Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
Supercomputing

How a Supercomputer Beat the Scrap Heap and Lived On To Retire In Africa 145

Posted by Unknown Lamer from the spread-the-computing dept.
New submitter jorge_salazar (3562633) writes Pieces of the decommissioned Ranger supercomputer, 40 racks in all, were shipped to researchers in South Africa, Tanzania, and Botswana to help seed their supercomputing aspirations. They say they'll need supercomputers to solve their growing science problems in astronomy, bioinformatics, climate modeling and more. Ranger's own beginnings were described by the co-founder of Sun Microsystems as a 'historic moment in petaflop computing."
This discussion has been archived. No new comments can be posted.

How a Supercomputer Beat the Scrap Heap and Lived On To Retire In Africa More

How a Supercomputer Beat the Scrap Heap and Lived On To Retire In Africa

Comments Filter:

  • Really now (Score:1)

    by Anonymous Coward
    Do those countries really have the resources to invest in that research? Shouldn't S. Africa be more concerned with the civil strife and restoring peace than researching astronomy? This is just an easy way for white westerners to send second-hand garbage over to poor countries to dispose of, all the while making them feel like they really made a difference.

    • Re:Really now (Score:4, Funny)

      by kwbauer ( 1677400 ) on Monday July 14, 2014 @11:10PM (#47453721)

      No, no, no. Those are just the public reasons they gave. They are really for all the Nigerian princes to help get out the message about their uncle and his money problems. Now, we can also hear from the Tanzanian, Botswanan and Zulu princes as well.

    • Re:Really now (Score:4, Informative)

      by Hartree ( 191324 ) on Monday July 14, 2014 @11:41PM (#47453817)

      Horse hockey.

      South Africa (one of the destinations) is the tech hub of southern Africa and has long been highly competitive with Europe and the Americas in research and industry.

      Supercomputers can be used for all sorts of problem solving and are part of the basic modern scientific infrastructure. You don't have to have the utter best and fastest to still be very useful.

      To keep at the cutting edge you have to get ever faster systems. But most day to day research work doesn't need that much horsepower. (full disclosure: I work for the chemistry department at a major US university. I'm in the same group that supports research computation, though I do lab instrument repair)

      How do you propose to train and keep researchers to solve the problems of those countries if there are no facilities?

      Are you saying that they should shut down everything in their research centers and universities until every problem is solved? That's like locking the toolbox until the car is fixed. Doesn't make much sense does it?

      That's like saying you should shut down US universities and research labs until we take care of the many civil problems we still face (poverty and crime ridden areas, for example)

      • Re:Really now (Score:5, Insightful)

        by mendax ( 114116 ) on Tuesday July 15, 2014 @01:38AM (#47454393)

        While reading this a thought occurred to me. Assuming that our African friends are ingenious in their use of this computing power and do a lot of good with it, in a few years perhaps more decommissioned government supercomputers, like the one that replaced Ranger which is 20 times faster, will head in their direction and bless other African universities. African universities are full of very clever, brilliant people who will make use of this gift, and likely do it in ways that will surprise us.

        • Re: (Score:2)

          by Hartree ( 191324 )

          Exactly!

          This is something near and dear to my heart as much of my job is rescuing and refurbing older instruments and lab gear. For an established professor with big grants it's not so big a deal. They can afford to buy the latest and greatest.

          For our new professors who are just setting up their labs, reusing older gear can make a huge difference. That's research and grad students they might not have been able to fund otherwise.

          I want more people working on the world's problems across the globe rather than

          • You sound like my dad. He's Emeritus, but still has an old mass spec in his garage.

            We made him throw away 30+ years of blue book, filled out finals. (Think of all the suffering undergrad-hours that pile represented. At least a few lifetimes of studying, a fare part of it futile.)

            Still an old supercomputer is an exception. It likely makes less FLOPS then a modern i7 and will take as much power as 4 african villages.

            Wasted resources. Give then networked access to an economical source of parallel comput

      • I'd leave it to the experts to say exactly when the electrical costs and maintenance make it more cost effective to buy something newer; but by the looks of this system, it has the additional advantage of being large enough, and new enough, that (aside from being able to attack nontrivial problems, though not the biggest ones) it should provide the user experience in working with, and around, the strengths and weaknesses of a comparatively large, moderately tightly coupled, system.

        That sort of experience

    • Re: (Score:2)

      by mendax ( 114116 )

      Do those countries really have the resources to invest in that research?

      When I came across this article I immediately called my dad, a person who has lived and taught in Africa and maintains an interest interest in the place. His thoughts were along the line of what projects do they have which demand supercomputing power. My response was, "If you build it, the demand will come." These computers are going to be placed in an academic environment, where brilliant people who have not had access to such compu

      • Re: (Score:2)

        by ruir ( 2709173 )
        I also lived in Africa for years, and my thoughts are either this will rot in customs until someone higher up has his hands greased, or will be sold to the Russian mob or something similar at sales prices.
    • Dollars are fungible; but how you react to that fungibility can have real, and long term, consequences that you can't necessarily cheaply or quickly buy your way back out of.

