Science and Tech Part 5: PARAM to the NKN

In one line: Science and Technology Part 5: Supercomputing questions in Indian exams are fact-cluster questions — the PARAM lineage, the National Supercomputing Mission’s numbers, AIRAWAT and the AI stack, and the National Knowledge Network that ties research institutions together.

Supercomputing questions in Indian exams are fact-cluster questions: the PARAM lineage, the National Supercomputing Mission’s numbers, AIRAWAT and the AI stack. Beneath them sits the National Knowledge Network, the high-speed digital spine that ties research institutions together. The underlying concepts — what FLOPS measures, what a petaflop actually is — frame the prelims one-liners. The mission architecture, meanwhile, feeds mains answers on digital infrastructure, research capacity and AI readiness. This note covers both layers.

Quick Answer: Supercomputing in exams = five fact clusters (PARAM history, NSM numbers, AIRAWAT/IndiaAI, NKN, global rankings) built on one concept (FLOPS). Learn them as a single “compute sovereignty” stack: NKN connectivity + NSM compute + IndiaAI GPU layer.

Table of Contents.

1. The Concept Layer: What a Supercomputer Is Measured In
2. The PARAM Lineage: India’s Supercomputer History
3. The National Supercomputing Mission: The Number Card
4. AIRAWAT and the AI Compute Stack
5. The National Knowledge Network
6. Where India Stands Globally
7. How Exams Probe This Topic
8. Quick Revision: One-Glance Facts

1. The Concept Layer: What a Supercomputer Is Measured In.

Before memorising mission numbers, anchor the topic in one concept: performance in FLOPS — floating point operations per second. A floating point operation is essentially a arithmetic calculation involving decimal fractions, and supercomputers are ranked by how many trillions or quadrillions of these they can perform every second.

• The performance ladder: mega (10⁶) → giga (10⁹) → tera (10¹²) → peta (10¹⁵) → exa (10¹⁸) FLOPS. Each step is a thousandfold jump — so a petaflop machine performs a quadrillion calculations per second, and an exaflop machine performs a quintillion.
• The global scoreboard: the TOP500 list, ranked by the LINPACK benchmark (a standardised dense linear-algebra test) and published twice a year, in June and November. Its complement is the Green500, which ranks machines by energy efficiency (flops per watt) — an increasingly important metric as power consumption becomes the binding constraint on frontier systems.
• Why HPC matters: weather and monsoon forecasting (the flagship Indian use-case), molecular drug design, computational fluid dynamics for aircraft and missiles, genomics (links to Genome India in Part 4), crash simulation, climate modelling and AI model training. The difference between a 5-day and a 7-day accurate monsoon forecast is, quite literally, supercomputer hours — better resolution models demand more compute.
• The architecture idea: supercomputers achieve speed by parallelism — lakhs of processor cores working on divided parts of the same problem — connected by high-speed interconnects. Raw single-core speed matters less than the network fabric and the software stack that keeps thousands of processors synchronised. This is why “indigenous supercomputer” claims often refer to the design, interconnect and software stack rather than every chip inside the machine.

2. The PARAM Lineage: India’s Supercomputer History.

The origin story is the classic exam hook, and it begins with denial. In the mid-1980s, the United States refused to sell India a Cray supercomputer (intended partly for weather forecasting), arguing the technology could have military applications. India’s response was to build its own.

• C-DAC — the Centre for Development of Advanced Computing, Pune, established in 1988 under (what is now) MeitY, with the late Vijay Bhatkar as its founding director — built PARAM 8000 in 1991: India’s first indigenous supercomputer. It was among the fastest machines in the world for its price class at the time.
• PARAM stands for Parallel Machine. The lineage ran PARAM 8000 → PARAM 8600 → PARAM 9000 → PARAM 10000 → the modern PARAM Siddhi-AI — a three-decade arc of continuous capability building that mains answers can cite as evidence of technology-denial-driven indigenisation.
• PARAM Siddhi-AI (2020, C-DAC) delivered 33 petaflops of peak AI performance (Rpeak; Rmax lower) and debuted at approximately 63rd on the TOP500 list in November 2020 — the flagship indigenous software achievement of the NSM era. A nuance exams have probed: it runs on NVIDIA hardware, so the “indigenous” claim is about the software and systems stack, not the processors.
• Other institutions in the game: IISc Bengaluru, the IITs and C-DAC host most of India’s top machines. The Indian Lattice Gauge Theory Initiative (ILGTI) machines serve fundamental physics research, and MoES runs dedicated forecasting systems such as Mihir and Pratyush for weather and climate.

3. The National Supercomputing Mission: The Number Card.

The National Supercomputing Mission (NSM) is the number-card topic — launch year, ministries, outlay, phases, indigenous components. Deploy these figures in exactly this order in mains answers.

