• Thu
  • Dec 18, 2014
  • Updated: 4:13am

China's world-beating supercomputer fails to impress some potential clients

The Tianhe-2 has again topped the global rankings, but a lack of investment on software limits its use, researchers say

PUBLISHED : Monday, 30 June, 2014, 4:25am
UPDATED : Monday, 30 June, 2014, 4:25am

The mainland's billion-yuan supercomputer might be the most powerful in the world, but some researchers say its benefit to them is limited by its high operating cost and a lack of software.

Tianhe-2 last week held onto its first-place ranking in the Top 500 charts, which measures the capacity of the world's supercomputers. It performed at a sustained 33.86 petaflops, or quadrillions of calculations per second.

But all that muscle is not translating into practical use, some potential clients say. One problem is cost. The electricity bill for the machine at the Sun Yat-sen University campus in Guangzhou runs between 400,000 yuan (HK$485,400) and 600,000 yuan a day, which ultimately falls to the user.

Another obstacle remains a lack of software, they say. Tianhe-2 has been used for railway design, earthquake simulation, astrophysics and genetic studies. But so far investment has focused on hardware, forcing clients to write the programmes to allow them to use it.

Some voiced their criticisms yesterday in a meeting with Guangzhou's deputy mayor, Wang Dong.

"It is at the world's frontier in terms of calculation capacity, but the function of the supercomputer is still way behind the ones in the US and Japan," said Chi Xuebin, deputy director of the Computer Network and Information Centre under the Chinese Academy of Sciences.

"It's like a giant with a super body but without the software to support its thinking soul," Chi said. Some users would need years or even a decade to write the necessary code, he added.

Tinahe-2 was developed by the National University of Defence Technology at a cost of 2.5 billion yuan, and began trial operations in April. It has so far served 120 clients at 34 per cent of its capacity.

Chen Dongmin, a professor at the Academy for Advanced Interdisciplinary Studies at Peking University, also voiced concerns over the cost to use Tianhe-2. "It should be truly applicable to fields beneficial to society," Chen said.

Chen urged the city government to come up with ways to attract international and high-profile users rather than focusing on covering the supercomputer's bills.

According to Professor Yuan Xuefeng, director of the National Supercomputer Centre that manages Tianhe-2 in Guangzhou, the machine takes one second to finish what 1.3 billion personal computers would need 1,000 years to complete.

Yuan said they were aware of Tianhe-2's shortcomings. "The bill might be large at first but calculations by Tianhe-2 can drastically shorten years of experiment time," Yuan said.

The centre would seek to cultivate young talent in software programming, Yuan said.

Wu Hequan, a fellow at the Chinese Academy of Engineering, said last month that China's lack of "core technology" posed a potential cybersecurity risk as it still relied heavily on imported silicon chips.


For unlimited access to:

SCMP.com SCMP Tablet Edition SCMP Mobile Edition 10-year news archive



This article is now closed to comments

"It's like a giant with a super body but without the software to support its thinking soul," Chi said."
China in a nutshell.
As usual, they have the hardware but not software. Everything in china is like that: exterior is so nice and grand, interior is just shxt, they use the cheapest things when you cannot see it, plus but they don't know how to use/operate it... Like the poor services in "nice" restaurants...
one step at a time fellows! the software will come! we should congratulate the Chinese for their accomplishment. Now, bring down the power use and increase the software function!
If they have the talent to build a supercomputer that can be No. 1 in the world, they'll have the talent to build the software too. Just a matter of time, money & priority. Belittling it only exposes your own intellectual incompetence. WT
The Tianhe-2 is obviously an engineering marvel which a lot of countries will want but cannot have because they don't have the capability. As for the software, supercomputing isn't something easily done. But further investments and research will solve the problem. If there are no gaps there can be no advance. But in any case a very good tool for China to have! Congratulations China. All that's need is to get the good software too.
Haha you're the exact opposite: full of hateful words but no actions to back it up! Keep hating on the Internet forums, you sad **** and keep collecting those 50 cents!
These reader comments are disappointing, since they betray a fascination with building large computers. In reality, you can achieve whatever position you want on Top500 - it only depends on how much you want to pay. Many machines have some novel engineering aspects, but they're all fundamentally very similar - most use exactly the same components.
The article itself is somewhat interesting, and demonstrates that building large computers is basically a publicity stunt, not driven by concrete demand for computation. The issue is not so much about whether novel software is required to use a large machine, but rather that there is little demand - important enough to *push* - for high-petaflop computing. Many fields *could* utilize the power, but few would run efficiently, and few could justify the cost.
Stakhanov Lives!
Thank goodness they didn't use this to do the hacking attacks on the Unofficial Referendum votes, the numbers alone would've physically wiped Hong Kong off the map.
Yes, congratulations! Tick....$600,000....tick....$600,000....tick.....$600,0000.....


SCMP.com Account