Saturday, February 9, 2013


On 8th Feb 2013, India unveiled its fastest supercomputer PARAM YUVA II. This post is all about it. 
Ever since India was denied CRAY supercomputer, we took upon to give ourselves a Supercomputer. That is how the first indigenously build supercomputer, PARAM 8000 was born(1991). 

See the whole evolution of PARAM Series in India, here.

In recent years, India had set a vision:

In 12th five year plan:
(1) "To identify 200 scientists and 2000 researchers dedicated to Fourth(4th) research paradigm i.e. Data Intensive Discovery. To create an ecosystem where 25% of the ecosystem of the country would be connected  by supercomputer activity. For this purpose a proposal has been received for establishing 200 centres across the country connected over the National Knowledge  Network." 
(2) "To have indigenous efforts to build Peta-scale supercomputer capacities and capabilities for the country’s requirements that will place India among the top five supercomputing power in the world." 
(3) To set up a top-tier of around six(6) supercomputers with about 3-6 Petaflops computing power in different locations, followed by 12-20 zonal supercomputers of 200-500 Teraflops. The third rung would involve 20-50 machines with 10-Teraflop speed and finally, 50-100 regional supercomputers of one Teraflops.
The Science, Technology and Innovation Policy 2013:
(1) lays lot of emphasis on R&D and talks about increasing India's expenditure in this area. 
[Note:  (1) 4th Research Paradigm - Data Intensive Discovery 
First Paradigm of Scientific Research - 1000s of yrs ago we had – Experimental Science - desc. of natural phenomena 
Second Paradigm of Scientific Research - Last few centuries ago we had – Theoretical Science – Newton’s Laws of Motion, Albert Einstein's Theory of Relativity etc. 
Third Paradigm of Scientific Research - Last few decades we had – Computational Science – Simulation of complex phenomena, constructing mathematical models to solve complex problems, linear programming etc 

Fourth Paradigm of Scientific Research - Current date we have – Data-Intensive Science, where scientists are overwhelmed with lots of data.

Data from many different sources

- Data captured by instruments

- Data generated by simulations

- Data generated by sensor networks 
which then needs to be data mined, analysed, visualized and exploration, draw meaningful information and finally communicated to wider audience. 
National Knowledge Network(NKN) is trying to do cater to this form of distributed research.   

(2) FLOPS - Floating Point Operations per second. 
It is the unit for measuring performance of a computer, just like we measure the performance of a car(mileage -Kms/litre or acceleration time) or performance of a printer toner(no. of pages to be printed).
So, the nominal speed will be some number of FLOPS and then we have the higher units.
MEGAflops - Millions - 10^6
GIGAflops - Billions - 10^9
TERAflops - Trillions - 10^12
PETAflops - Quadrillions - 10^15
EXAflops - Quintillion - 10^18 
Note: You're going up by a factor of 1000(10^3) each time. :)  
Flop measurement of a computer is always its peak performance size. So when I say a computer can scale or work at 500 Teraflops, that is its peak calculation performance. Marketing people or even by other people for normal information purposes, use flops measurement system to tell about computer performance, but this can at times, be a misnomer.
While we talk about the peak performance of machine(in flops), but what we are really bothered about is how fast the science is delivered on the machine, or how fast the weather model runs on the computer. This is called the Time to SolutionFlops and Time to solution are related but they are not the same thing.
Note: Today a typical personal computer has a computing power in the order of GIGAflops. Samsung Galaxy S2 is ~ 85 Gigaflops !! :)  
1. PARAM YUVA II supercomputer is developed by CDAC Pune
Remember: CDAC is R&D arm of Department of Electronics and Information Technology (DeitY), Ministry of Communications & Information Technology (MCIT).
Also, CDAC is not the only organization which develops super computers. ISRO, CSIR, BARC also develops super computers. 

2. PARAM YUVA II is the seventh(7th) in the PARAM series. At ~ 500 Tera Flops, its the fastest supercomputer in India, which is being indigenously developed. 

