effort to share genome based cancer research.  This is an interesting article on the simulation of DNA repair mechanisms.  Gene research becomes a data mining exercise.

Product Design

Batteries are a foundational technology for many products and of special interest in new automobile design; this is a nice write up on how supercomputers are being used for better battery design.

Oil & Gas

The oil and gas industry uses supercomputer simulation for exploration (processing seismic data) and production (reservoir simulation), but this story talks about a new computer at the University of Regina that is being used for research into greener methods for petroleum processing, even the manufacturer of the machine is Canadian.

Space

ACT_TelescopeAlthough supercomputers are usually thought of for solving computationally intensive algorithms, this article talks about the processing needs of massive amounts of data for space research.  Canada also uses supercomputers for processing telescope data.

Environment

Weather simulation has been run on supercomputers for decades, but more recently, climate simulation is another growing area for supercomputer research.  Canada should play a leading role in modeling the changes in arctic climate.  This article makes reference to chemistry research using high performance computing for CO2 capture, and important component of climate modelling.  There is an interesting animation of an ocean current simulation that is attempting to predict the spread of the BP oil spill.  It has been suggested that disaster scenarios like this should be simulated as part of the due diligence for all new ocean oil drilling platforms.  With all the interest in drilling in the arctic, it would be in Canada’s best interest to not only model disaster scenarios for Canadian rigs in the arctic, but also the oil drilling sites of our arctic neighbours as well.

Canadian contributions to Supercomputing technology

Canada’s biggest supercomputer belongs to the SciNet consortium. Although this CBC article written last year states that it would be in the 15 machines in the world, it actually ranked 22nd when it first came online, and when the list was updated last month, it dropped to 28th.

I wrote a blog post on an innovative supercomputer built in Quebec.  Although this is a truly innovative supercomputer design, it is currently only ranked 72nd.

This article mentions Canada’s contribution to a multi-national effort to develop exascale computing software, although it is not specific on what Canadian institutions are participating in this development.

Although a CATA article from last year, it paints a sad picture for companies that develop new technology in Canada; the case in point here is a Canadian company that was developing a supercomputer system, of which I have written about in another post.  This company did not survive, but I wonder if it would have if it could have had a GoC reference account a year earlier than it did.

Back to the Future

Posted on March 18, 2010 in HPC, Innovation, Public ICT policy, social media by ronvanholst

I’m oDoc Brown Pours Beer into Mr. Fusion in Back to the Futureverdue for a blog post.  I’ve been very busy following up on some interesting opportunities and have also been “making my mark” with comments on a new Canadian blog based news service. This is a great site for Canadians wishing to make their mark by commenting on posts relevant to their expertise. One of my comments got a bit long and I noticed the posting lost my paragraph breaks, so I thought I would edit it a bit and elaborate further for my blog.

Remember “Mr. Fusion” from the movie “Back to the Future”?  Well the promise of cheap limitless energy from nuclear fusion has been “almost here” for a long time.

I’ve been reading articles on nuclear fusion since I was in high school.  In hind site it’s pretty funny, a couple of nerdy high school kids talking about nuclear fusion after reading about it in a Popular Science magazine they found in the library (PopSci have made their archives available on line BTW, so cool).  Thirty years later, I can’t honestly say that I know much more about nuclear fusion, but an article on the subject is just as temping now as then, so when I saw one in the Mark, I was hoping to discover a Canadian connection.  Failing to find one, I provided one of my own in my comments recaptured below:

Check out the National Ignition Facility home page, it has a great video of how this fusion reaction will be achieved.

I was hoping to see a Canadian angle in the above article on the subject of nuclear fusion. I don’ t know of any, but I’ll tell you about a Canadian connection that might have been.

So how is such a system designed with any confidence that it will work? More importantly, how can they be sure the reaction will not get out of control venting a nuclear cloud over suburban Livermore, California? The answer is supercomputing (often called High Performance Computing or HPC).

