PARAM Siddhi- India's fastest Supercomputer
Updated: Jan 1
We find computers everywhere today. They have brought a revolution in the way we look at things. Our lives are no more normal without computers. With #computers around, everything is just a click away. They have put an end to many tedious jobs like hand-written documents, countless iterative calculations, storing hard copies of documents, and many more. In addition to all these, we have Supercomputers that are way more beneficial than general-purpose computers.
A #supercomputer is a high-level performance device that usually runs on a Linux based operating system. They are normally used for many high-end applications. One such supercomputer is the PARAM Siddhi-AI, which is currently India's fastest supercomputer. It is built as a part of the National Supercomputer mission( NSM ) by the Centre for Development of Advanced Computing( C-DAC) that is executed collectively by the 'Ministry of Electronics and Information Technology' and the 'Department of Science and Technology'. It is the 63rd fastest supercomputer amongst the world's top 500 powerful non-distributed computer systems. These rankings are released by Atos, a global Information Technology provider.
National Supercomputer Mission(NSM)
In the late 1980s, the USA refused the exportation of a Cray supercomputer as the device can have a dual-use in weapon development. This was when the need for self manufacture of supercomputers became obvious. The objective of the National Supercomputer Mission is to build a pack of 70 supercomputers that connect various educational and research institutes across the country.
This is carried out at the Centre for Development of Advanced Computing(C-DAC), Pune, and the Indian Institute of Science(IISc), Banglore. It was approved in April 2015 with a total budget of rupees 4500 crores for a period of 7 years. The plan was to build supercomputers with a performance ranging from a few Tera Flops to hundreds of Tera Flops, and also 3 devices with a performance of more than 3 Peta Flops, across the educational and research institutes of national importance.
All the built supercomputers would be connected across the National Knowledge Network(NKN). The NKN is a government program that connects various academic institutes, Research and Defence laboratories over a High-Speed network. This enables the smooth operation of all the interconnected systems. Thus, India would be self-sufficient in the supercomputing infrastructure.
Features of PARAM Siddhi
It is a non-distributed computer system. In a non-distributed system, all the components are present at the same physical location, unlike the Distributed systems where components are present at various physical locations and are connected across a #network. This supercomputer has 4.6 petaflops of Rmax and 5.267 petaflops of Rpeak. These technical terms are explained below.
Rmax - It specifies the maximum achieved performance.
Rpeak - It is the maximum theoretical performance that can be achieved.
Rmax and Rpeak are usually measured in TeraFlops or PetaFlops.
The term Flops refers to 'Floating Operations per Second'.
Hence the ability of the computer system to perform a quadrillion floating operations per second is PetaFlop.
A PetaFlop is equal to thousand TeraFlops.
PARAM Siddhi-AI is built with NVIDIA DGX A100 systems in connection with the NVIDIA Mellanox HDR InfiniBand Network. It uses the C-DAC HPC-AI Engine, Cloud Platform, and AI Software Stack. It is based on the DGX SuperPOD reference architecture( a turnkey AI infrastructure), which enables organizations to install powerful computer systems in a shorter period of time.
PARAM Siddhi has a wide range of applications:-
Simulate real-world problems and provide more appropriate and efficient solutions.
Tackle new social problems.
In the discovery and extraction of new resources.
Mathematical Applications and Computational Physics
Empower Academic Institutes, Research and Defence Laboratories across the country.
Solve the challenges of Agriculture, Health Care, Cybersecurity, Space, and Energy.
Visual Computing, Deep Learning, Virtual Reality, and Graphical Visualisation.
MSME's, Natural Language Processing, Surveillance, and Artificial Intelligence.
Other computers of the PARAM series
PARAM is a Sanskrit word which means Supreme. These series of computers are designed and assembled by the C-DAC at Pune, India. The first machine in this series is the PARAM 8000. It has a reconfigurable interconnection network and distributed memory MIMD architecture. The next machine PARAM 8600 is an improvement over PARAM 8000.
A few other machines in the #PARAM series are PARAM 9000, PARAM 10000, PARAM Padma, PARAM Yuva, PARAM Yuva II, and PARAM Ishaan. The latest supercomputer of the PARAM series is the PARAM Brahma. It is supported by a cooling system called the Direct Contact Liquid, which is the first of its kind.
We live in an era of Technology. Using #technology to the maximum extent is necessary. Making tasks easier and faster has been the most significant purpose of computers. Hence deploying supercomputers would enable us in solving the most complex problems and also boosts the nation's economy. They also fulfil the goals of Digital India and Make in India, India's two important missions.