Sibal’s Supercomputer Dream: Putting It in Perspective

So, Kapil Sibal has decided on his new obsession for the next few months. And this time, it is not a still cheaper tablet or, for that matter, a new mobile operating system to challenge Apple or Android. The lawyer-turned-minister has set his sight on nothing less than building the fastest supercomputer on earth. Sibal has reportedly written to the Prime Minister detailing a plan prepared by the state-owned research & development outfit, Centre for Advanced Computing (C-DAC) to achieve this feat by 2017, at a cost of some Rs 4,700 crores.

Critics may link it to the speculation of his possible removal from HRD ministry in the imminent cabinet reshuffle. But to be fair to Sibal, reshuffle or no reshuffle, he is never short of big ideas.

Kapil Sibal is a dreamer. That is a good thing. Few of the politicians at his age are. And we surely need a few of those dreamers.

But that also is his problem. He still has the hangover of his extremely successful past as a lawyer, and often has excessive confidence in his own ideas and abilities. So, even when his basic intention is laudable, it is seen as maverick-ism. While he has the ideas, the dreams, the passion, and a rare sincerity of approach, he lacks the vision to realize those dream. All his dreams, from Right to Education to low cost laptop, are low on vision. They lack a practical approach (that is they do not take into consideration the ground realities), but more importantly, they are not aligned with the shared vision of the government. So, depending  on who is talking, these ideas get dubbed as wishful thinking to megalomania, and a lot many things in between.

Take the Aakash tablet.  Using affordable technology to enhance education quality is a great dream. The government stepping in to help in whatever way possible to the private sector to make that happen is also a good approach. But why should the government align itself to a single brand? A single project? No one could explain this to Mr Sibal.

Now comes the supercomputer dream. While we don’t know the exact details of the “blueprint”, based on whatever media has reported, it already sounds flawed. Here is why

1. The focus is purely on speed. It is a petaflops speed supercomputer that the minister and C-DAC want to build. The application is secondary. While performance is not a bad objective to have, spending Rs 4700 crore to just be on top of the table sounds a little too much. I am still refraining from dubbing it megalomania. But will not quarrel with those who do.

2. Why do the vision and nuts-and-bolts have to come together? Why should it be assumed that C-DAC will build it? The same question was asked in case of Aakash. No bias against C-DAC. They have great capability. But for those interested in facts, though, C-DAC Param has not featured among the top 500 supercomputers in the last two years.  But should not a thrust on high performance computing through policy initiatives be a better approach to encourage the building of such supercomputers than adopting a project?

3. The focus is entirely on the speed of one supercomputer. What India needs is many such supercomputers in all aspects of our economy: oceanography to identity verification; drug research to weather forecasting. Just for the record, in the last six-monthly list (June 2012) of world’s 500 fastest supercomputers,  India had just 5 of them, up from 2 in November 2011 list. In contrast, China had 68. And we thought China just scores in physical infrastructure, India is the IT superpower!

In 2007, China was just slightly ahead of us. In November 2007, China had 10 of the world’s fastest supercomputers, while India had 9.  In June that year, China had 13, while India had 8.

See how we compare now.

At one time, China was just a little ahead of India. Now, it has overtaken Japan to be at No 2 position

China comparisons apart, India’s supercomputing journey has not been anything laudable as such.  Based on the Top 500 data, India’s share of fastest supercomputers in the list has not really increased. The average number of India’s supercomputers in the Top 500, between 2003 to 2007 was 6.8, with as many as 11 featuring in June 2006. Between 2008-2012, that average has come down to 4.4 per list, with highest being 8 in November 2008.

INDIA’S PERFORMANCE

MONTH NO OF SYTEMS IN THE LIST RMAX (TFLOPS) TOP RANK ORGANIZATION
Jun 2012 5 303.9 58 CSIR
Nov 2011 2 132.8 85 Tata Sons
Jun 2011 2 132.8 58 Tata Sons
Nov 2010 4 132.8 47 Tata Sons
Jun 2010 5 132.8 33 Tata Sons
Nov 2009 3 132.8 26 Tata Sons
Jun 2009 6 132.8 18 Tata Sons
Nov 2008 8 132.8 14 Tata Sons
Jun 2008 6 132.8 9 Tata Sons
Nov 2007 9 117.9 4 Tata Sons

Even the top performance has not seen any great improvement. In November 2007, the world’s fastest supercomputer was about four times faster than India’s fastest supercomputer. In June 2012, that ratio had increased to 54. Yes, the fastest supercomputer on earth was 54 times faster than the fastest in India.

Indian computing has not been able to keep pace with the world.

All this is not to suggest that Indians do not have capability to build a fast supercomputer or the dream to build a fast supercomputer is misplaced.
Far from that, India needs a lot of effort in high performance computing so that Indian companies and organizations can build many such supercomputers for application in all areas. The country badly needs that. And if the fastest one happens to be Indian, that would be an icing on the cake. The icing, however, should not be confused with the cake!
Advertisements

Leave a comment

Filed under Digital Economy, Technology & Society, Uncategorized

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s