IIT-KGP to get its own supercomputer: Here are 5 most powerful supercomputers in India

India is planning to build the world's fastest supercomputer, an exascale machine that is expected to cost Rs 4,700 crore over the course of five years.

The Indian Institute of Technology, Kharagpur, is the first institution to receive a supercomputer under the National Supercomputing Mission. The supercomputer is a petascale system, and will be used for research in bio molecular simulations, drug design and bio-informatics, climate change and digital earth, geo-scientific exploration, infrastructure design and sustainable cities. The National Supercomputing Mission has received approval for installing eighty supercomputers across the country in the next seven years. Some of them will be imported, some will be indigenously made.

Here though, is a list of the five most powerful computing systems currently used in India.

5. PARAM Yuva - II at the Centre for Development of Advanced Computing

The PARAM Yuva - II is the fastest supercomputer built indigenous in India, with a peak computing power of 520.4 Tera Flops. The supercomputer was specifically designed for the benefit of the High Performance Computing (HPC) community in India. Housed in the C-DAC premises on the Pune University campus, the supercomputer is a general purpose research platform for accelerating the solving of scientific problems, and providing an opportunity for scientific breakthroughs.

IIT-KGP to get its own supercomputer: Here are 5 most powerful supercomputers in India

The PARAM Yuva - II is the most energy efficient supercomputer in India, although not the most powerful. It provides more computation power per unit of electricity consumed, and is more conservative on power usage than even the more powerful imported supercomputers in India. A number of scientific packages have been deployed on the supercomputer. Researchers have studied Material Science, Quantum Chemistry, Molecular Modeling, Climate Modeling, Computational Fluid Dynamics and Electronics on the PARAM Yuva - II.

4. HP Apollo 6000 at the Indian Institute of Technology, Delhi 

The HP Apollo 6000 Xl230/250 supercomputer at Delhi has a peak performance of  524.4 Tera Flops. With 2880 CUDA cores, the high performance computing cluster is the most powerful GPU based supercomputer in the country. The supercomputer is available to developers and researchers, who can queue jobs in the supercomputer, which then executes them sequentially. The supercomputer has 500 TB of on-board storage, and 1,000 TB of scratch disk space.

HPE_Apollo6000_Pane_2x_md

The supercomputer has been used to study viruses, molecular simulations and folding proteins. The computer has also been used for artificial intelligence purposes, with an artificial neural network set up to classify images. The usage of the supercomputer is charged to the various department that use the facility of the supercomputer. The supercomputer has allowed for the publication of a number of research papers in chemistry and physics, with four research papers already published in 2017.

3. Cray XC30 at the Tata Institute of Fundamental Research 

The Cray XC30 is located at a facility in Hyderabad and has a peak performance of 730.7 Tera Flops. There are 476 nodes on the supercomputer, each with a Intel Ivy Bridge 10-Core 2.8 GHz CPU and a Nvidia Tesla K20x. The supercomputer is used at the TIFR balloon facility, which studies the upper atmosphere, space plasma, astrobiology, interstellar dust, high energy physics and cosmic rays. All aspects of scientific ballooning are integrated into one facility.

Cray_XC30

Balloons are released from the facility in two seasons of the year, with balloons collecting a wide variety of data according to the mission. The data is stored and processed on a data center in the Hyderabad facility, and the Cray XC30 has been installed for this data center. The supercomputer is the third most powerful supercomputer in India, and ranks at one hundred and forty fifth worldwide. At the time of installation, the Cray XC30 at TIFR was the second most powerful supercomputer in India, and one hundred and thirteenth in the world.

2. Aaditya at the Indian Institute of Tropical Meteorology

The Aaditya supercomputer at the Indian Institute of Tropical Meteorology in Pune has a peak performance of 790.7 Tera Flops. The Supercomputer uses the iDataPlex dx360 M4 compute nodes from Lenovo, and is designed to be efficient in both space utilisation and power consumption. Aaditya is a common supercomputing facility for all the organisations under the Ministry of Earth Sciences, and is part of the back end for the Central Information Processing System (CIPS) of the Ministry of Earth Sciences.

aaditya

Then Cabinet Secretary Shri Ajit Kumar Seth visiting Aaditya in 2015. Image: IITM.

The CIPS ingests data from various sources, including satellites, ground based sensors and radar. The computer is one of four high performance computing systems across the country, that work together for weather prediction, which is then used by the states for land and ocean forecasting services. The supercomputers are used for dynamic weather models for short term and medium term weather forecasting, as against the traditional statistical models. By the next year, the supercomputers are planned to be used for long term forecasts as well.

1. Sahasrat at the Indian Institute of Science 

The Sahasrat at the Indian Institute of Science is the most powerful supercomputer in India, with a peak capacity of 1244.2 Tera Flops. The supercomputer is the SERC - Cray XC40, which is the next generation of supercomputer than the one used by the balloon facility of Tata Institute of Fundamental Research in Hyderabad. The Sahasrat is housed in Bengaluru and has two petabytes of on board memory.

The Sahasrat is the first petascale system in India.

The SahasraT. Image: The Indian Institute of Science.

The SahasraT. Image: The Indian Institute of Science.

Sahasrat has been used for research into climate science and astrophysics. Twenty four cores make up a single node on the supercomputer, and four such nodes make up one blade. 16 blades are assembled on a chassis, known as a rack. There are 8 such racks. If you do the math, that works out to 33,000 cores. The Sahasrat uses Intel’s Haswell processors. Each node can have its own software and operating system, which means that essentially each node is an independent computing system. The different nodes are connected with the Dragonfly network topology.

India is planning to build the world's fastest supercomputer, thirty times more powerful than the Sunway TaihuLight in China, which is currently the most powerful supercomputer in the world. The project is expected to cost Rs 4,700 crore over the course of the next five years. The resulting machine would be the first indigenously built exascale supercomputer.

Tech2 is now on WhatsApp. For all the buzz on the latest tech and science, sign up for our WhatsApp services. Just go to Tech2.com/Whatsapp and hit the Subscribe button.






also see

science