Supercomputers: India pitches for $1bn leap to make up for lost ground | Mail Online
India pitches for $1bn leap in supercomputers
Almost two and a half decades after entering the supercomputing race but losing ground thereafter, India is now rebooting with a vengeance. The government has discreetly embarked upon a billiondollar initiative to create next-generation supercomputers.
But unlike the `1,750 ($35) tablet PC, which is unfolding under the glare of high-voltage publicity, the supercomputing programme is shrouded in secrecy. So much so, the plan details have not yet been shared with scientists who have developed such mega machines in the past.
The government has committed `5,000 crore (nearly $1billion) for the plan, making it the largest ever grant for a single research programme since Independence. The money is likely to start flowing during the 12th five-year plan period.
The only jarring aspect of the project is that its reins are being handed over to the Indian Institute of Science (IISc), Bangalore, which has only been a user and not a designer or developer of supercomputers.
India had begun its foray in supercomputing nearly 25 years ago after the U.S. denied it a supercomputer meant for weather research. When the first Indian supercomputer, PARAM, was launched in 1991, India was only the third country in the world to possess supercomputing power after the U.S. and Japan. China joined the race 10 years later than India but surpassed it soon.
The current initiative, conceived in the Prime Minister's Office and nurtured by the Planning Commission, is a direct response to the Chinese challenge.
Currently, Indian supercomputers are at a teraflop stage. This means they can perform several trillion floating point operations per second. China, Japan and the U.S. have already achieved petaflop capability. Such supercomputers are able to perform a thousand trillion floating point operations per second.
These countries have also begun work on the next level - exascale - to develop machines with the capacity of an exaflop or one million teraflops.
In order to remain in the race, India would first need to develop a petaflop machine and then aim for exascale. When contacted, Dr T. Ramasami, secretary, department of science and technology, confirmed the development, revealing that the programme had been conceived with a view 'to position India strongly in supercomputing' and would be implemented in mission mode.
A detailed project report was ready and loose ends, if any, would be tied up soon.
Earlier this month, Prime Minister Manmohan Singh had announced at the Indian Science Congress at Bhubaneswar in Odisha that the IISc would spearhead the project.
'All knowledge institutions would be linked, but will be coordinated by the IISc,' Ramasami elaborated.
While the decision to rejoin the race could be a good strategic move, the choice of IISc as the leader of this ambitious programme has left experts baffled. The institute has never planned, designed and developed supercomputers in the past. In fact, its supercomputer centre is mostly focused on academic research and education. The Centre for Development of Advanced Computing (CDAC) - it has developed a series of supercomputers, including PARAM, and already prepared a roadmap for a petaflop machine - would strangely have to play second fiddle to the IISc.
Other scientific agencies such as those working in the fields of defence, atomic energy, space and aerospace labs, too, have proven expertise in supercomputer design and architecture. In the private sector, Eka - a supercomputer developed by Tata's Computational Research Laboratories - became the world's fourth most powerful system in 2007.
'We lost ground to late entrants such as China only because of inadequate investment and a fragmented approach to supercomputing. And when the government is ready to invest, it has decided to make a non-player the captain of the game,' a leading scientist, who has been involved in developing supercomputers, said.
The government, he added, had overlooked indigenous design expertise developed over the past 25 years.
'The IISc has its own strengths, but somehow it doesn't spring to mind as the first choice for the nodal agency,' Anand Parthasarathy, formerly a scientist in the supercomputing team of the Defence Research and Development Organisation (DRDO), felt. He is currently the editor of IndiaTechOnline.com.
The idea of bringing back India into the mainstream of high performance computing, he said, could be best implemented through a consortium of agencies such as CDAC, DRDO, BARC (Bhabha Atomic Research Centre) and Tata's CRL.
All of them have proven expertise in designing and developing supercomputers. 'They should come together in an "ego-less" public-private alliance to pool their strengths and deliver a truly Indian platform for exascale computing,' Parthasarathy added.
After China emerged a global leader in supercomputing in 2010, sources said the pioneer of CDAC supercomputers, Dr Vijay Bhatkar, had shared a blueprint on 'Building Exascale Supercomputing Capability' with government agencies.
The plan had mentioned an investment of Rs 500 crore to build a petascale facility in two years and Rs 5000 crore for reaching the exascale capacity by 2020.
Bhatkar, however, did not comment on the latest development.
'Right now, we are in the dark about what the PM's initiative is all about. No timelines, responsibilities and technological options have been revealed to anyone. Such a large scientific programme should be carried out in an open, transparent manner and should be peer-reviewed,' a Bangalore- based scientist said.