US supercomputers have been the world's most powerful since the first high-performance machines analyzed virtual nuclear blasts, climate change and the makeup of the universe.
Now, one built in Japan with an "old" design runs five times faster than the previous record holder, a machine that simulates nuclear tests at Lawrence Livermore National Laboratory.
PHOTO: AP
Japan's Earth Simulator supercomputer hasn't quite rattled the US like the Soviet Union's launch of Sputnik in 1957. But it does highlight some drawbacks of recent US machines -- and it has made more than a few scientists envious.
"This machine is powerful enough that a researcher who uses it can do in one day what it takes a researcher in the US to do in one month," said Jack Dongarra, a University of Tennessee professor who tracks the world's 500 speediest computers.
More than national pride is at issue. Certain research is better suited to machines like the Earth Simulator, whose design was abandoned by most US manufacturers in favor of one that melds better with the rest of their computing businesses.
Supercomputers are built with thousands of processors that work in tandem to analyze the most complex issues -- including nuclear test simulations, aircraft designs, drug creation and others -- for governments, research centers and corporations.
Livermore's ASCI White supercomputer, like most recent US-built machines, used off-the-shelf processors rather than custom parts specifically geared for high-performance scientific jobs.
The trend took off in the 1990s, as such processors grew increasingly powerful, making it difficult to justify the cost of developing chips just for the small scientific supercomputer market.
Now, supercomputers like ASCI White use the same chips -- but thousands more of them -- as do servers sold to businesses.
Such machines now make up nearly 92 percent of the top 500 supercomputers worldwide. In 1993, they made up only about 27 percent.
But some say the move away from custom processors places business concerns ahead of scientific needs.
"The arguments are all based on strange economic theories -- none of them are based on technical grounds," said Burton Smith, chief scientist at Cray Inc, which still sells custom supercomputers.
NEC-built Earth Simulator, which will be used in climate and earthquake studies, is faster than all 15 of the biggest supercomputers in the US combined, Dongarra said.
It performs 35.9 trillion calculations a second with 5,104 processors. ASCI White, by contrast, performs 7.2 trillion calculations a second with its 8,192 processors.
"US scientists want to use it," said Kiyoshi Otsuka, leader of the Earth Simulator's research exchange group in Yokohama, Japan.
The high-performance title isn't expected to stay in Japan forever. IBM, which built ASCI White, Hewlett-Packard Co and other US supercomputer makers say they are working on even more powerful machines.
"We could do that in a heartbeat and we could do that for a lot less money," said Peter Ungaro, vice president of high performance computing at IBM, which sells supercomputers to foreign nuclear governments, research centers and corporations.
Officials at the Livermore Lab say their machine and others like it offer better price performance than those designed like the Japanese supercomputer. Earth Simulator cost about US$350 million, compared with ASCI White's US$110 million.
"For global climate, [Earth Simulator] is a great advance," said David Nowak, the ASCI program leader at Livermore. "It's just a very expensive machine."
But supercomputers built with off-the-shelf parts can be more expensive in the long run, said Guy Robinson, research liaison at the Arctic Region Supercomputing Center, where scientists study everything from climate change to how galaxies form.
Price isn't the only factor, however. Scientists say certain research problems work better or are easier to program on one supercomputer type over the other.
Custom supercomputers, for instance, have bigger data pipes, known as memory bandwidth, which is critical in climate modeling and some nuclear research.
That can be a drawback for off-the-shelf processors.
"In one sense, we have a tiny straw for data," Dongarra said. "And we have the processor, which would like a fire hose of data to come at it."
But off-the-shelf machines work best on data analysis such as in genetic research. It's a case of finding the best fit for the research -- and budget, said Fran Berman, director of the San Diego Supercomputer Center.
"From a political and financial perspective, we're not in an environment where we could build all these different kinds of tools with equal emphasis," she said.
J. David Neelin, atmospheric scientist at the University of California, Los Angeles, said computing is ultimately just another tool, like a laptop is for a student assigned to write a paper.
"Are they going to get that essay out any faster or is it going to be any better? Well, their spell checker is going to run a little faster," he said. "In the end, it's the thought that goes into it that really makes the essay."
WASHINGTON’S INCENTIVES: The CHIPS Act set aside US$39 billion in direct grants to persuade the world’s top semiconductor companies to make chips on US soil The US plans to award more than US$6 billion to Samsung Electronics Co, helping the chipmaker expand beyond a project in Texas it has already announced, people familiar with the matter said. The money from the 2022 CHIPS and Science Act would be one of several major awards that the US Department of Commerce is expected to announce in the coming weeks, including a grant of more than US$5 billion to Samsung’s rival, Taiwan Semiconductor Manufacturing Co (TSMC, 台積電), people familiar with the plans said. The people spoke on condition of anonymity in advance of the official announcements. The federal funding for
HIGH DEMAND: The firm has strong capabilities of providing key components including liquid cooling technology needed for AI servers, chairman Young Liu said Hon Hai Precision Industry Co (鴻海精密) yesterday revised its revenue outlook for this year to “significant” growth from a “neutral” view forecast five months ago, due to strong demand for artificial intelligence (AI) servers from cloud service providers. Hon Hai, a major assembler of iPhones that is also known as Foxconn, expects AI server revenues to soar more than 40 percent annually this year, chairman Young Liu (劉揚偉) told investors. The robust growth would uplift revenue contribution from AI servers to 40 percent of the company’s overall server revenue this year, from 30 percent last year, Liu said. In the three-year period
LONG HAUL: Largan Energy Materials’ TNO-based lithium-ion batteries are expected to charge in five minutes and last about 20 years, far surpassing conventional technology Largan Precision Co (大立光) has formed a joint venture with the Industrial Technology Research Institute (ITRI, 工研院) to produce fast-charging, long-life lithium-ion batteries for electric vehicles, mobile electronics and electric storage units, the camera lens supplier for Apple Inc’s iPhones said yesterday. Largan Energy Materials Co (萬溢能源材料), established in January, is developing high-energy, fast-charging, long-life lithium-ion batteries using titanium niobium oxide (TNO) anodes, it said. TNO-based batteries can be fully charged in five minutes and have a lifespan of 20 years, a major advantage over the two to four hours of charging time needed for conventional graphite-anode-based batteries, Largan said in a
Taiwan is one of the first countries to benefit from the artificial intelligence (AI) boom, but because that is largely down to a single company it also represents a risk, former Google Taiwan managing director Chien Lee-feng (簡立峰) said at an AI forum in Taipei yesterday. Speaking at the forum on how generative AI can generate possibilities for all walks of life, Chien said Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) — currently among the world’s 10 most-valuable companies due to continued optimism about AI — ensures Taiwan is one of the economies to benefit most from AI. “This is because AI is