How different are supercomputers to normal computers?, Pg11

Key Highlights:

• Supercomputers solve complex problems ordinary computers can't, like weather forecasting and simulating nuclear reactions.
• They use parallel computing, employing thousands of processors simultaneously to tackle small parts of a problem.
• India began developing indigenous supercomputers in the late 1980s, leading to the PARAM series.
• The National Supercomputing Mission (NSM), launched in 2015, aims to build a network of over 70 high-performance computing facilities across India.
• AIRAWAT-PSAI at C-DAC in Pune is currently India's fastest supercomputer, ranking among the top 100 worldwide.
• Future supercomputers may include quantum computers and neuromorphic computing designs for improved efficiency.
• JUPITER in Germany became Europe's first exascale supercomputer on September 5, powered by renewable energy.

Detailed Insights:

• Supercomputers consist of processors (CPUs and GPUs), nodes (processors bundled with memory), and a high-speed network for data transmission.
• They require specialized cooling systems and efficient power distribution due to the enormous heat and electricity generated.
• Parallel programming languages like MPI and OpenMP are used to control how processors work together, ensuring efficient task scheduling and memory management.
• Load balancing is crucial to prevent processors from idling while others are overloaded, optimizing speed and power consumption.
• Supercomputer performance is measured in flops, with top supercomputers operating in the exaflops range (quintillions of operations per second).
• Users interact with supercomputers remotely via secure network connections, submitting job scripts to a scheduler that manages tasks.
• The NSM emphasizes indigenously developed processors like the Rudra server and AUM high-performance computing nodes.
• Indian researchers use supercomputers for diverse applications, including weather modeling, molecular dynamics, drug discovery, and simulating black holes.
• Exascale computing is pushing the limits of classical machines, while quantum and neuromorphic computing offer potential future advancements.

Scientific/Technical Concepts Involved:

• CPU (Central Processing Unit): Handles general computing tasks in a computer.
• GPU (Graphics Processing Unit): Excels at repetitive mathematical operations, useful in scientific simulations.
• Parallel Computing: Using multiple processors simultaneously to solve a problem faster.
• Exaflops: A measure of computing speed, representing quintillions (10^18) of floating-point operations per second.