IBM Warns of China Closing the Supercomputer Gap 238
eldavojohn writes "China is digging a massive hole to house a computer building with the intent of usurping the United States' lead in the field of supercomputing, claims IBM. As of earlier this year, Oak Ridge Lab was beating China's Shenzhen Center. But now, an IBM representative has said to a Washington, DC forum, 'You have sovereign nations making material investments of a tremendous magnitude to basically eat our lunch, eat our collective lunch.' China has long been a contender in this regard, and Europe and Japan have similar goals to build an exascale supercomputer. To achieve this by 2020, the US will need to focus on 'co-design,' where hardware is developed in tandem with every other aspect of the computer, from applications down to optics. This isn't the first time a 'space race' style supercomputing push has been spurred by international competitiveness."
To compute what? (Score:4, Insightful)
So everyone's trying to make a big, fast computer.
What's at stake? What does the winner win?
Plus ca change.... (Score:2, Insightful)
Apparently there is nothing new under the Sun. The reader of this PR to help IBM sell more of their HPC machines should read http://en.wikipedia.org/wiki/Missile_gap [wikipedia.org] first.
Government Bailout for IBM (Score:3, Insightful)
In other words, IBM wants the government to give them lots of cash so they can ship more jobs over to India.
Re:To compute what? (Score:5, Insightful)
So everyone's trying to make a big, fast computer.
What's at stake?
Bragging rights? China beats IBM, we can no longer say that we're the most technologically advanced country and that's what I want. If that happens, maybe we'll get a boost in science education like post-Sputnik.
What does the winner win?
The best and brightest immigrants?
Re:"My supercomputer is bigger than yours!" (Score:2, Insightful)
I understand IBM did some nice marketing in the "HPC is for Chess" area, but "surprise!": real world HPC is not used for chess playing.
It's for serious research, nowadays mostly nukes (design stuff to go BOOM) and flow modelation (climate research, stealth research, building better cars,planes and other machines), biochemistry (genetic engineering), cryptography and probably dozens of others things.
If they are worried... (Score:5, Insightful)
Since when does IBM care about the U.S.? (Score:4, Insightful)
Really, since when does IBM care about what happens in the U.S.? Aren't they the same company that recently told some of their top researchers that they could either move to China, Poland, or a couple of other countries on their own dime and work for 'local wages' or be out of a job?
Re:To compute what? (Score:2, Insightful)
The biggest, baddest nuclear bombs, the deadliest engineered plagues, the shortest cryptography decryption times and the goodwill of all mankind (everybody is very very nice to you if you have the aforementioned weapons).
Re:Government Bailout for IBM (Score:3, Insightful)
Why is it a bailout? Why do people just generically call things they don't like a bailout? A bailout implies just a handout to keep something from failing. This doesn't sound like a bailout. This sounds like a an investment, and for that money, we'll get a more powerful supercomputer and the knowledge and research and know-how that comes with it.
Re:Since when does IBM care about the U.S.? (Score:5, Insightful)
PR: "But now, an IBM representative has said to a Washington, DC forum, 'You have sovereign nations making material investments of a tremendous magnitude to basically eat our lunch, eat our collective lunch.'"
Translation: "But now, an IBM lobbyist has said to a Washington, DC forum 'Other countries are doling out sweetheart contracts to manufacturers and designers of expensive computers. Give us a giant pile of money or the chinks win."
Re:Since when does IBM care about the U.S.? (Score:5, Insightful)
Re:Remember the 1980s? (Score:3, Insightful)
Re:eh (Score:3, Insightful)
(sigh)
Low taxes do NOT mean low government revenue. The US government pulled in more money AFTER Bush's tax cuts than any time in history. (unfortunately, they spent even more, but that's a different story)
From USA Today [usatoday.com], Feb 12, 2007:
The continued strong growth in revenues reflects the record profits corporations have been recording in recent years and low levels of unemployment, which means more Americans are working and paying taxes.