• Launched in 2015 by MeitY and the Department of Science and Technology (DST), implemented through C-DAC and its partner IISc, with an outlay of ₹4,500 crore. The original aggregate target of 33 petaflops was later enhanced to approximately 45+ petaflops as more systems were commissioned.
• The three-phase plan: Phase I — assembling imported machines and installing them at institutes (imported hardware, some domestic subsystems). Phase II — partially indigenous design and manufacture, featuring the RUDRA family of servers built by C-DAC and the Trinetra interconnect (the indigenous elements exams most often cite). Phase III — a fully indigenous supercomputer, design to chip.
• Reach: 24–30+ institutions connected with NSM HPC facilities — IITs (Kharagpur, Chennai, Bombay and BHU among the early hosts), NITs, IISc, C-DAC centres and national research laboratories, with the network connectivity supplied by the National Knowledge Network (below).
• The applications flagship: the MEGHADUTU cloud-HPC platform, which lets researchers access supercomputing as a service; weather and climate modelling in partnership with MoES; bioinformatics and drug-discovery workloads; and the Garuda grid for federated computational resources. These are the applications-layer names that lift a mains answer from generic to specific.

4. AIRAWAT and the AI Compute Stack.

The topic has evolved from pure HPC to an HPC-plus-AI convergence story, and recent exams reflect this.

• AIRAWAT — the AI Research, Analytics and Knowledge Dissemination Platform — is a National Programme on AI initiative anchored at C-DAC. It provides AI-dedicated compute capacity, including a 200+ petaflops AI-capable system being scaled under the IndiaAI mission. PARAM Siddhi-AI belongs to the same AI compute line.
• The IndiaAI Mission (approved 2024) carries a ₹10,371 crore outlay with a dedicated compute pillar: a target of 38,000+ GPUs for shared national AI compute, offered as common compute at subsidised rates to startups, researchers and academic institutions. This is the current-affairs update that supercomputing mains answers now need.
• The exam link: HPC + AI convergence is the standard framing. “Supercomputing is the substrate of India’s AI ambitions” — pair NSM with IndiaAI and the GPU sovereignty debate (a strand of the wider digital infrastructure discussion, including large-scale national infrastructure programmes).

5. The National Knowledge Network.

NKN is the least glamorous and most examinated component of the stack — and the easiest to score.

• NKN, launched in 2010 and implemented by NIC under MeitY, is a multi-gigabit national research and education network (NREN) with a backbone of up to 100 Gbps. It connects universities, research labs, libraries and hospitals — over 1,700 institutions in all.
• What it enables: collaborative research access (shared supercomputing under NSM literally rides on NKN’s pipes), virtual classrooms and e-learning, grid computing (the Garuda grid), e-governance data exchange, and international connectivity to global research networks such as GEANT (Europe), Internet2 (US) and TEIN (Asia-Pacific).
• The exam line: NKN is the plumbing of Indian e-science; NSM provides the engines. Mains answers should present them as one integrated stack — NKN connectivity + NSM compute + the IndiaAI/GPU layer for AI workloads. A system is only as fast as the network that feeds it.

6. Where India Stands Globally.

• India’s fastest systems sit in the TOP500’s upper-middle ranks — PARAM Siddhi-AI debuted at ~63rd in November 2020, but subsequent lists have seen it and the Mihir-era machines slide as global frontier systems pulled away. The world’s frontier is now exascale: Frontier (US, the first exascale machine, 2022), Aurora, and El Capitan; China’s exascale-class entries have famously gone undeclared on the list.
• Frontier-class machines exceed 1 exaflop; India’s mission targets aggregate petaflops — a two-orders-of-magnitude gap that honest mains answers should acknowledge, alongside the genuine indigenisation progress (RUDRA servers, Trinetra interconnect, the C-DAC software stack). Framing the gap plus the trajectory is stronger than either alone.
• The quantum adjacency: the National Quantum Mission (2023, ₹6,000 crore) is the next compute frontier, and exams increasingly pair supercomputing and quantum computing as “compute sovereignty” questions. Keep the two missions as companion entries in your notes.

7. How Exams Probe This Topic.

• Prelims one-liners: “PARAM 8000 was developed by —” (C-DAC, 1991). “The National Supercomputing Mission is implemented by —” (MeitY + DST via C-DAC and IISc). “FLOPS measures —” (floating point operations per second). “NKN is implemented by —” (NIC). “The TOP500 uses which benchmark?” (LINPACK).
• Statement checks: “PARAM Siddhi is fully indigenous hardware” — false (indigenous software stack on NVIDIA hardware). “NKN connects only universities” — false (also research labs, libraries and hospitals). “Green500 ranks raw speed” — false (it ranks energy efficiency).
• Mains framing: “The National Supercomputing Mission is central to India’s research and AI ambitions. Discuss its progress and challenges.” Deploy the number card (2015, ₹4,500 crore, three phases, RUDRA/Trinetra), name applications (monsoon forecasting, genomics, drug discovery, MEGHADUTU), acknowledge the exascale gap, and close with the IndiaAI GPU pillar as the forward link.
• Current affairs hooks: IndiaAI mission compute targets, new TOP500 appearances of Indian systems, and NQM linkages — always check the most recent TOP500 list before your exam, since rankings shift every June and November.

