Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Supercomputing

India Plans To Build Fastest Supercomputer By 2017 135

Posted by samzenpus from the top-of-the-list dept.
First time accepted submitter darkstar019 writes "India is planning to build a computer that is going to be at least 61 times faster than the current fastest super computer, IBM Sequoia. Right now the most powerful supercomputer in India is 58th in the list of top 100 supercomputers. From the article: 'Telecom and IT Minister Kapil Sibal is understood to have written to Prime Minister Manmohan Singh sharing the roadmap to develop "petaflop and exaflop range of supercomputers" at an estimated cost of Rs 4,700 crore over 5 years.'"
This discussion has been archived. No new comments can be posted.

India Plans To Build Fastest Supercomputer By 2017 More

India Plans To Build Fastest Supercomputer By 2017

Comments Filter:

  • Nonsense (Score:1, Insightful)

    by GeekWithAKnife ( 2717871 )
    It would be far wiser If they just spend all those millions on poverty programs. [slashdot.org]

    • Re: (Score:2)

      by nashv ( 1479253 )

      Obligatory "a-country-which-I think-it-underdeveloped-and-full-of-poor-naked-children-is-thinking-of-beating-us-at-hi-tech-stuff-zOMG!" troll.

    • It would be far wiser If they just spend all those millions on poverty programs.

      Countries have an interest in making their citizens feel proud and confident. Building supercomputers and spacecraft are much better ways of doing that than the traditional methods of starting a war with Pakistan and/or detonating another atomic bomb.

      • I'd more confident if the US had a higher standing in education, or a lower poverty rate, or lower infant mortality. But hey, one mans asshole is another....something like that.
    • Wait Mr. GeekWithAKnife. I don't know who you are but this is worst insightful text one could type for this post. First delete your impression about India in your mind If(you==Indian){ I feel sorry about the fact that you didn't have clear idea about your nation.} else{ Don't judge India by movies. Visit India and live there and then see what is actually happening there. } Note:It is not the land of homeless people (There are lot of homeless in US but not in Canada). In India, there are about one third of

  • Crore = 10^7 (Score:2, Informative)

    by Anonymous Coward

    http://en.wikipedia.org/wiki/Crore

    Nice to see the editors making sensibly proof-read, accessibly written summaries, rather than the usual treasure hunt for the true meaning.

    • Re: (Score:2)

      by 1s44c ( 552956 )

      Indeed. The weird Indian numbering system is meaningless for almost everyone outside India. The more common three orders of magnitude system ( thousand, million, billion, trillion ) makes a lot more sense.

      • Re: (Score:3)

        by nashv ( 1479253 )

        It's not that hard. The only difference being that the Indian system uses 10^ odd numbers, more often primes, as a reference.

        And thus : 10^0,1,2,3,5,7 all have names - 10^5 being a lakh and 10^7 being a crore. A complete list is rather interesting, showing that the system predates Western mathematical formulations: http://en.wikipedia.org/wiki/History_of_large_numbers [wikipedia.org] and having pecularities like bodhisattva ( or ) —10^37218383881977644441306597687849648128.

        You gotta ask...why 10^372183838819776444413

      • Indeed, a bit like the Imperial (non-metric) distance measuring system is meaningless for almost everyone outside the USA. The more common decimal three orders of magnitude system (millimeters, meters, kilometers) makes a lot more sense. But we try to get by.

        • Re: (Score:2)

          by 1s44c ( 552956 )

          Indeed, a bit like the Imperial (non-metric) distance measuring system is meaningless for almost everyone outside the USA. The more common decimal three orders of magnitude system (millimeters, meters, kilometers) makes a lot more sense. But we try to get by.

          This is of course true. Everything is easier in metric so hanging onto a dumb system that everyone else has abandoned due to its awkwardness is just, well, dumb. But at least most of the world had some idea what a mile or yard is, if they don't know how long is it they at least know it's a measurement of distance.

          Crore is just meaningless to almost everyone.

  • 1 crore is 10 million, so this comes out to be 875 million USD roughly.

    http://arstechnica.com/information-technology/2012/06/with-16-petaflops-and-1-6m-cores-doe-supercomputer-is-worlds-fastest/ [arstechnica.com] says that livermore spent 250 mil on sequoia (which seems like a bit of a lowball to me, given the K computer's price at 1 billion), so throwing a lot more money at the problem would seem to give better performance.

