Why is the nation's premier institution for developing supercomputers, being used as a clearing agent for distributing imported systems?
INDIATECHONLINE C O M M E N T
Bangalore, November 23 2018: Elsewhere on this page we have carried a report from Paris, France, about the contract with Atos, makers of the BullSequana supercomputer and the Centre for Development of Advanced Computing headquartered in Pune to supply an unspecified number of the French systems for installation at India's leading academic and R&D institutions. ( Read our story here).
There is a sad irony in the agency that India's National Supercomputing Mission has selected, to procure and distribute these supercomputers. C-DAC has a special place in the history of India's indigenous development of supercomputers. It was the agency set up in 1988 to design and develop India's own supercomputer -- when the nation was denied the import of such high performance computing ( HPC) platforms by the US government. Under the leadership of its first director Dr Vijay Bhatkar, C-DAC delivered the first Indian supercomputer in the incredibly short span of 3 years -- Param 8000 -- created by stringing together 8000 pieces of the basic building block, the transputer, in a parallel computing configuration. The machine was benchmarked at 5 giga flops in 1991, making it among the world's fastest machines in the world at the time... a triumph of indigenisation in a high tech arena throttled by geopolitical restrictions.
Since then C-DAC has continued to develop a series of faster, more efficient Param supercomputers right up to 520 teraflop Param Yuva II in 2013-14. In subsequent years, indigenous supercomputers were largely out of the global Top 500 rankings of the world's fastest computing platforms.... though last year C-DAC did roll out an innovative "supercomputer in a box" called Param Shavak.
The agency was the natural choice for any new Indian supercomputing initiative. Yet bizarrely when the National Supercomputing Mission was relaunched in March 2015 by the present government at the centre, the stated purpose was subtly changed "to connect national academic and R&D institutions with a grid of over 70 high-performance computing facilities at an estimated cost of Rs 4,500 crore. The Cabinet Committee on Economic Affairs, chaired by Prime Minister Narendra Modi, approved the launch of the mission that will enable India to leapfrog to the league of world-class computing power nations. Rs 2,800 crore will come from the Ministry of Science and Technology and the rest, about Rs 1,700 crore, from the IT ministry"
Since then, the funds of the NCM have largely if not entirely, gone into the procurement of supercomputers from abroad: the two 3.7 petaflop Cray XC40 machines being used for weather forecasting and modelling ( see our story here) -- and now the order on Atos for an unstated number of Bull Sequana systems. Going by past annopuncements of NCM, there will be around 6 systems bought from Atos for IITs at BHU Varanasi, Kanpur, Kharagpur and Hyderabad, the Indian Institute of Science Education and Research Pune and the Indian Institute of Science, Bangalore. The first three will be outright imports by C-DAC, the next three will be assembled in India under C-DAC supervision.
There has been no mention of any allocation for C-DAC or any other Indian institution to develop or enhance an indigenous system. It is a sad descent for C-DAC, from pioneering the nation's first indigenous supercomputer -- and then improving the system with multiple iterations for over 30 years -- to acting as an agent to buy and redistribute imported systems. If not anything else, this is an outrageous and cynical misuse of the hard core of talent in the hardware and software of high performance computing systems that C-DAC has built up since 1988. It is also a good indication of how the those at the helm of the NCM today prioritize the spending of the Rs 4500 crore kitty at their disposal. It seems to be 'buy first -- then only ( try to) make'.
Today's global market and political environment, may allow us to shop for what we want -- supercomputing-wise. How long will it take for these geopolitical winds to change direction, at the next international flashpoint? US trade policy is already seeing the effects of having at the helm, a president whose policies are driven by old fashioned self interest rather than a nuanced world view. India has survived earlier spells of what used to be known as the Missile Technology Control Regime (MTCR) when dozens of devices and systems of possible strategic value were denied.
Who can say with any certainty that major supplier nations will not regress again into such trade-limiting regimes? If and when that happens, we will rue the day when those shaping our supercomputing policies decided that it was easier to be buyers than creators.