The Livermore facility is home to some of the largest supercomputers ever built by the US Department of Energy. When nuclear testing was banned, facilities like the Lawrence Livermore National Laboratory were built to turn the task of nuclear research over to computer simulation. To accurately simulate nuclear reactions in a computer requires machines with thousands of identical processors running in perfect synchronization. In March 2005 LLNL built Blue Gene/L with IBM the first supercomputer to exceed 100 TeraFLOPs of performance. But at LLNL, as soon as one machine is built, the plans for the next one begin.

A small Canadian start-up company from Ottawa was invited to propose a computer architecture for a PetaFLOP machine. This proposal was delivered along with a small (refrigerator sized) prototype supercomputer built in Kanata, Ontario by former Nortel engineers. Although the prototype machine was installed at LLNL, and early prototypes achieved many of the performance targets set by LLNL, a system was not purchased. That’s the Canadian connection to nuclear fusion that might have been.

Unfortunately the dream of a Canadian designed and built supercomputer died with the start-up that built them a few weeks ago when they ran out of funds. This company was able to sell a few systems for non-supercomputing tasks, but it was never able to win a supercomputing sale. I often wonder if the Canadian government had purchased a supercomputer from this little company, if it would have been enough to bring it to self-sustainability. Unfortunately our government does not have procurement vehicles to help along start-ups like this. Although Canada may not be directly involved in solving the science and engineering of nuclear fusion, there are other such “grand challenge” problems that are being solved by supercomputing technology.

Canada can have a role if it chooses to invest in supercomputing technology. We toss a bit of money for universities to build some nice systems to support academic research, but there is no national strategy to advance supercomputing technology for industry in Canada. It should be a component of Canada’s Digital Economy Strategy, it is for practically all other advanced nations. Maybe energy comes too easy for Canada, with clean hydro, lots of fossil fuels, and lots of uranium; the relative comfort of the present doesn’t impel us to invest as strongly in the future as we should. “Mr. Fusion” will not likely be invented in Canada, but if some of the new game changing technologies of the future are not invented here, Canada will cease to be one of the best places in the world to live.

Go for Gold

Posted on January 26, 2010 in Innovation, Public ICT policy, social media by ronvanholst

winter_2010-vancouver-olympics-medals-gold-silver-bronze

Maclean’s published a great article on Canada’s preparation for the Olympics, using innovative technology and a goal of being number 1, likening the secrecy behind these advancements as a new cold war (especially fitting for a winter Olympics).

We all look forward to seeing our athletes on the podium soon and I’m hoping that will inspire us as a nation to drive to be most innovative nation in the world.  Right now the Conference Board of Canada ranks our performance in Innovation a “D”, we’re ranked 13th worldwide in innovation; 6th among the G-8 countries.  An OECD survey ranked Canada 14th in R&D spending relative to GDP.  As exciting as it is for us as a nation to be cheering for our athletes to be first and win gold, especially as a host nation,  our standard of living in Canada depends directly on innovation and we can choose to aim for the gold there as well, if we focus on it.  We can be the most innovative nation in the world if we’re willing to invest in research and innovation.

The Minister of Industry, Tony Clement, promised a High Tech Plan to promote business innovation by year end, and we’re still waiting.  It looks like leadership in this area will come from the private sector, through a local technology hero Sir Terrance Matthews who is in addition to all his corporate accomplishments, spokesperson for CATA.  Every world class athlete needs a world class coach and mentor and we’re fortunate to have such an accomplished technology entrepreneur speak up for innovation in Canada.  CATA president John Reid is tapping into social media to get feedback on Mr. Matthew’s comments about a structural deficit in innovation; he is specifically quoted as saying, “While we recognize that there have been a number of distractions and that significant discussions are occurring on whether a structural budget deficit exists, everyone agrees that Canada has a structural deficit in innovation that continues to impede growth in many of our communities across the country.”

A structural deficit means that spending > revenue over the long term. This term normally applied to public sector deficits, is used here to draw attention to Canada’s Innovation Gap.  Thus, the supposition is that Canada is systematically spending more on innovation than the revenues or benefits that we derive from it.

I agree with this supposition.  Not because Canadians are not innovative, but because others benefit more from our innovations than we do.