    • Crore is a counting number i should say. 1 crore of something is 10 million of that something, not 1 crore INR = 10 million USD.

      • Re: (Score:3)

        by Matheus ( 586080 )

        You are correct sir. Although apparently since I was last there the value of the Rupee has fallen *severely so...

        At current exchange R4,700 Crore ~= $87,231,060.00 USD.

  • Yawn (Score:2, Insightful)

    by Anonymous Coward

    More dick waving.

  • Obsolescence (Score:2, Interesting)

    by Anonymous Coward

    The best part is by the time 2017 rolls around other countries would be doing the same so their fast computer turns out not to be THE FASTEST.

    • Re: (Score:2)

      by 1s44c ( 552956 )

      By the time 2017 comes your iphone 20s will be about as fast as this supercomputer.

  • Considering Moore's law (Score:4, Informative)

    by SpaghettiPattern ( 609814 ) on Monday September 17, 2012 @11:05AM (#41363561)
    Considering Moore's law that's just about to be expected.

    In 5 years we have 3 x 18 month period. The level of improvement in hardware is multiplies by 2^3. Then I'd expect level of parallelism to affect the process by the same magnitude bringing the total to 2^6 = 64.

    • Re: (Score:3)

      by tlhIngan ( 30335 )

      Considering Moore's law that's just about to be expected.

      In 5 years we have 3 x 18 month period. The level of improvement in hardware is multiplies by 2^3. Then I'd expect level of parallelism to affect the process by the same magnitude bringing the total to 2^6 = 64.

      Except Moore's Law only indirectly applies. All it states is that the number of transistors you can squeeze doubles every 18 months or so.

      Number of transistors only really is a passing indication of CPU power. Especially these days where the th

    • Sorry, but you're milking the cow twice: process shrinks allow us to pack more transistors on a chip. This would amount to a growth of about 2^3 = 8 in a period of 5 years, as you correctly estimate. But this already includes the increase in parallelism. Today's supercomputers apparently don't grow much more racks:
      • Roadrunner: 296 racks
      • Jaguar: 150 racks IIRC
      • K computer: 768 racks (huge exception)
      • Sequoia (Blue Gene/Q): 96 racks

      The reason why we can't just buy an infinite amount of racks and netw

  • For these 'fastest' metrics: (Score:3)

    by fuzzyfuzzyfungus ( 1223518 ) on Monday September 17, 2012 @11:11AM (#41363629) Journal

    How well do the 'fastness' metrics used to rank computers in e-peen order capture some of the messier variables of assorted cache speeds and sizes, latencies and throughputs of network interconnects, dubiously general; but very high speed for certain purposes GPU or fpga elements vs. generic CPUs, and so on?

    Obviously, the people building these things to get work done have an incentive to make them actually useful; but is the benchmark itself much of a test of dreadful interconnect design or other serious issues, or could you just buy your way to a shiny spot at the top by shoving together enough gigE connected 1Us?

    • Re:For these 'fastest' metrics: (Score:5, Insightful)

      by serviscope_minor ( 664417 ) on Monday September 17, 2012 @11:28AM (#41363841) Journal

      Depends on the test.

      The classic LINPACK benchmark will stress most of the parts which will mean the result is a combination of raw FLOPS with memory bandwidth, cache and etc. LINPACK doesn't stress the interconnects particularly highly and is very regular. As a result, it tends to favour computers that have more FLOPS but cheaper interconnets.

      That said, it's not terrible, which is why the computers also have the efficiency (theoretical peak FLOPS/actual flops) listed. Compare the Tinhae-1A computer which was heavy on GPUs versus with 46% efficiency the K computer which has lots of wide SIMD cores with a very tightly coupled interconnect which achieved 93%.

      So even LINPACK which is generally considered as "too easy to be a useful test" still can distinguish between raw peak FLOPS and sustained performance.

      In practice, some tasks will depend heavily on the interconnect. Others, like protein folding are so embarressingly parallel that the interconnect basically a non-issue which is why floding@home works.

