Indian advancements in Supercomputing

[archie]

In 2012, Chinese R&D spending was $294 billions while Americans spent $454 billions. Now, tell me how much did you IPL spend?

Are you saying 294 Billion$ just for supercomputing research??? I doubt that number very much Intel Market capitalization is 148 Billion $ they feature on most top 500 Supercomputers..

IPL 2014 Brand value is estimated to be USD 3.2 Billion Player , Season 7 Prize money alone was USD 6.3 Million .brand endorsments are more...

 

[archie]

All it takes is about 500 million USD for a spanking new supr computer at US cost of living .. Im sure India can make it much cheaper

U.S. plans to build world's fastest supercomputers with $425 million

 

[no smoking]

Are you saying 294 Billion$ just for supercomputing research??? I doubt that number very much Intel Market capitalization is 148 Billion $ they feature on most top 500 Supercomputers..

We am talking about total R&D here.

IPL 2014 Brand value is estimated to be USD 3.2 Billion Player , Season 7 Prize money alone was USD 6.3 Million .brand endorsments are more...

So, a small part of your IPL SPENDING can't even cover a single supercomputer?

 

[no smoking]

All it takes is about 500 million USD for a spanking new supr computer at US cost of living .. Im sure India can make it much cheaper

U.S. plans to build world's fastest supercomputers with $425 million

It is only direct cost of designing and assembling this computer.
There are lots more hide in other departments which are producing the core parts of this computer.

 

[archie]

Your just talking crap here.. 294 Billion$ just for super computing research is impossible Do you even know how much your entire country's economy is valued at??

Can you list the other departments.. ??

 

[no smoking]

Your just talking crap here.. 294 Billion$ just for super computing research is impossible Do you even know how much your entire country's economy is valued at??

Once again, we are talking about total R&D here, not only super computing.

Can you list the other departments.. ??

Chips, Routers, Connections, Sensors, Coolers...
the material to build these Chips, routers, Connections, Sensors, Coolers...
the machine to build these Chips, routers, Connections, Sensors, Coolers..
the machine to produce these materials...
the machine to build these machines...
On and on.
Why do you think it is USA, China and Japan occupy the top 3 in the supercomputing list? They are the top 3 spenders on R&D.

 

[archie]

Once again, we are talking about total R&D here, not only super computing.



Chips, Routers, Connections, Sensors, Coolers...
the material to build these Chips, routers, Connections, Sensors, Coolers...
the machine to build these Chips, routers, Connections, Sensors, Coolers..
the machine to produce these materials...
the machine to build these machines...
On and on.
Why do you think it is USA, China and Japan occupy the top 3 in the supercomputing list? They are the top 3 spenders on R&D.

All that is called Computing fabric .. 457 Million is all inclusive of that you dont need a billion $ for that... Do you even know how much a billion $ is??

Cost of he top chip from Intel is about 900$ in 1000 unit Quantity and for a super omp they dont use top chip they use a scalable chip with lower power consumption a full blade machine with 6-12 unit processsing package will fit under 1000$ for 1000 unit quantities.. each of those 6-12 will contain about 4,6 or 12 processing cores ..

with the advent of computing Gfx cards and co processors the cost of supercomuting has fallen manyfold ..
Do you know how much Gflops a single nvidia Co processor adds to an overall system ??

High Performance Computing (HPC) and Supercomputing | NVIDIA Tesla | NVIDIA

Learn up before talking crap... the networks and switches that you talk of is is here is bilt for scalability inexpensive multiple unit modules and not single expensive unit

SeaMicro SM15000 Fabric Server and Freedom Fabric Storage

450 Billion $ is many times over of the entire economy of a small nation "maldives economy is $2.754 billion" sane people in chinese govt dare not spend that much money on one thing..

By the way if you all all the manufacturing equpment you are insane cause thats all of computing industry .. thats like counting the cost of a Aircraft with that of "cost of Aerospace industry"

 

[no smoking]

All that is called Computing fabric .. 457 Million is all inclusive of that you dont need a billion $ for that... Do you even know how much a billion $ is??

Cost of he top chip from Intel is about 900$ in 1000 unit Quantity and for a super omp they dont use top chip they use a scalable chip with lower power consumption a full blade machine with 6-12 unit processsing package will fit under 1000$ for 1000 unit quantities.. each of those 6-12 will contain about 4,6 or 12 processing cores ..

with the advent of computing Gfx cards and co processors the cost of supercomuting has fallen manyfold ..
Do you know how much Gflops a single nvidia Co processor adds to an overall system ??

High Performance Computing (HPC) and Supercomputing | NVIDIA Tesla | NVIDIA

Learn up before talking crap... the networks and switches that you talk of is is here is bilt for scalability inexpensive multiple unit modules and not single expensive unit

SeaMicro SM15000 Fabric Server and Freedom Fabric Storage

450 Billion $ is many times over of the entire economy of a small nation "maldives economy is $2.754 billion" sane people in chinese govt dare not spend that much money on one thing..