8. Quick Revision: One-Glance Facts.

• FLOPS ladder: peta = 10¹⁵, exa = 10¹⁸; TOP500/LINPACK, published June & November; Green500 = energy efficiency.
• Cray denial → C-DAC (1988) → PARAM 8000 (1991); PARAM Siddhi-AI (2020, ~63rd on TOP500 debut, 33 PF peak AI performance).
• NSM: 2015, MeitY + DST, implemented via C-DAC/IISc, ₹4,500 crore, three phases, RUDRA servers + Trinetra interconnect, ~45+ PF aggregate target.
• AIRAWAT (200+ PF AI compute); IndiaAI Mission 2024 (₹10,371 crore; 38,000+ GPUs).
• NKN: 2010, NIC, up to 100 Gbps backbone, 1,700+ institutions; the connectivity layer NSM rides on.
• Exascale era (Frontier, 2022); India operates at petaflop aggregate — the gap mains answers must state honestly.

Supercomputing is a five-fact topic wearing a technology costume: the PARAM origin story, the NSM number card, PARAM Siddhi’s ranking, the AIRAWAT/IndiaAI compute layer, and NKN. Learn those five as a single “compute sovereignty” stack — connectivity (NKN), compute (NSM), AI capacity (IndiaAI) — and keep one TOP500-current update in your notes. That covers both the prelims one-liners and the mains framing this topic reliably generates.

Read next: The Indus Waters Treaty, 1960-2026: Survival, Abeyance and the Exam Questions, Exam-Ready Notes

Related exam guides.

• Black Holes, LIGO and Lagrange Points: Gravitation’s Research Frontier.
• The Indus Waters Treaty, 1960-2026: Survival, Abeyance and the Exam Questions, Exam-Ready Notes.
• Junagadh and the Integration of the Princely States: From the 1948 Plebiscite to the 2025 Supreme Court Verdict, Exam-Ready Notes.

Frequently Asked Questions.

What is a supercomputer’s performance measured in?

FLOPS — floating point operations per second. The ladder runs mega (10⁶) → giga (10⁹) → tera (10¹²) → peta (10¹⁵) → exa (10¹⁸) FLOPS. The TOP500 list, ranked by the LINPACK benchmark and published twice a year (June and November), is the global scoreboard; the Green500 ranks energy efficiency.

Who built India’s first indigenous supercomputer and when?

After the US denied India a Cray supercomputer in the mid-1980s, C-DAC (Centre for Development of Advanced Computing, Pune, established 1988 under what is now MeitY) built PARAM 8000 in 1991 — India’s first indigenous supercomputer, among the fastest in the world for its price class at the time.

What are the key numbers of the National Supercomputing Mission?

Launched in 2015 by MeitY and DST, implemented through C-DAC and IISc, with a ₹4,500 crore outlay and an aggregate target enhanced from 33 to ~45+ petaflops. The three phases move from assembling imported machines, through partially indigenous design (RUDRA servers, Trinetra interconnect), to a fully indigenous supercomputer.

What are AIRAWAT and the IndiaAI Mission?

AIRAWAT (AI Research, Analytics and Knowledge Dissemination Platform) is a National Programme on AI initiative anchored at C-DAC providing AI-dedicated compute, including a 200+ petaflops AI-capable system being scaled under IndiaAI. The IndiaAI Mission (2024) carries ₹10,371 crore with a compute pillar targeting 38,000+ GPUs as shared national AI compute at subsidised rates for startups and researchers. PARAM Siddhi-AI is part of this AI compute line.

What is the National Knowledge Network?

NKN — launched in 2010, implemented by NIC under MeitY — is a multi-gigabit national research and education network (NREN) with a backbone of up to 100 Gbps, connecting over 1,700 universities, research labs, libraries and hospitals. It enables collaborative research (shared NSM supercomputing rides on it), virtual classrooms, grid computing (Garuda), e-governance exchange, and international links to GEANT, Internet2 and TEIN.

Related explainers

Environment Part 9: The 500 GW Push — Solar, Wind and Storage

iPhone Price Hike & PLI: Electronics Manufacturing Notes

Gaganyaan G1 & Vyommitra: Human Spaceflight Exam Notes