Mumbai: Vijay Pandurang Bhatkar, one of India’s most acclaimed scientists, is best known as the architect of India’s first supercomputer—PARAM 8000. The PARAM series of supercomputers have been designed and assembled by the Pune-based Centre for Development of Advanced Computing (C-DAC), of which Bhatkar was the founder executive director. He is also credited with the creation of several national institutions, including the Electronics Research and Development Centre (ER&DC) in Thiruvananthapuram, the ETH Research Laboratory in Pune, the International Institute of Information Technology (I2IT), also in Pune, and the India International Multiversity. With the help of this Pune-based varsity, Bhatkar aims to resurrect India’s ancient ‘Gurukul’ system of learning that originated in the Vedic times.
Bhatkar won the Padma Shri in 2000. He will also be presented the Sitaram Jindal Foundation Award-2012 on 3 December in New Delhi for his work in science, technology and environment. The award, which includes Rs.1 crore in prize money, was given to former Indian president A.P.J Abdul Kalam in 2011. In a telephone interview, Bhatkar discussed the future of supercomputing in India and his current research interests. Edited excerpts:
India’s Eka, ranked the world’s fourth largest supercomputer in November 2007, was ranked 129 in the June 2012 Top 500 Supercomputers list that is published twice in a year. Has India missed the high-performance computing (HPC) or supercomputing bus (Eka has been built by Computational Research Laboratories Ltd (CRL), now a subsidiary of Tata Consultancy Services Ltd)?
India does not lack the technology. The capacity you want to build depends a lot on the budget that is available with you. For instance, it will take Rs.300-500 crore to build a petaflop (1,000 trillion flops, or floating point operations, or calculations per second) supercomputer today. India is in the process of building petaflop supercomputers. (The supercomputer from the CSIR Centre for Mathematical Modelling and Computer Simulation, or C-MMACS, was ranked 58 in the June 2012 list, and the institute is one among those in the process of building more powerful supercomputers by 2017). These are primarily research and scientific projects that are funded by governments. In India, these kinds of budgets were not available. Eka met with success since it was a private initiative by the Tata group.
With the government announcing Rs.5,000 crore for supercomputing in the 12th Five-Year Plan, are things set to change?
Yes. It will significantly change things. In this context, I had proposed to the government that we move from petascale supercomputers to exascale (one quintillion flops) ones. A project of this scale requires massive government support. On 3 January, during the Indian Science Congress, Prime Minister Manmohan Singh approved the move in principle. From the 1990s, computing power has been increasing by a factor of 1,000.
In the next 10 years, the objective is to reach exascale but building these machines will require a new set of technologies. Current technologies will not work since they will require gigawatts (GW) of power. Currently, a petascale machine requires around 5 megawatts (MW). With our current technology, an exascale machine will devour around 5,000MW. This is just not sustainable. Current scientific understanding is that the power requirement should be brought down to around 20GW for an exascale supercomputer and, hence, the need to change the technology.
However, over a period of 10 years, even with the new technologies, India will require around Rs.11,000 crore to build these exascale machines over the next 10 years. This will require several breakthroughs, including the development of new architectures for multi-core chips (with thousands of cores on a chip). A billion nodes will need to be connected, for which we will require a new network, new software, and new applications will need to be built.
But we will do well to remember that yesterday’s supercomputers are today’s laptops and tomorrow’s supercomputers will be commoditized. These exascale supercomputers will become tomorrow’s machines for everyday use.
Supercomputers, combined with artificial intelligence, are gradually emulating the human brain. But will they ever match the power of the brain?
Supercomputers are being used to understand the brain even, in some cases, by reverse engineering it. We can simulate the brain using supercomputers but the brain remains a mystery. As I see it, many functions of the brain will be mimicked by supercomputers but the brain will continue to remain a mystery for a long time to come.
What are supercomputers primarily used for in India?
Supercomputers are primarily used for weather forecasting, which requires a lot of computing power. They are also being used for oil exploration for companies like the Oil and Natural Gas Corp. Ltd and Indian Oil Corp. Ltd. Climate modelling to detect trends like global warming is another area. Supercomputers are also needed for space programmes, nuclear reaction simulations, bio-technology and gene sequencing and a whole range of scientific applications (highly calculation-intensive tasks such as problems involving quantum physics, weather forecasting, climate research, molecular modelling and physical simulations). All these applications are connected by C-DAC on the national knowledge network (NKN) and use the grid computing model to fire the applications from anywhere while combining the computing power from these different groups. This has been taking place since 2005 when the grid was introduced.
Do we have enough researchers and scientists for this job?
We need a large number of researchers with doctorates. The number needs to be increased manifold—tens of thousands of skilled people to work on applications, in mathematics, algorithms, software, modelling, etc.
You work on supercomputing and also with the education-to-home, or the ETH Research Lab. How do these activities complement each other?
This is where the Jindal award recognition makes sense to me. I received the award for applying science and technology at the grassroots level. So you have things like weather forecasting, predicting disasters like tsunamis, etc. accurately. National resources mapping for agriculture is another area. Also, in 1987-88, when the supercomputer was being built at C-DAC, my concern was that supercomputers were being used only in English and European languages. So we, at C-DAC, initiated moves to have supercomputing in Indian languages and the mission was accomplished. Another irony is that while India is being recognized as an information technology powerhouse, around 25 crore people do not know how to read or write. So my efforts are concentrated on providing the benefits of ICT (information and communications technology) to the masses.
As a scientist, you now also talk about the melding of science and spirituality...
From an Indian perspective, right from Vedic times, we have not differentiated between these two knowledge systems—spirituality and science. Both these systems, in my view, attempt to understand reality in the universe. One (spirituality) explores the inner world while the other (science) explores the outer world. The inspiration was quantum physics. Both these paths can help us get a complete understanding of reality since they complement each other. Take, for instance, climate change. Reckless use of technology can harm the world we live in. We need a lot of discerning power to use technology, and spirituality could help us here. We are also trying to design courses on these lines.