      IOW YMMV HTH HAND

    • More importantly, what do these supercomputers do? i love computers as much as the next guy (or gal) on slashdot, but i'm unaware of anything that they do. they don't seem to be involved in landing robots on mars. they don't seem to be involved in making a self driving car. I guess they will do environmental simulations and then show you the results you want after you tweak the input data enough. Oh, and they can win at jeopardy and chess.

      • Re:For these 'fastest' metrics: (Score:5, Informative)

        by cwebster ( 100824 ) on Monday September 17, 2012 @12:14PM (#41364353)

        Modeling.

        Weather modeling (solving navier-stokes and a few other equations on a discrete cartesian grid or on a spherical grid in spectral space). Add in land surface models, ocean models, data assimilation, chemical processes, and then crank the resolution way up and you need a lot of power.

        DNS (direct numerical simulation) -- if you want to simulate a fluid flow with turbulence and you want to resolve the turbulence explicitly you need to have a grid spacing in your model that is smaller than the kolmogorov scale. For some flows this may produce a grid spacing measured in millimeters. If you want any decent sized model domain, this produces a lot of grid points.

        Monte-carlo type simulations -- i.e., run a simple simulation but do it 1e50 times to amass a statistical representation of the process.

        and lots of other types of modeling. Basically if you have a set of partial differential equations that tell us something and you need to solve them numerically (no analytic solutions, etc) and need to do it on very large domains at high resolution and your neighbor grid dependencies are such that your problem is parallel, then a supercomputer is for you.

      • They model protein folding, cosmic events such as the big bang, creation of black holes, stellar collapse and creation,they are used in modeling atomic bomb blasts, and better electrical grids, they are used to work out highly complex math and number crunching such as finding holes in encryption schems, they are used for analyisis of high speed partical excelerator data. Pretty much anything that requires lots of compex math.

      • A few points:
        * The weather report
        * Finding oil and gas
        * Supernova research
        * Big bang simulations
        * Designing rockets for the space program
        * Simulating nuclear fusion

        In general, solving large problems where there is turbulence, other complicated phenomena or the data set is hard to search

      • They are, in fact, used for landing robots on Mars. I worked on a supercomputer with my professor during my physics undergrad working on fluid-structure-interaction (FSI) code. The supersonic parachute used to land Curiosity was simulated using a FSI code, a simulation which my professor helped with. Cars these days often use fluid dynamics during the design process and structure code as well (which can be just as complicated when you're simulating a collision, as is often done these days).

        There's really a

  • Difference between 2007 and 2012 was only 34x grea (Score:3, Interesting)

    by Turboglh ( 816701 ) on Monday September 17, 2012 @11:11AM (#41363631)
    http://www.top500.org/lists/2007/11 [top500.org]
    http://www.top500.org/lists/2012/06 [top500.org]
    Should be interesting to see them double the rate of growth over the preceding five years

  • And there will be (Score:5, Insightful)

    by onyxruby ( 118189 ) <onyxruby@ c o m c a s t . net> on Monday September 17, 2012 @11:19AM (#41363735)

    One supercomputer to outsource them all

    • Re: (Score:3, Funny)

      by account_deleted ( 4530225 )
      Comment removed based on user account deletion
    • I have read that the Sequoia supercomputer can do 2 Gigaflops per watt, I think this is at least an order of magnitude greater than any home computer. So when they get into the Exaflop range I would think that they would again be better by an order of magnitude. So that would mean they could do 2 Teraflops per watt. My question is what are they going to do with all this power? I would think that the federal government would get a Watson like computer for both the executive and legislative branches. I

  • Please it is a Joke (Score:1)

    by Anonymous Coward

    Our Minister Kapil, is a good comedian. So please dont take this seriously.
    Check out what happened to his $35 laptop! http://en.wikipedia.org/wiki/Aakash_(tablet)
    Our closed room scientists in CDAC will have plenty of money to play around for 5 years!
    Thank god, Minister Kapil wont be there that long.

    • Our Minister Kapil, is a good comedian... Minister Kapil wont be there that long.

      hey... celebrating world optimism day today are we?
      you see, just like managers, these dickheads are expected to come up with "next big idea", spent a lot of money and move on. and that is what this is, just like the akash tablet project, the complete literacy project, remove proverty project and what not.

