Another superflop in the making? This is the first thing that came to mind when I heard of telecom and information-technology minister Kapil Sibal’s supercomputing plans. For a man who has promised but failed to deliver the world’s cheapest computers over the last couple years, the grand plan to build one of the world’s fastest computers by 2017 sounds far fetched. Currently, India’s fastest supercomputer is at number 58 in the Top 500 list compiled twice a year. Number one on the list is IBM’s Sequoia. India plans to build one that is 61 times faster than Sequoia. Of course, by the time 2017 rolls around, there is no way to tell how fast the fastest supercomputer will be since the race is always on.
I saw a supercomputer for the first time in 2000, when a derailed train scrapped my holiday plans and instead of heading for the Andaman Islands, I ended up in the office of a supercomputing start-up in Bangalore. The machine looked like a sleek refrigerator and I was clueless about its capability. But I soon discovered that these machines can help us in ways that I had no reason to think of before.
Supercomputers are needed by a small set of scientists working on highly calculation-intensive tasks such as weather forecasting, molecular modelling, quantum physics and physical simulations including nuclear simulations. The animation industry, too, needs supercomputers for rendering or filling in of the characters. The faster the better to make progress in these areas and for quick turnaround in the animation business.
I remember being amused when Rajeev Gowda, a professor at the Indian Institute of Management, Bangalore, and co-founder and director of a supercomputing company, told The Economic Times newspaper in 2002 that “with the initiation of the CSIR-CDC linux partnership, we have started our mission of putting a supercomputer on every Indian scientist’s desk”. It will be interesting to know what happened to that project.
Getting back to the present, one thing we all know is that India’s supercomputers have not helped in getting the weather right. This year in particular, much has been said about the failure to predict the monsoon accurately. Faster machines may help, but plans to spend Rs.4,700 crore in five years to develop India’s next-generation supercomputers, need to be closely examined. Ongoing costs to maintain the systems will run into millions of dollars per year. A cost to benefit analysis should be made public before such amounts are allocated for this venture that could easily become the next scam to be born.
The funniest thing about all this is the discovery of the acronym DEITY for the department of electronics and information technology that is to oversee the implementation of this plan. I’m certain I’m not the only one wondering, “Why the Y?”