By the way if you all all the manufacturing equpment you are insane cause thats all of computing industry .. thats like counting the cost of a Aircraft with that of "cost of Aerospace industry"

Ok, I give up.
Based on your theory, the only conclusion is: either your scientists are too stupid to build a supercomputer with peanut budget, or your government doesn't even want to spend that peanut budget on it. Just pick one you like.

 

[Compersion]

Why India's Mars Orbiter Mission Cost Less Than 'Gravity' Movie - NBC News

Chinese space program - Wikipedia, the free encyclopedia

PRC is saying that they can spend money (forget about its efficiency) ...

the

PRC will build many super fast computer and spend billions.

India will build one super fast computer and spend millions.

what India is now thinking to do is build more super fast computers and with ambition ... and one would not be surprised if it was "cost-effective" ... the quality and use is different matter all together

"As missions go, Maven represents a Mercedes 'S' class to the Volkswagen that is MOM," Pillai wrote.

The latter company some say is better to the former ...

but sure we also need to drive the former and will get there

 

[archie]

Ok, I give up.
Based on your theory, the only conclusion is: either your scientists are too stupid to build a supercomputer with peanut budget, or your government doesn't even want to spend that peanut budget on it. Just pick one you like.

The basic question is what is the end result of building such a system will it recover the investment ?? Do we have software that is capable of extrating that much out of such a system ??
Building a super computer is easy . Operating such a system is expensive the one US is contemplating uses 24MW a small powerstation is needed to power it..

only if the industry has confidence in the uses of such a thing will they build it... or they can invest in software to farm the computing resources from several centers "distributed computing"

Its not a question of funding but "rate of return" I guess you do not understand all that.

Your statement of "294 Billion or 450 Billion" is factually incorrect.

 

[archie]

Upcoming super computers

Intel & Cray Land Contract for 2 Dept. of Energy Supercomputers

 

[Chanakya's_Chant]

India's C-DAC Launches World's Most Compact Supercomputing (HPC) System 'PARAM Shavak'

http://defenceforumindia.com/forum/...ompact-supercomputing-system.html#post1027267

 

[no smoking]

The basic question is what is the end result of building such a system will it recover the investment ?? Do we have software that is capable of extrating that much out of such a system ??
Building a super computer is easy . Operating such a system is expensive the one US is contemplating uses 24MW a small powerstation is needed to power it..

only if the industry has confidence in the uses of such a thing will they build it... or they can invest in software to farm the computing resources from several centers "distributed computing"

Oh, the demands for such a powerful computer is always there:
Don't your military department need it to simulate nuclear explosion?
Don't your agriculture department need it to calculate their latest weather model?
Don't your health department need it in their research on genetic project?

Its not a question of funding but "rate of return" I guess you do not understand all that.

Your statement of "294 Billion or 450 Billion" is factually incorrect.

Oh, it is the problem of funding---the funding to every departments.
As I said "294B" is the total R&D for the whole country not supercomputing only.
But when we discuss supercomputing, we can't ignore those key techs coming from other departments. These are the real problems that restrict scientists.

 

[archie]

Oh, the demands for such a powerful computer is always there:
Don't your military department need it to simulate nuclear explosion?
Don't your agriculture department need it to calculate their latest weather model?
Don't your health department need it in their research on genetic project?

There are many ways to achieve an end .. SUpercomputing is not the only way .. Planes were designed by paper and drawing they still flew.. now its easier and cost effective to design on a software package . The key word is "cost effective". No stupid person will spend more money to get it done when there is a cheaper means.

Oh, it is the problem of funding---the funding to every departments.
As I said "294B" is the total R&D for the whole country not supercomputing only.
But when we discuss supercomputing, we can't ignore those key techs coming from other departments. These are the real problems that restrict scientists.

The Whole Industry compries of Computers,laptops,custom fab, Mobilephones and any number of electronic devices and was built over decades .. Silicon valley is an entire industry.. Super computer is a single unit of High performance computer. Are you mad to say just to build a super computer you need investment in all that?? The thread topic is Only the Niche market of Supercomputers..
If you dont know what your talking about first accept it.. Please dont keep trollng with Wrong info...

 

[no smoking]

There are many ways to achieve an end .. SUpercomputing is not the only way .. Planes were designed by paper and drawing they still flew.. now its easier and cost effective to design on a software package . The key word is "cost effective". No stupid person will spend more money to get it done when there is a cheaper means.

Well, so far, computerized designing is way cheaper and better than manual calculation on paper. That is why countries like US, Russia and China are stupid enough to spend millions to equip their aviation designing department with supercomputer.

