Subtle shift seen in Indian supercomputing mission

24th July 2017
Subtle shift seen in Indian supercomputing mission

Task split into 3 phases starting with import of 6  machines, followed by  indigenous manufacture of  switches and networks.  Full Indian  manufacture  relegated to last phase.

Bangalore, July 24 2017: India's National Supercomputing Mission -- the ambitious  Rs 4500 crore project to make India a leading supercomputer nation -- seems to have been subtly tweaked, to embrace near term goals, where  'buy for India' replaces 'Make in India'.
|The mission was first launched by the Manmohan Singh government in  2012 with Rs 50 billion (Rs 5000 crore)  in the kitty  "to build national capacity and capability in supercomputing".
The plan was very similar to --  and indeed may have been derived from --  a scheme  for India to  achieve exascale capability submitted to the  Prime Minister's  Science Advisory Council   by Dr Vijay Bhatkar former Director General of the Centre for Development of Advanced Computing (CDAC) and currently  VBice Chancellor, Nalanda University,  who  is known to be a strong proponent of India building its own supercomputer.  
Not much progress was made for the next few years and the new government headed by Narendra Modi,  dusted and relaunched the supercomputing mission in 2015, resetting the clock with a new a 7 year time line and  a budget that had by then shrunk  to Rs 45 billion ( Rs 4500 crores). The Centre for Development of Advanced Computing (C-DAC) and the Indian Institute of Science (IISc), Bangalore  were designated as the  twin operational hubs of the project. 
Now, two years after that,  there is some clarity about how  the Mission intends to  move forward -- but it also reveals a major re cast of priorities.   Developing and building an indigenous system is now no longer a priority: it has been relegated to the third and last phase of the 7-year programme ( 5 years remaining).
Prior to that India will shop in the global market and buy 6 supercomputers.  It will begin by importing three  ready to use systems and  another three will be manufactured abroad but assembled in India  -- under the supervision of C-DAC.
PTI quotes the secretary, Ministry of Science and Technology to  say that these six machines are intended for the IITs at   BHU  Varanasi, Kanpur, Kharagpur and Hyderabad, the Indian Institute of Science Education and Research Pune  and the Indian Institute of Science, Bangalore.  Only two of these machines will be of  2 petaflop capacity.
IISC is already home to a Cray XC40 system  rated at 901.5 teraflop  -- at the Supercomputing Education and Research Centre( SERC). It is the fastest of four  India-based supercomputers  ranked among the world's  500 fastest machines -- all of them imported.
Indigenous manufacture will be undertaken in the second phase but that too will be restricted to  some select sub assemblies like  switches, nodes and networks. 
The game plan revealed by  the Ministry of Science and Technology,  appears to indicate a rethinking of the original goal  of  the National Supercomputing Mission to create national capacity in supercomputing.  And indeed this is a reversal of a national endeavour that goes back almost 3 decades when   C-DAC launched  the Param supercomputer series of  indigenous machines  with Param 8000, whose latest avatar is the   388 teraflop Param Yuva  II.
(Jargon: A teraflop is  one trillion computer operations per second