3. This supercomputer uses Hybrid Technology - which is network of high performance computers. This hybrid technology has enabled the up-gradation of existing 54 Teraflops to 500 Teraflops peak computing power of Indian Supercomputers. 

This up-gradation of computing power was taken up as part of National Supercomputing Mission which was part of National Policy on Information Technology (NPIT) - 2012.

4. As mentioned above, India has a vision of creating an ecosystem of High Performance Computers in the country with tiered architecture. PARAM YUVA II will be incorporated in the National Knowledge Network (NKN) which will help majority of research institutes and scientific community to leverage the high computing powers of this computers for their work.

5. Param Yuva II has high storage capacity(200 TB), it supports parallel computing(infact all Supercomputers must support this feature). 

Typically most of the data centers (places where large mainframe or super computers are located) requires a lot of cooling devices, as it generates a lot of heat due to high number of processors running. So these data centers are in a way primary sources of Green House Gases. In this respect Param Yuva II is made energy efficient and consumes around 35% less power.

6. Param Yuva II is purported to be used in areas like weather forecasting, Monsoon Mission(I have covered monsoon mission of GoI here), seismic data analysis, Cancer Bioinformatics, Genome Sequencing, Discovery & Extraction of new energy sources like oil and natural gas etc. Apart from this, even pharmaceutical companies can use it for drug discovery (Open Source Drug Discovery is an important candidate here). So the scientific institutes, research laboratories and universities will be benefited a lot from this supercomputer. 

E.g. - 

(i) The increase in peak computing power in Param Yuva II will help reduce the time required for weather predictions. 
(ii) Also, if supposedly the current system was used to predict weather conditions in a 6 km area, Param Yuva II would help in predicting weather conditions for a 10km area etc. 
(iii) Also, I remember reading somewhere that for things like Indian Forests Survey or say Coral Reef Study etc where complex remote sensing data is sent back to Earth through the G-SAT or G-SLV satellites(BTW the one to go up soon will be the SARAL - Argos/Altika, read about it) will require a high performance computers to churn GBs of data and use complex models to draw scientific understanding from it. 

Remember accelerating a nation’s R&D programs by increasing the productivity of scientists and researchers will in turn lead to both scientific advancement and economic competitiveness of a nation !!! 


Some mock questions on Supercomputers for Civil Services Prelims Exam 2013. 

All The Best !! :) 

Question I : Consider the following statements:
1. PARAM series of high performance computers were the first supercomputers India had developed. 

2. Basic principle behind any PARAM computers is that calculations are broken up into segments and solved parallel to one another.

3. PARAM YUVA II ranks in top 50 fastest supercomputers in the world.

4. In order to boost domestic supercomputing manufacturing facility, Govt. of India, in its National Information Technology Policy 2012, has identified, Centre for Development of Advanced Computing(CDAC), as the nodal government agency authorised to develop supercomputers for all government departments in India. They are also authorised to export supercomputer to other countries in the World. 
Which one of the above statements are not incorrect?

(A) Only 1, 3 and 4

(B) Only 1 & 2

(C) Only 2 and 3

(D) Only 2 and 4  

Question II: Consider the following statements:
1. India's supercomputing development has seen an evolution from the one driven by weather forecasting in 1990s to currently being involved in Data Intensive Discovery and distributed computing. 

2. The 12th Five Year Plan earmarks Rs 5,000 crore, which would ensure that at least 25% of the scientific population of the country has access to a supercomputer.

3. The supercomputing demand in India is primarily driven by the Private Industries need for mainstream supercomputing, involving 2-10 Teraflops machines, as is also seen in US, China and Japan.

Which one of the options below are correct?

(A) Only 1 and 3

(B) All 1, 2 and 3

(C) Only 1 and 2 

(D) Only 3 

(E) None of the options (A), (B), (C) or (D) are correct 


  1. Answers:
    Question 1: B - Bcoz we are not allowed to export?

    Question 2: E ???

  2. good info.can you clarify about SAGA 220 also?difference and speed.


Note: Only a member of this blog may post a comment.