The rise of artificial intelligence, machine learning and quantum computing, and the availability of huge amounts of data have necessitated the need for supercomputers more than ever before. The first supercomputer in the world was developed in America, called CDC 6600, developed by Control Data Corporation (CDC), by Seymour Cray in 1964. It was designed to handle complex calculations much faster than other computers of its time, achieving up to 3 million instructions per second.
Later, America advanced significantly in supercomputing with systems like the Cray-1, also designed by Seymour Cray in 1976. The Cray-1 became the benchmark for high-speed computing and cemented the US’s leadership in the field. Supercomputers enable cutting-edge innovation and research, especially in artificial intelligence (AI), quantum computing and other areas.
Processing large amounts of data at a speed of one petaflop can facilitate the discovery of new materials, such as catalysts critical for emission control, while contributing to solutions for other global challenges, like renewable energy and climate change. Other important uses supercomputers have been put to are defence applications, oil and gas exploration, genome sequencing, aviation engineering, and space technology.
India started working seriously on supercomputers in 1987 when America refused to sell Cray X-MP. The journey of Parallel Machine (PARAM) supercomputers, developed by the Centre for Development of Advanced Computing (C-DAC) in India, is a significant milestone in India’s high-performance computing (HPC) landscape. The series of supercomputers was launched in 1988 and has seen remarkable growth over the decades.
In 2015, the National Supercomputing Mission was launched in India to enhance research capabilities under the government’s vision of a ‘digital India’ initiative. It is jointly steered by the Department of Science and Technology and the Ministry of Electronics and Information Technology, supported by indigenous technology and powerful computing capabilities. This has opened new opportunities for research in astronomy, material science, and several other fields of physics and medicine.
Leading computing systems such as Fugaku from Japan and Summit of America are designed for large-scale applications such as AI research and simulations across various scientific disciplines. PARAM Rudra is highly cost-effective compared to these.
A supercomputer’s calculating speed is measured in petaflops. One petaflop is a thousand trillion flops, which is 1015 flops. This is used to measure the performance of the computer’s processor. PARAM Rudra computers have a processing speed of one petaflop. Achieving such a high processing speed was a feather in India’s cap, though it lags behind Fugaku and Summit-type computing systems, which have processing speeds of 442 and 200 petaflops, respectively.
Siddhi AI, launched in 2020, has a speed of 210 petaflops—the highest among all PARAM supercomputers. It focuses exclusively on AI. Better computational ability in astronomy will enable scientists to pay more attention to cosmic phenomena and comprehend the universe better. Supercomputing systems will also bring breakthroughs in material science and physics.
Next-generation supercomputers
Various public and private institutions in India have developed many other supercomputers. Pratyush and Mihir help with weather forecasting.
AI Research Analytics and Knowledge Dissemination Platform (AIRAWAT) was launched to advance research and development in artificial intelligence (AI) under the National AI Mission (AIM) and the National Supercomputing Mission (NSM). Installed at C-DAC, Pune, it was ranked 75th in the world in May 2023 in the Top 500 Global Supercomputing List at the International Supercomputing Conference in Germany.
On September 26, 2024, Prime Minister Narendra Modi dedicated the next generation of High-Performance Computing systems, Arka and Arunika, to the nation and PARAM Rudra supercomputer. PARAM Rudra system will be used in Pune, Delhi, and Kolkata. At Pune, the Giant Metre Radio Telescope will use it to study fast radio busts and other astronomical phenomena. Inter-University Accelerator Centre in Delhi will use it to strengthen material science and atomic physics research. S N Bose Centre at Kolkata will use it in Physics, Cosmology and Earth Sciences studies.
Arka and Arunika aim to increase predictability in adverse weather conditions, soil nutrient analysis, and processing other agriculture-related data, making them useful in agriculture and disaster management.
National Supercomputing Mission (NSM) aims to install 70+ supercomputers across India by 2025 with a combined capacity of over 50 petaflops, focusing on indigenising hardware and software to reduce dependency on imports. Educational and research institutions must gear up to ensure students take advantage of the country’s best supercomputers for research. The research budget must be released to the institutions in time. India hosts one-sixth of the global population; any positive changes due to cutting-edge research will likely benefit the world.
(With inputs from Shree D N)
(The author is retired Principal Chief Conservator of Forests (Head of Forest Force) Karnataka)
==
The PARAM Series
1. PARAM 8000 (1988): India’s first indigenously built supercomputer, developed in response to the denial of Cray supercomputers by the United States. Based on Inmos transputers and UNIX operating system.
