What's new

India Greenlights $730 Million Supercomputing Grid

thank you.

these institutions like cdac, iis, iit, iiit, drdo etc should rather be designing microprocessor and operating system than be putting together a room full of western computers and giving them sanskrit names.

supercomputers are a obsolete way of computing but which in modern times have become a means of nationalistic self-glorification among china, india and usa.

if the method of computing ( problem construction, os design, processor design, language etc ) are radically simplified, it wouldn't really need a room-full of computers... i myself have been working on processor design which is innovative and much simplified... sophistication is simplification and not over-complication.

what became of drdo's announcement in 2010 about them writing a "hack-proof, crash-proof" operating system?? what became of the "india microprocessor program" announced by the min. of info. tech in 2009?? until 2011, every year 300,000 computer science "graduates" came out of indian colleges... and computing has been there in india for about 40 years... yet no processor and operating system... shameful... as long as india follows the indian way of education and research, no processor or os will come out of india... such things take a different thinking.

That is right, why invent and build a super computer grid when you can instead create a Ummah ?

Why would you ever need brute force calculations ? especially when we have to collate data from hundreds of thousands of weather stations around the world, each one recording dozens of variables like wind speed/direction, temperature, humidity, pollen count etc.

Why would we need a super computer to calculate the impact of all of this data and compared it with historical data, all of which must be fetched from huge memory and storage banks. :cheesy:

Why do we even need to predict the weather, especially when Agriculture and Fishing depend on it. :lol:

Why would you ever need a super computer to compute a new design for a more efficient nuclear device, or for a brute force cracking of a 256 bit encryption ? All we need to do is trust in god :P

Why would you need a super computer to molecular dynamic modelling to create new drugs and compounds ? :cheesy: ..... we need to use test tubes like in the 50's.

All we need is a Calculator and a quran. Right ? :D
 
That is right, why invent and build a super computer grid when you can instead create a Ummah ?

Why would you ever need brute force calculations ? especially when we have to collate data from hundreds of thousands of weather stations around the world, each one recording dozens of variables like wind speed/direction, temperature, humidity, pollen count etc.

Why would we need a super computer to calculate the impact of all of this data and compared it with historical data, all of which must be fetched from huge memory and storage banks. :cheesy:

Why do we even need to predict the weather, especially when Agriculture and Fishing depend on it. :lol:

Why would you ever need a super computer to compute a new design for a more efficient nuclear device, or for a brute force cracking of a 256 bit encryption ? All we need to do is trust in god :P

Why would you need a super computer to molecular dynamic modelling to create new drugs and compounds ? :cheesy: ..... we need to use test tubes like in the 50's.

All we need is a Calculator and a quran. Right ? :D

sigh...

do read the rest of my posts at least, before commenting.
 
Its current top system, a 719-teraflops IBM/Lenovo iDataPlex installed at the Indian Institute of Tropical Meteorology, has slid from 36th to 71st position since it making its TOP500 debut in 2013. And the nation’s second-fastest number-cruncher, the 388-teraflops PARAM Yuva II, has gone from 131 to 69 in the same timeframe.


India's 2nd fastest supercomputer, PARAM Yuva II, with Rmax at 388.4 TFLOPS/sec, ranked #131 not #69.

India's 1st fastest supercomputer, iDataPlex runs at 719.2 TFLOPS/sec, ranked #71. Note the world #1 supercomputer (China Tianhe-2) has Rmax at 33,862.7 TFLOPS/sec, i.e. 47 times that of India's fastest.

India has 9 supercomputers in TOP500:

Supercomputer_Share_Top_500_by_Country_Jun_2014.png

According to an official press statement from India’s Cabinet Committee on Economic Affairs, the mission involves both capacity and capability machines. Earlier reports stated that the first order of business would be raising India’s supercomputing ranking by standing up three petascale supercomputers, some 40-times faster than the country’s current fastest.


