As I mentioned, I attended a lecture by one of the top minds at IBM at last week’s SuperComputer conference. It is always a pleasure to listen to a lecture by a person dripping with intelligence, as was the case here.
His lecture was pretty wide-ranging, and I’m not sure I can do it justice. It encompassed a big-picture view of economic trends, global competition, the next growth engine and the organization of society. Here were his key points:
-one of the key drivers for the next spurt of economic growth will be SuperComputing and its applications. Supercomputing has been around for a while, but it has mostly been applied in governmental and scientific applications. Then, the focus of the IT world was shifted to the internet and the dissemination of information. This internet revolution enabled globalization and the “earth is flat†phenomenon that Thomas Friedman has written about.
-the IT industry is now returning it focus to Supercomputing (there were 20,000 people at this conference) and its application in private industry. The application of Supercomputing to private industry will drive the next spurt of growth. As an example, one of the industries that has already adapted Supercomputing is the US oil and gas sector. With Supercomputing, they have been able to perform geological simulations which would have been unthinkable a few years ago. In doing so, they have been able to find new reserves in areas that were previously considered barren. Industries that are gearing up include automotive (cars designed and simulated before a single sheet of metal is cut), pharmaceuticals, bioengineering and genetic engineering.
-Today’s Supercomputers can run at 1 petaflop (1,000,000 billion calculations per second). To reach their true potential, they need to operate at an exaflop (one billion, billion calculations per second). Such a machine cannot be constructed with today’s economic structure, in which companies like IBM, HP and Cray do the system-level design, and then “farm-out†the sub-designs. At today’s level of technology, such a machine would consume about 10 megawatts (the energy consumption of a mid-size town) and its reliability would leave a mean time between failure of a few seconds – both unacceptable. In order to engineer such systems, we will have to go back to centralized, integrated organizations, like IBM of a few decades ago, or NASA for the Apollo program.
-In essence, he argued, we are going “back to the future†with this phenomenon. Supercomputers will not be designed and built on a global basis. Rather, nation-states will compete to set up the necessary centralized organization just discussed. That is why China just built their first “Top 10†supercomputer, to fire a warning shot that they understand the new game. In true China fashion, however, it was a cheap shot. Rather than assembling a Supercomputer from Intel-based CPU’s, they assembled them from Nvidia GPU’s (Graphics’ Processing Units) that are used on video games. It’s the equivalent of building a car with a humongous motor, and poor controls.
-The critical weakness of the US is at the component level. Our technology is not energy efficient. This was the challenge he laid before those assembled. As far as China, he was not optimistic. He reminded us that the most critical part of IT is the software, and software development required an environment that is supportive of intellectual property rights, out-of-the-box creative thinking, and collaboration: not strong suits of China.
If you are interested in following this field from an investment POV, I would advise that you keep an eye on two companies: Cray and an American-Israeli company called Mellanox. Most Supercomputers in the US fall into two categories: Cray’s unique architecture, or an open architecture based upon Mellanox’s switch (IBM and HP, for instance, use Mellanox’s switch). Moreover, Cray retains much of that centralized organization discussed, and their machines are taking an increasing share of the Top 500 Supercomputers.
Bxgrl, the idea is that you can stick a pin in an explosive situation by just being there and distracting everyone, best by acting like a good natured flake who talks a lot.
Green beans (esp. the slim haricots verts) w/ brown butter & slivered almonds make a nice veggie dish.
Bird’s nest for rent on bushwick rooftop:
http://ny.curbed.com/archives/2010/11/23/organic_architecture_for_rent_in_brooklyn_must_have_wings.php#more
As I mentioned, I attended a lecture by one of the top minds at IBM at last week’s SuperComputer conference. It is always a pleasure to listen to a lecture by a person dripping with intelligence, as was the case here.
His lecture was pretty wide-ranging, and I’m not sure I can do it justice. It encompassed a big-picture view of economic trends, global competition, the next growth engine and the organization of society. Here were his key points:
-one of the key drivers for the next spurt of economic growth will be SuperComputing and its applications. Supercomputing has been around for a while, but it has mostly been applied in governmental and scientific applications. Then, the focus of the IT world was shifted to the internet and the dissemination of information. This internet revolution enabled globalization and the “earth is flat†phenomenon that Thomas Friedman has written about.
-the IT industry is now returning it focus to Supercomputing (there were 20,000 people at this conference) and its application in private industry. The application of Supercomputing to private industry will drive the next spurt of growth. As an example, one of the industries that has already adapted Supercomputing is the US oil and gas sector. With Supercomputing, they have been able to perform geological simulations which would have been unthinkable a few years ago. In doing so, they have been able to find new reserves in areas that were previously considered barren. Industries that are gearing up include automotive (cars designed and simulated before a single sheet of metal is cut), pharmaceuticals, bioengineering and genetic engineering.
-Today’s Supercomputers can run at 1 petaflop (1,000,000 billion calculations per second). To reach their true potential, they need to operate at an exaflop (one billion, billion calculations per second). Such a machine cannot be constructed with today’s economic structure, in which companies like IBM, HP and Cray do the system-level design, and then “farm-out†the sub-designs. At today’s level of technology, such a machine would consume about 10 megawatts (the energy consumption of a mid-size town) and its reliability would leave a mean time between failure of a few seconds – both unacceptable. In order to engineer such systems, we will have to go back to centralized, integrated organizations, like IBM of a few decades ago, or NASA for the Apollo program.
-In essence, he argued, we are going “back to the future†with this phenomenon. Supercomputers will not be designed and built on a global basis. Rather, nation-states will compete to set up the necessary centralized organization just discussed. That is why China just built their first “Top 10†supercomputer, to fire a warning shot that they understand the new game. In true China fashion, however, it was a cheap shot. Rather than assembling a Supercomputer from Intel-based CPU’s, they assembled them from Nvidia GPU’s (Graphics’ Processing Units) that are used on video games. It’s the equivalent of building a car with a humongous motor, and poor controls.
-The critical weakness of the US is at the component level. Our technology is not energy efficient. This was the challenge he laid before those assembled. As far as China, he was not optimistic. He reminded us that the most critical part of IT is the software, and software development required an environment that is supportive of intellectual property rights, out-of-the-box creative thinking, and collaboration: not strong suits of China.
If you are interested in following this field from an investment POV, I would advise that you keep an eye on two companies: Cray and an American-Israeli company called Mellanox. Most Supercomputers in the US fall into two categories: Cray’s unique architecture, or an open architecture based upon Mellanox’s switch (IBM and HP, for instance, use Mellanox’s switch). Moreover, Cray retains much of that centralized organization discussed, and their machines are taking an increasing share of the Top 500 Supercomputers.
bxgrl, wife knows my feeding of eyes are harmless – 100% window shopping.
china is all lip service with NK.
Bxgrl, the idea is that you can stick a pin in an explosive situation by just being there and distracting everyone, best by acting like a good natured flake who talks a lot.
I just read that China is very unhappy with NK. Very unhappy.
And I’m sure, M4L, your wife would love that 🙂 For those who have never met Ms. M4L- she’s very lovely.
DH, if there were models, I would’ve asked you to get a photo for me
South Korean leaders are calling for retaliation.
Anybody have some popcorn?