India’s $700 million Supercomputer to be Unveiled by Year End

Article By : Sufia Tippu

December targeted by government and National Supercomputing Mission

BENGALURU — Finally, after three years there is some news about it. India is expected to have its first supercomputer by December this year.

Three years ago, in March 2015, the Indian government has approved a seven-year National Supercomputing Mission (NSM) supercomputing program which will comprise a supercomputing grid of 70 geographically-distributed high-performance computing centers linked over a high-speed network.

This was estimated to cost about $700 million (₹4,500-crore).

But until March this year, just a measly sum of $30 million had been disbursed, which had made the going pretty tough.

The entire project was shrouded in a deathly silence – not because of any top secrecy element attached to it but simply because no one seemed to know anything about it. When questioned about the progress of the project, government sources would say on record that it was work in progress but off the record there would be non-committal shrugs because no one really knew where it was heading.

But recently it was reported in The Economic Times that India is expected to have its first supercomputer by December. The first phase of the NSM is focused on assembling the supercomputers while the build element will be part of the second phase. If things stay on track, IIT-Kharagpur will have a 1.3 petaflop machine and IISER Pune and IIT-BHU will have a 650 teraflop computer each by the year-end, according to the report.

The Centre for Development of Advanced Computing (C-DAC) is evaluating the technical bids by firms such as Acer, Fujitsu, IBM, HCL, TCS, Dell and Netweb and the contract is likely to be awarded soon.

“The project is not just building a supercomputer but also applications and that work hasn’t stopped. All other parts of the process are on track,” Hemant Darbari, director general, C-DAC, said in the report.

C-DAC, along with the Indian Institute of Science (IISc) in Bengaluru, is spearheading this project which was announced with much fanfare in 2015.

As per the initial plan, while some components would be imported, some, like server-board assemblies, cooling solutions, power supply and storage systems would be manufactured in India with an aim to make 50% of the components locally over time.

The slow climb up the supercomputing chain

Sometimes, when it comes to the history of supercomputing in India one tends to be sceptical about it given the tardy and sluggish progress it had made over the past few years.

India’s supercomputer program was started in late 1980s because Cray supercomputers could not be imported into India due to an arms embargo imposed on India, as it was a dual-use technology and could be used for developing nuclear weapons.

Due to this technology-denial move, India set up the Centre for Development of Advanced Computing (C-DAC) in March 1988 with the clear mandate to develop an indigenous supercomputer to meet high-speed computational needs in solving scientific and other developmental problems where fast number crunching is a major component.

From the late 80s we have been hearing of Indian supercomputers, starting with PARAM 8000, considered to be India’s “first supercomputer”. It was indigenously built in 1991 by the Centre for Development of Advanced Computing (C-DAC) and was replicated and installed at ICAD Moscow in 1991 under Russian collaboration.

Then came Aaditya from the Indian Institute of Tropical Meteorology, Mumbai. It had a theoretical peak of 790.7 teraflop/s which is used for climate research and operational forecasting. It ranked 96th on the June 2013 list of the world’s top 500 supercomputers.

This was followed by the supercomputer SahasraT from the Indian Institute of Science – Bangalore (IISc.) in 2015. The Supercomputer Education and Research Centre (SERC) of IISc inaugurated the Cray XC40 petaflop supercomputer, christened as SahasraT had nothing to do with the NSM which had just been announced. It was an independent effort from SERC who was spear-headed by Professor N Balakrishnan and his team.

Today, SERC is the country’s leading computing centre having the state-of-the art computing facilities of high performance computing for scientific and engineering research. The supercomputing facility is a symbiosis of computing, network, graphics, and visualization. It has sophisticated software packages which work in a functionally distributed supercomputing environment connected by a powerful high- speed network and is working together with C-DAC.

But after a long gap, it was only in January this year that Pratyush, a Cray XC40 system – an array of computers that can deliver a peak power of 6.8 petaflops – was launched by the Indian Institute of Tropical Meteorology (IITM)

According to a statement by IITM, Pratyush is the fourth fastest supercomputer in the world dedicated for weather and climate research, after Japan, US, and the UK. More important, this development would move an Indian supercomputer from the 300s to the 30s ranking in the Top500 list, an international tracker of the world’s fastest supercomputers.

Challenges that waylaid the project

But as said earlier, it wasn’t easy to even make this snail-like progress.

Almost every single step had been dogged by inordinate delays and according to industry experts there were three key reasons for this slow start.

One, there is no designated team leader or a dedicated person heading it; two, lack of adequate funding; and three, no co-ordination between the various government departments involved in this program.

The NSM, announced three years ago with an intent to propel India higher in the global supercomputer rankings, literally seemed to be a non-starter. None of the government sources would make any concrete announcements about the progress and it was rumoured that Param series was a dying breed of supercomputers and there was nothing on the Indian high performance computing (HPC) horizon to replace it.

Until the January announcement of Pratyush – a 10-petaflop machine where a key function of the machine’s computing power would be monsoon forecasting using a dynamic model.

Professor S. Sadagopan, director, International Institute of Information Technology, Bengaluru, (IIIT-B) who has been closely following this project was quite optimistic about this project and felt that a three-year delay “was actually not bad, given traditionally these projects used to take much longer.”

C-DAC was set up with a mission to focus on supercomputing, but over the years it has been asked to do too many other things, which has impacted its performance.

“Moreover, getting good talent is a challenge for all of us. Most of the brainpower comes from the Indian Institutes of Technology (IITs) IIITs and (National Institutes of Technology) (NITs. And today, it is a herculean task for government organizations such as C-DAC to compete with research grants from multinational companies such as Google which are more attractive to the scientists,” Sadagopan pointed out.

Leave a comment