(It's amazing how short our memories are. All I hear about today is how bad the economy was under Bush, yet from 2003-2007, we were booming, but no one care remember anything more than 2 years back)
See: Laffer Curve.
IBM, helping China beat America (Score:5, Insightful)
This is the same company that sold most of it's commodity hardware business to Red China. The same company that's heavily investing in research... in China and India. The same company that continued to sell the Nazi's computing hardware used against allied forces and for managing the Holocaust via their their Brazilian unit. IBM has had a long history of selling out America in order to maximize profits.
Re:To compute what? (Score:4, Insightful)
The huge debts that sovereign nations tend to rack up trigger the same moral instincts that petty consumer debt does; but it isn't at all clear that they work anything like the same way, economically.
China: "Dear US, we are cashing in the giant pile of debt you owe us."
US: "Shucks, China, it looks like we spent all our money on increasingly elaborate pyramid schemes and shitty exurbs that nobody wants. Anyway, thanks for all the free stuff over the years, and I hope you don't find the sudden transition from high-employment export economy to moderate-unemployment internal economy too jarring... TTLGTG!"
Re:What do they exepect? (Score:5, Insightful)
IBM is entitled to all the handouts they want - it's only the unemployed and welfare mothers that aren't entitled to handouts. Christ are you some kind of socialist?
Re:To compute what? (Score:5, Insightful)
IBM is not an American company. They've said so repeatedly, every time they been asked about all the thousands of jobs they've off-shored.
What they ARE, though, is a large multi-national trying to stir up fear and pseudo-patriotism in the hopes of snagging huge, profitable government contracts for projects to build things we really don't need right now.
Re:To compute what? (Score:5, Insightful)
But more importantly, there are reasons to have such big computers and it isn't bragging rights. That's what the managers use to measure by and get their bonuses, but the real value in a super is what you can calculate with it.
As they get more powerful, you can quit doing various approximations and do real calculations. Your simulations get more accurate. You can also do those simulations quicker and do more of them as you explore optimization strategies. While these computers are expensive, being able to trim years or decades off of research programs pays back many times more in first to market, market dominance, etc.
You can do things in simulations that aren't even practical or even feasible in real life. Depending on the problem, it can be that you cannot even simulate it at all without a computer of such a scale.
One thing to keep in mind is that the researchers who use these things, when they get more power, generally are able to make pretty amazing new discoveries. For anyone using these, the advantages are obvious. And so are the opportunites if they can get their hands on even more computing power.
The quest for more computing power is in no way simply a bragging rights kind of thing. There are huge advantages to being able to run on the fastest computer in the world.
Re:What do they exepect? (Score:1, Insightful)
But US smart people are studying law & finance. Their smart people are studying computers & engineering.
Re:To compute what? (Score:3, Insightful)
China beats IBM, we can no longer say that we're the most technologically advanced country and that's what I want. If that happens, maybe we'll get a boost in science education like post-Sputnik.
We don't need more science education (except maybe educating legislators). We need more science investment and employment to clear out the backlog of science postdocs that have been educated in numbers far in excess of jobs that require that sort of qualification.
Re:To compute what? (Score:5, Insightful)
The huge debts that sovereign nations tend to rack up trigger the same moral instincts that petty consumer debt does; but it isn't at all clear that they work anything like the same way, economically.
It should be clear and obvious they don't work the same way. After all, the US owes China in US dollars, not Euros, not RMB.
So it's more like an amusement park owing suppliers massive debts payable in amusement park tokens (except amusement park tokens cost more to make than "electronic" US dollars).
Or like you owing trillions in fuzzyfuzzyfungus dollars. You can create as many as you need. Sure the smart ones may never lend you money again, but maybe the smart ones wouldn't have lent you trillions payable in fuzzyfuzzyfungus dollars right? So the dumb ones might actually say "thank you!" when you go up to them and repay them :).