The Whole Industry compries of Computers,laptops,custom fab, Mobilephones and any number of electronic devices and was built over decades .. Silicon valley is an entire industry.. Super computer is a single unit of High performance computer. Are you mad to say just to build a super computer you need investment in all that??

Yes, supercomputer is only a single unit of computer, but this computer covers almost every sub-departments of the whole industry in each country. In order to build the next generation supercomputer, almost every department is pushed to challenge its own existing technology limit SIMULTANEOUSLY. This kind of systematic pushing can bring larger advancement than the independent effort of each department alone.
Meanwhile, supercomputer just like other cut-edge technology projects-such as weapons, space craft, requires parts with much higher tech standard than those in civilian market. The technology for these projects can be easily transferred into civilian usage by lowing down the standard. But it won't be so easy to upgrade civilian products to military standard. That is why countries like to use big project like supercomputer to push their tech instead of mobile phone, lap top.

The thread topic is Only the Niche market of Supercomputers..
If you dont know what your talking about first accept it.. Please dont keep trollng with Wrong info...

Wow, you call this trolling? Did I said anything bad about India to hurt your fragile heart?

 

[cobra commando]

IISc's supercomputer SahasraT adds to Indian computing might

BENGALURU: India's most- powerful supercomputer is now housed in the city's Indian Institute of Science, strengthening the country's position in the global high-power computing race that is currently led by China. On May 11, the IISc's Supercomputer Education and Research Centre (SERC) discreetly inaugurated the Cray XC40 petaflop supercomputer, christened as SahasraT. "This unprecedentedly large, powerful computer gives our scientists the opportunity to do all the things they could not previously," IISc Director Anurag Kumar told ET. SahasraT's computational abilities has been tested with promising results. An entire landing sequence of a high-lift wing was simulated using complex physics, and so was the overlap of supernovae forming a hot over pressured bubble. "The system will help scientists in complex weatherclimate modelling, molecular and materials research and aerospace engineering. Every field has computational problems, so not a single field can be left out," Kumar said, adding that the SahasraT cost around $13 million (82.70 crore). The SahasraT overtakes Aaditya at the Indian Institute of Tropical Meteorology, Pune, which was India's fastest supercomputer ranked 71 globally by Top500, an agency that ranks supercomputers.SahasraT clocks a total peak perfor mance of 1.46 petaflops, where one petaflop is 10 to the power of 15 mathematical computations per second.But with 33.86 petaflops, the world's fastest supercomputer is the Tianhe-2 at National Supercomputer Center in Guangzhou, China.


Read more:
IISc's supercomputer SahasraT adds to Indian computing might

 

[power_monger]

I think the SahasraT is a an imported supercomputer and not an indigenious developed one.

 

[Srinivas_K]

India to have 70 super computers for high-level research

Published 24-May-2015 10:43 IST


New Delhi: The government has begun the process of having 70 super computers in the country, which will enable high-level research in different fields for not only government agencies but also the private sector. The super computer will help India do research in different fields ranging from climate, defence and other areas, which a normal computer is ill-equipped for.

The computer would have a capacity (speed) of half petaflop to 20 petaflop, but by the time the project reaches its full peak, the speed of the super computers can go up to 50 petaflop.

A petaflop is a measure of a computer's processing speed and can be expressed as a thousand trillionFLOATING POINT operations per second.
"The project, which costs around Rs 4,500 crore is in its nascent stage.

It will come up with association of the Department of Science and Technology (DST) and the Department of Information and Technology (DIT). These super computers will be in different parts of country," Ashutosh Sharma, Secretary of DST, said.

The government has formed modalities to discuss nitty- gritty of the project. However, sources pointed out that it is expected to take seven years for the project to complete.

"We will be having these computers across the country where scientists can come and do their research. These can do calculations which your normal desktop or laptop cannot. It can help in defence simulation, computing, biology and material science and earth science," Sharma said.

He said even scientists/researchers other than those affiliated to government institutions can use the super computers. "Many companies have scientists doing their research. They need not go abroad for their research and can instead use these super computers," Sharma said.

However, there are challenges before the plan is undertaken at a full-scale level. First would be building both software and hardware infrastructure for such a large-scale project and making applications for these super computers.

More importantly, having man power for running these super computers would also be a challenge as it would require training them.

• 
  http://www.eenaduindia.com/News/Nat...70-super-computers-for-highlevel-research.vpf

 

[LETHALFORCE]

https://www.armscontrol.org/issuebriefs/Science-Replaces-Nuclear-Tests

Science Replaces Nuclear Tests

 

[LETHALFORCE]

http://www.hpcwire.com/2015/03/26/india-greenlights-730-million-supercomputing-grid/

India Greenlights $730 Million Supercomputing Grid