India is today home to 23 supercomputers — powerful computers that are primarily used for scientific and engineering work that demands ultra high-speed computations.
Although indigenous development of supercomputers began in 1980 — with the involvement of organisations such as BARC, C-DAC and C-DOT, among others — it was the launch of the National Supercomputing Mission (NSM) in 2015 that accelerated efforts in a big way.
“Compared to five or 10 years ago, India’s supercomputing journey has been quite successful,” says Faisal Ahmad, co-founder and CEO of BIS Research, a market intelligence and advisory firm focusing on deep technology. “Until 2016, India had only four supercomputers.”
The superfast machines are in use in the field of computational chemistry, material science, quantum mechanics, and more, with nearly 5,000 users executing close to 8 lakh jobs on them.
On the flip side, however, use of supercomputers is currently limited to research institutions.
In terms of supercomputing capacity, India has made commendable progress, though it still lags other leading nations.
As of June 2022, China boasted a staggering 173 of the world’s 500 most powerful supercomputers, while the US had 128. From India, only three systems — Param Siddhi AI (ranked 111), Pratyush (132), and Mihir (249) — made it to the list.
“Our focus should be to equip supercomputers with better facilities, rather than achieving a global rank,” says Rupesh Nasre, Faculty, Department of Computer Science and Engineering, IIT-Madras.
While exascale computing — involving billions of computations per second — is evolving rapidly, India has no exascale supercomputers yet. “We are still looking at petaflops [quadrillion flops, where a ‘flop’ — or floating point operations per second — is a measure of computer performance]. But we should be looking at exaflops now, because the world is already there,” says Nikhil Malhotra, Chief Innovation Officer, Tech Mahindra.
- Also read: In-memory computing. A chip off the AI block
The Indian government has initiated efforts to develop indigenous exascale computing capabilities through NSM by 2024. Does this delay signify an oversight on the part of the country?
Ahmad says, “Param-Shankh, India’s new indigenous exascale supercomputing monster from C-DAC, is set to launch in 2024. Thus, India has not ignored the exascale revolution. Under the NSM scheme, C-DAC is aiming to install 70 supercomputers pan India.”
Beyond research and academia, however, supercomputers have found limited adoption in industry. Lack of awareness is one reason for this. “Both industry and research institutes need to come together to solve this challenge,” says Malhotra.
India currently lacks the infrastructure to produce the semiconductor devices required for the development of supercomputers.
Moreover, due to friendly trade agreements, companies such as Intel, Qualcomm, Nvidia, and others have access to Indian markets, adds Ahmad.
While the country still relies on imports for some components, indigenisation efforts are on, too.
“We have seen in recent times that India has started building the hardware required for supercomputers, but it is still early days and would need a push,” says Devroop Dhar, co-founder of advisory firm Primus Partners.
As per government data, India’s network of research institutions, in collaboration with industry, is scaling up the technology and manufacturing capability to make more and more parts in India.
In phase I of NSM, 30 per cent of value addition was done in India, and this has been scaled up to 40 per cent in phase II. India has developed an indigenous server, Rudra, which can meet the high-performance computing needs of government bodies and public sector undertakings.
- Also read: Do AI machines have feelings and rights?
The quantum jump
Supercomputers continue to be highly relevant even in the era of quantum computing. While quantum computers possess immense potential for certain types of calculations, supercomputers excel at tackling a broader range of complex problems.
Malhotra says the link between supercomputers and quantum computers is hierarchical.
“If supercomputers don’t yield the desired effect, I’ll explore the option of quantum computers. It’s a hierarchical scale that I can leverage,” he says.