Speed: 1 gigaflop (1 billion floating-point operations per second).
2. PARAM 8600 (1994): An upgraded version of PARAM 8000 with better processors and interconnects.
Speed: Around 5 gigaflops.
Details: Focused on scientific research and complex simulations.
3. PARAM 9000/SS (1996): Incorporated vector processors for enhanced performance.
Speed: Approximately 20 gigaflops.
Purpose: Used in weather forecasting and molecular modelling.
4. PARAM 10000 (1998): India’s first teraflop-capable system. Utilised the Sun UltraSPARC II processor and high-speed interconnects.
Speed: 100 gigaflops peak performance.
Details: Marked a leap in computational power, supporting diverse scientific applications.
5. PARAM Padma (2002): Based on Itanium 2 processors and high-speed interconnects.
Location: C-DAC, Pune.
Speed: 1 teraflop (1 trillion floating-point operations per second).
Purpose: Climate modelling, seismic analysis, and bioinformatics.
6. PARAM Yuva (2008): Focused on energy efficiency with improved cooling mechanisms.
Speed: 54 teraflops.
Purpose: Targeted research in weather forecasting, space research, and nanotechnology.
7. PARAM Yuva II (2013): Ranked among the world’s top supercomputers at the time, enabling breakthroughs in engineering, drug discovery, and aerodynamics. Used Intel Xeon processors and GPU accelerators.
Speed: 524 teraflops
8. PARAM Kanchenjunga (2016)
Location: National Institute of Technology (NIT), Sikkim.
Speed: 15 teraflops.
Purpose: Regional focus on weather forecasting and regional scientific research.
9. PARAM Brahma (2017)
Location: IISER, Pune.
Speed: 797 teraflops.
Purpose: Designed for complex simulations in physics, chemistry, and biology.
10. PARAM Shivay (2019): Partially built with indigenous components under the National Supercomputing Mission (NSM).
Location: IIT (BHU), Varanasi.
Speed: 833 teraflops.
Purpose: Multidisciplinary research in areas like engineering and biotechnology.
11. PARAM Sanganak (2020)
Location: IIT Kanpur.
Speed: 1.6 petaflops.
Purpose: Research in AI, machine learning, and data analytics.
12. PARAM Siddhi-AI (2020): Ranked among the top 100 supercomputers globally in 2020
Location: C-DAC, Pune.
Speed: 210 petaflops (AI-focused performance).
Purpose: AI, big data, drug discovery, and societal applications.
13. PARAM Ganga (2022)
Location: IIT Roorkee.
Speed: 1.66 petaflops.
Purpose: Facilitates advanced research in AI, engineering, and healthcare.
14. PARAM Ananta (2022)
Location: IIT Gandhinagar.
Speed: 3.3 petaflops.
Purpose: Advanced research in energy, agriculture, and environmental modelling.
Other supercomputers
Pratyush (2018)
Location: Indian Institute of Tropical Meteorology (IITM), Pune.
Speed: 4 petaflops.
Purpose: Weather forecasting, climate modelling and monsoon prediction. Aids in studying phenomena like cyclones, thunderstorms, and air quality.
Mihir (2019)
Speed: 2.8 petaflops.
Location: National Centre for Medium-Range Weather Forecasting (NCMRWF), Noida.
Purpose: Medium-range weather forecasting upto 10 days complements Pratyush for nationwide coverage and simulation.
AIRAWAT (2020)
Location: Centre for Development of Advanced Computing (C-DAC), Pune
Speed: 210 AI-specific petaflop
Purpose: AI-based research used in healthcare, agriculture, smart cities, education and governance and climate research
SahasraT (2016): Used in advanced materials science, life sciences, and engineering research.
Speed: 1.27 petaflops.
Location: Indian Institute of Science (IISc), Bengaluru.
HP Apollo 6000 (2015): Contributes to AI and ML research, engineering simulations
Speed: Over 1 petaflop.
Location: IIT Madras.
Arka (2020)
Location: Indian Institute of Technology (IIT) Ropar, Punjab
Speed: 1.35 petaflops
Purpose: Engineering, data science, artificial intelligence (AI), machine learning (ML), simulation-based studies and complex data processing
Arunika (2021)
Location: Indian Institute of Technology (IIT) Roorkee, Uttarakhand
Speed: 1.66 petaflops
Purpose: To support AI, machine learning, and engineering simulations, also used for research in structural engineering and materials science.