Despite huge fund of $730M, even at 40-times faster than current 719 TFLOPS/sec, that's around 28,760 TFLOPS/sec, i.e. still cannot catch up with existing fastest supercomputer (November 2014). More importantly, what is the time frame? Computing speed increases exponentially, it has increased almost 1,000,000-times in the past 12 years, so what the $730M initiative is achieving?

Supercomputers-history.svg.png

Rapid growth of supercomputers performance, based on data from top500.org site. The logarithmic y-axis shows performance in GFLOPS.
BLUE: Combined performance of 500 largest supercomputers
RED: Fastest supercomputer
YELLOW: Supercomputer on 500th place​
 
Last edited:
Despite huge fund of $730M, even at 40-times faster than current 719 TFLOPS/sec, that's around 28,760 TFLOPS/sec, i.e. still cannot catch up with existing fastest supercomputer (November 2014). More importantly, what is the time frame? Computing speed increases exponentially, it has increased almost 1,000,000-times in the past 12 years, so what the $730M initiative is achieving?

730 million is just the starting money of this project.GOI has plans to spend about 3 billion on this project.
 
730 million is just the starting money of this project.GOI has plans to spend about 3 billion on this project.

That's exactly the issue. With $730M ~ $3B, the future target is only 40-times of existing 719 TeraFlops (iDataPlex, India's fastest), i.e. 28 PetaFlops, which is still slower than old computers in other countries say. As I questioned, people should ask the government, what is the time frame for these 3 Petascale computers? Cos supercomputing speed across the world grow at exponentially (see above chart), by last year the top ones were already at Petascale (China, US, Japan), and now looking at next level i.e. EXAscale.

Top 5 of TOP500, November 2014:
1
23px-Flag_of_the_People%27s_Republic_of_China.svg.png
Tianhe-2, 33.863 PetaFlops
2
23px-Flag_of_the_United_States.svg.png
Titan, 17.590 PetaFlops
3
23px-Flag_of_the_United_States.svg.png
Sequoia, 17.173 PetaFlops
4
23px-Flag_of_Japan.svg.png
K Computer, 10.510 PetaFlops
5
23px-Flag_of_the_United_States.svg.png
Mira, 8.586 PetaFlops​
 
well, i have actually mentioned it quite a few times but...

it is clock-less in execution of instructions, less than 25 instructions, risc, very wide instructions ( 160 bit ), no floating point operations, asynchoronous i/o etc.

it is designed to be used within a portable computer ( also being designed ) which is universal and meant to replace desktops, servers, embedded systems, cell phone, tablet etc.

a new operating system specific to this processor and computer will be written.

a socialist processor, socialist portable computer. :enjoy:


Yaar stop your socialism communism rubbish. Tell me which muslim communist will give woman equal rights and allow her to have 4 husbands?
 
That's exactly the issue. With $730M ~ $3B, the future target is only 40-times of existing 719 TeraFlops (iDataPlex, India's fastest), i.e. 28 PetaFlops, which is still slower than old computers in other countries say. As I questioned, people should ask the government, what is the time frame for these 3 Petascale computers? Cos supercomputing speed across the world grow at exponentially (see above chart), by last year the top ones were already at Petascale (China, US, Japan), and now looking at next level i.e. EXAscale.

Top 5 of TOP500, November 2014:
1
23px-Flag_of_the_People%27s_Republic_of_China.svg.png
Tianhe-2, 33.863 PetaFlops
2
23px-Flag_of_the_United_States.svg.png
Titan, 17.590 PetaFlops
3
23px-Flag_of_the_United_States.svg.png
Sequoia, 17.173 PetaFlops
4
23px-Flag_of_Japan.svg.png
K Computer, 10.510 PetaFlops
5
23px-Flag_of_the_United_States.svg.png
Mira, 8.586 PetaFlops​
India's Supercomputer program was temporarily halted due to various reasons but now work is going on at a very rapid pace and C-DAC may deliver a high petascale computer by 2020.
Besides,not having a semiconductor foundry is a very big impediment towards a successful electronics development environment.China sure is lucky on that regard.
 
Back
Top Bottom