For example, think of all the intellectual property developed by Nortel and Bell-Northern Research, most of it in Canada supported by GoC research tax credits. With the demise of a “star athlete” in Canadian R&D, most of this innovation has now been auctioned off to foreign ownership for pennies on the dollar. Also consider the many start-ups launched here, although some will inevitably fail, of the ones who succeed, most are bought by foreign companies; thus practically all of the wealth generation potential leaves Canada. Think of it this way; let’s say we design the best hockey skate in the world that gives players a significant competitive advantage, but then only team USA gets to use them in the Olympics; it is like this all too often in the world of technology start-ups.  Thankfully there are exceptions to this such as RIM, our new star athlete in R&D and global competitiveness, but these exceptions are too few.

I personally believe that Canada needs to spend more public and private funds on innovation, but the bigger challenge is to find ways of nurturing the wealth generation opportunities to maturity at home. We used to talk about the brain drain to the US, but the real problem is the drain of home grown wealth generating ideas that are just starting to bear fruit. The brains can often be encouraged to come home, bringing more experience and wealth on the return trip, but it is much harder to bring back the profits from our ideas once they’ve crossed the border.

There are a lot of good discussions about improving programs like IRAP and SR&ED, but there would be even greater benefit if we had “Buy Canadian” policies in government. Naturally we can’t disregard trade agreements, but many start-ups would have a much greater chance to grow organically at home if they could get a little preferential treatment with government procurement processes. Certainly the majority of government spending should be focused on best value for money spent, but a portion reserved for higher risk purchases in support of Canadian innovators would go a long way in reducing our innovation deficit.

If we can support our athletes to go for gold, let’s support small business and start-ups in Canada to also go for gold, to be the best in the world at what they do.  It’s not just about national pride, but the standard of living for the next generation of Canadians depends on it.

A cheap place to build a big data centre?

Posted on November 24, 2009 in HPC, Public ICT policy by ronvanholst

There has been much talk about Iceland as a cost effective place to build large data centres, due to the access of cheap and reliable electricity, cool air, and an economy eager for new business opportunities.  A limiting factor would be fibre communications, some undersea cables have been built to Europe, but this is a limited resource.

Why not make Canada a destination for such large systems?  We have a better fibre infrastructure and close proximity to the world’s largest consumer economy.  Americans don’t have to be sold on the availability of cold air, as their meteorologists frequently remind them of this.  As for cheap power, although I’m not an expert, I would expect there are places where it can be had reliably and economically, perhaps even places where we have a surplus.  I’m thinking of the Chalk River facility up the Ottawa valley, where we have nuclear power from test reactors, but probably not much in the way of local demand for power (especially in an area where hydro electricity is easy to exploit).  You also should have infrastructure and expertise to support advanced technology.  Although our economy probably hasn’t been hit as hard as Iceland’s proportionately, as a nation we do need to make investments in forward looking businesses like better data centres and advanced computing facilities which improve Canadian competitiveness globally.

Update:

I found another good article on data centre location. It mentions several things like community incentives, local construction costs, cost of electricity, but also the need to avoid areas like airport flight paths, and proximity to volatile industries such as an explosives manufacturing site.  I suppose a nuclear test reactor wouldn’t be viewed so favourably for that type of analysis.

Race to PetaFLOPs

Posted on November 5, 2009 in HPC, Public ICT policy by ronvanholst

200px-Top500_logoSo far, only the US has a few systems in the PetaFLOP range as of June 2009, the list will be updated at SC09 in a couple of weeks.

As nations position themselves for technological leadership, they will aim for a top 10 machine, Germany has two for now.

China will be staking a run for a spot in the top 10 in the next list with their “Tianhe” machine described in the insideHPC article.

Canada’s best showing is currently a U of T machine built by IBM that comes in at 16th for the June list, I’ll be watching how much it drops in the next list.  So when will Canada have a PetaFLOP machine?  Should we have at top 10 supercomputer available for Canadian research?  What would we do with such a machine?  Climate research?  Biotechnology research? Geophysical research?