  • PARAM and beyond (Score:5, Informative)

    by unixisc ( 2429386 ) on Monday September 17, 2012 @11:57AM (#41364175)

    India's current supercomputer - one that it's developed since the 80s - is the PARAM [wikipedia.org], which has had 6 generations to date. The first was based on the Inmos Transputer, the second on an Intel i860, the third on a SuperSPARC II (and it even had an Alpha variant), the fourth on an UltraSPARC II, the fifth on an IBM POWER 4, and the most recent - unveiled in 2008 - was based on the Intel Xeon (Tigerton 73xx). They are currently working on one that's supposed to break the 1 petaflop barrier (that would be 10 crore crore flops for Indians). So this new announcement would be the successor to that.

    So it's not like they're new @ this, and what is impressive is that they've used a wide variety of CPUs from different vendors. For this next one, they might want to do that w/ an Itanium III or a POWER7 (unless POWER8 is anywhere close). It would seem that for that, they might get some Intel/HP expertise to help w/ that. I have no idea how good they are @ writing compilers. But yeah, planning a supercomputer based on this CPU and tossing in enough of them should enable them to achieve that goal. Put Debian on it, and then use it for whatever they need - weather forecasting, nuclear simulations or whatever they want to use it for. A lot of the 52 PARAMs that they've manufactured & sold have been sold to other countries.

    I just wish that aside from the Indian government, there were a few companies in India that made supercomputers.

    • Comment removed based on user account deletion
  • The problem with India is, "India wants to do it". It is not the scientists or researchers in a university or institute in India who will do it, but a telecommunication minister. Till this mentality dissipates and the government bets on the independent institutions in India to come up with such headlines, India is going nowhere.

    • The problem with India is, "India wants to do it". It is not the scientists or researchers in a university or institute in India who will do it, but a telecommunication minister. Till this mentality dissipates and the government bets on the independent institutions in India to come up with such headlines, India is going nowhere.

      Yeah. As I sit here posting on the internet via the World Wide Web I have to ask, what innovations has government ever produced?

    • Aside from the six previous supercomputers their government has built, you mean?

      True, there are some things the private sector is better-suited for. But you make the mistake of infering from this that governments are incapable of doing anything.

  • I don't really see a distinction. Computer chips keep adding cores. It's all networked together at varying speeds be it bus or Ethernet signaling. It's all cloud one way or another.

  • They're going to build it completely out of Nan. It will be fast and delicious!

    • Re: (Score:1)

      by CQDX ( 2720013 )
      And when the benchmarks spit out a bunch of NaNs, I don't think the supercomputing community will be impressed.

    • Re: (Score:2)

      by fatphil ( 181876 )
      I hope the result of the computation won't be NaN.
      I expect to see the results published at the HotChipatees conference.
      I wonder if they'll use off-the-shelf processors, or build their own Sag-ALU, or even utilise Ghee-P.U.s

      Thank you, thank you, I'm here all week.
  • So Doc Brown is going to bring us the MR Fusion Upgrade -- supposed to have taken place in 2012. Maybe India could use the MR Fusion to power their new Super Computer?? It would kill 2 birds with 1 stone...........Just sayin.
  • Yeah, this is just about as believable as the Aakash being produced for $35. That they SAY they're going to do it shouldn't be read to imply they actually can.
  • Just run a distributed computing client on every call center machine...

  • Basic necessities, infrastructure and other issues that people pointed out are issue in India - but how does that equate to wasting money on building a supercomputer? Would you rather have it that they spend much more buying when they can build it locally for cheaper? Also after all the dust of 'they should rather help the poor instead' argument has settled you can see that despite so many issue they still need to predict the weather.

  • The politicians in India are talking about investments in space and supercomputers because its sexy and gets people's attention. Whether or not these investments are going to materialize, or if they're even wise, remains to be seen. Perhaps all my fellow Americans will read about an Indian supercomputer one day, while we enjoy the comforts of our future lives on Moonbase Gingrich.

  • I hadn't realized that you needed so much computing power to run a call center.

Slashdot Top Deals

To stay youthful, stay useful.

Close