As long as the dollar remains the main currency used to trade oil and other commodities, the USA gets a cheap/free ride. The people who keep saying "the USA would be better off with the gold standard" should consider this and other important factors :).
Re:To compute what? (Score:5, Insightful)
Re:Nice marketing strategy (Score:4, Insightful)
Unilateral disarmement never works. If the other side wants to compete you can either compete or forfit. There are no other options. You might think you can choose to sit out, but there is no material difference between that and forfitting.
In theory, the USA has the greatest potential because (historically) you have had self-directed businesses that can rely on the stock market for capitialization and low taxes and low regulatory hurdles so a business can put incredible resources behind something that makes sense. There are plenty of examples of this happening where the government has stood aside while businesses gathered the capital and manpower to do really big things.
There was no government-financed support for transistors, lasers, integrated circuits or anything that led to the technology boom from 1950 to 1980 or so.
Where we are today is that everyone is looking to the government for direction and support. Solar power isn't practical on a large scale without massive government subsidies, so there are few businesses involved in this and some of the ones that are are pure scam. Electric cars might have a future, but there are so many regulations in place now that it is very difficult to manufacture anything involving those nasty things called "chemicals" that might get loose and destroy the environment - so other countries are building battery manufacturing plants and are fully prepared to sell the USA better, cheaper batteries while we fuss around. The result will be their batteries will always be better and cheaper.
The US Government is pretty much at the point of saying that China can have bigger, better, faster supercomputers because we will have bigger, better and likely more ponderous social support programs. The result will be a continuing slide towards 30-40% unemployment (we're at 20% now) and everything being made outside the USA. Hopefully, there will be plenty of jobs parking cars for foreign executives who come here to dictate terms.
Re:To compute what? (Score:1, Insightful)
you did not give one SPECIFIC example of why we need the fastest machine on the planet, much less a cost/benefit tradeoff - If I invest x dollars in a marginally faster computer that will be out of date in 5years, could I get a better ROI some other way ?
Re:To compute what? (Score:3, Insightful)
OK, been reading through all this stuff and I think it's time to chime in. So far as I can tell, the boogey man here is that China will: a) build a more powerful Supercomputer than the most powerful one we have, b) build a bigger data center than ORNL to house supercomputers, or c) both. As you point out, there are lots of things you can do with HPC assets, and having them is good. Here's where things begin to break down in the doomsday scenario. ORNL is one of a dozen or so similar sites in the US.
None of them is *quite* as large as ORNL (right now.. they actually switch around which is the biggest and best fairly regularly), but any one of them would be the biggest HPC site in the world if ORNL went away tomorrow. If China builds a bigger site than ORNL, will it be bigger than ORNL *and* Lawrence Livermore combined? How about if we add in Argonne? The DOE alone has like 5 of the top ten performing computers in the world. Then there's all of the DoD sites. We don't have to have the *fastest* or *biggest* computer in the world in order to have *way more computing power* than anyone else.
Currently we have 43 (if I counted right) of the top 100 supercomputers in the world here in the US, and 8 of the top 10. While having the single biggest is nice, not one is even close to us in total HPC assets. If we stopped buying HPC assets completely, right now, in two or three years we *might* be getting to the point where someone else was close.
China is capitalism as motivation for the state... (Score:4, Insightful)
Put another way, decades of observation of America taught China that you cannot depend upon "enlightened self-interest" or "their responsibility to shareholders" to keep humans motivated by greed on the high road, but if you shoot those who drift off the road you've chosen, you don't need sidewalks.
Put another way, China weaponized trade and used American greed against us. And now such as IBM wish to complain about the consequences of their eagerness to be fabulously wealthy victims? Who built Lenovo? Little green men from Mars?
Put another way, IBM whining about China investing in making them obsolete after a decade or two of IBM trying to make technology jobs in America obsolete is not American capitalism, it is American greed - and all that we have left.
Put another way, America herds cats, while China trains a tiger.