Despite living in the US for 25 years, Rajeeb Hazra has fond memories of the time he spent in Mumbai and Kolkata. Hazra’s tryst with the US began when he decided to pursue a doctorate in Computer Science.
Today, he is the Vice-President in the Intel Architecture Group and serves as a General Manager for the technical computing segment in the Datacenter and Connected Systems Group at Intel Corporation. Before assuming his current position, Hazra was the director of Supercomputing Architecture, where he was responsible for research and engaging with Intel's supercomputing customers.
Intel provides its technology for several home-grown supercomputers such as Param Yuva (developed by the Centre for Development of Advanced Computing) and EKA (from the Tata Stable).
In an interaction with Business Line, Hazra throws more light on how India could play a pivotal role in the supercomputing space by leveraging its expertise and talent pool. Edited excerpts:
Today, there are 11 supercomputers from India in the top 500, which is much better than what it used to be. Still, there is a long way to go…
We are in a field where you can be the 50th fastest supercomputer in the world today, but you will be soon retired, as you are making way for the second or third fastest. That is how quickly technology becomes obsolete in this space. India’s best days in high performance computing (HPC) are around the corner, given the tremendous industrial base the country has.
As part of your job, you interact with various arms of the Government to ensure continuity in investments related to high-performance computing (HPC). Could you throw some light on this, especially from an Indian standpoint?
We believe that governments can play a key role in making HPC ubiquitous for the masses. Our objectives are fairly simple when we call upon the government folks. We provide them insights from our experience in addition to keeping them advised on what is coming next from the perspective of capabilities and price points. While governments can fund the development of supercomputers, like it happens in the US and China, it is just as important for them to deploy it for the masses by making it a national infrastructure.
In India, we are deeply invested in creating inexpensive learning tools, developing indigenous applications and training people on HPC. We want to see India’s tremendous workforce accelerate the use of digital technology through the use of modelling and simulation techniques. We have partnerships with multiple educational institutions to develop turnkey appliances that can be used by small and medium businesses.
I think India has a great role to play in HPC given the sheer talent that it can offer in software, operating systems and even hardware architecture. India can use HPC to improve its industrial base, whether it is small and medium businesses or large corporate. India is well positioned not just as a consumer but also as a participant in building that capability for the future.
Are you working with the Government already?
We are in conversations with multiple parts of the Government. Again, our role is to provide data when asked and provide a perspective. We would love to continue that and build in India for India, what would be a wonderful and economically-viable supercomputing strategy.
Media reports have suggested that the Indian Government wants to produce the world’s fastest supercomputer by 2017. Do you think it’s a realistic ambition given the infrastructure issues and the power constraints the country faces?
I could answer that question in different ways, but let me answer it in two or three parts. Is it possible to have the world’s fastest supercomputer in a country? The answer is of course! It is just a question of priorities and funding.
At the end of the day, if the real goal is to have the world fastest supercomputer, you build a captive power plant and minimise all dependencies on external power.
In my view, the right question for India to ask is what will we do that will be unique with the world’s fastest supercomputer. A great deal of planning needs to happen for ensuring the presence of scalable applications (development focused) that will justify the huge investments in supercomputing.
We all know that high performance computing is an expensive affair. And you really need in a world market for getting a return on those investments. You need scale and if you cannot stay at the top of the heap, it is not a profitable business for too long. I hope that the way the problem is looked at in India, is which investments will have long term capability to uplift the people.
(The writer was recently in the US at the invitation of Intel Corp)