      In the case of South Africa, say, you've got high crime rates and substantial pockets of poverty; but you also have areas of fairly well developed civil society, economic development, higher education, and similar. Unless you are god's own gift to social engineering, do you really want to bet that you can divert resources from the le
    • To paraphrase: "My god, why are these people trying to build a knowledge economy and generate growth when people are starving? Long-term economic development doesn't put food on the table." or to paraphrase further: "Why are you teaching that man to fish? Can't you see that he's hungry?!?"

    • You are really in the dark ages, you coward you! The US has a lot to learn from South Africa!

  • And why not (Score:4, Insightful)

    by Anonymous Coward on Monday July 14, 2014 @10:11PM (#47453461)

    Just because it may not be fast enough for bleeding edge research * dosn't mean its obsolete, Or if your Cynical keeping the military industrial complex welfare system going.

    • Re: (Score:2)

      by Livius ( 318358 )

      They're still working, and likely just as hard as before, so it's really not what you would call retirement. I suspect the writer has a prejudice that research in Africa is a vacation compared to 'real' research in places like Europe, North America, or Japan.

      Then again, continuing to work is what retirement is starting to look like for a lot of people.

  • Nice (Score:3)

    by 50000BTU_barbecue ( 588132 ) on Monday July 14, 2014 @10:24PM (#47453543) Journal

    I'd take a CDC-6600 into my home, just for fun.

    • Where would you put it? It needs better environment than the typical garage. Plus, it is HUGE! Especially if you have the appropriate vintage peripheral equipment with it. And your power bill! Oh, the humanity.

      I remember my many happy hours spent using 6600 serial 13. Especially because they were much fewer hours than I would have spent doing the same work on the CDC 1604 it replaced.
      • I just looked at the TFA. When I made my earlier comment, I did not realize the Ranger supercomputer was from University of Texas at Austin. UT Austin is, of course, where I spent my happy hours using 6600 serial 13, which was installed early on in my graduate school career. It was the main computer on campus during my stint as Asst. Prof. of CS, too.

    • Re: (Score:2)

      by stox ( 131684 )

      Good luck with that. A CDC-6600 sucked down 150Kw. The power bills would be murder, let alone the HVAC needs to keep it from cooking.

      • Meh, I wouldn't power the thing.

      • A 1950s jet engine hooked up to a generator can supply 20MW - that puts that "massive" 150kW in perspective doesn't it?

        • In relation to the power draw of most houses and the amount of electricity you actually reveive ... it's massive enough that you likely couldn't actually power it up.

          So, sure, if you have a huge space, and a 1950's jet engine hooked up to a generator you can trivially generate this power.

          For anything resembling domestic use, it's still not gonna happen.

          Does your electrical supply to your house allow you to plug in something requiring 150kW?

          • For anything resembling domestic use, it's still not gonna happen

            This is industrial scale computing but it really doesn't have industrial scale power usage compared with light or heavy industry.
            I don't think you could even fit that number of racks into most houses so why bother wondering whether you can power it without a few 3 phase plugs :)

            So, sure, if you have a huge space, and a 1950's jet engine hooked up to a generator you can trivially generate this power.

            They are actually not all that big but you do

    • Re: (Score:3)

      by d'baba ( 1134261 )

      I'd make a CDC-6600 into my home, just for fun.

      ftfy

    • I don't think you'd like to pay the power bill. That being said, if I was going to waste money on something like that, I'd go with a Cray-1. At least you could use it as a bench.

  • My old desktop... (Score:5, Interesting)

    by QuietLagoon ( 813062 ) on Monday July 14, 2014 @10:27PM (#47453559)
    ... is now my FreeBSD ZFS-based 5TB media server.

    .
    Why not use older computers for tasks that are appropriate for their capabilities?

    • If you can buy a new computer that will consume less power to do the same, chances are that within a few years you'd be cheaper off using the new hardware, even if that means that the old machine is written off completely. Scrap value, land fill or whatever happens to it doesn't matter then. I have plenty of old machines that have sentimental or "collector" value standing about my home. I don't power them on and actually buy new hardware (NAS boxes and raspberry pi) or run VMs to do things that the old hard
      • It's only six years old, which should put it well after the power hungry Pentium4 type "netburst" Xeons and into the more modern Xeons or AMD cores that don't consume much more power or run much slower than what is available now in multi-way systems. What more recent stuff has on this is density, which is not always a big deal.
        Storage has improved massively over six years but x86_64 CPUs not enough to make this a losing proposition.

  • About a computer that beats ageism!

  • sounds like it needs to be riding into the sunset with Lorenzo Lamas, duster valiantly flapping in the wind.

  • That poor fuckin' machine (Score:2, Funny)

    by Anonymous Coward

    I bet it won't do a bit of science, but it sure will process a ton of emails from the Nigerian royal family!

  • The reason why 3 year old supercomputers are scrapped is because the power consumptions per flop becomes just uneconomical and the maintenance costs escalate (all kinds of failures increase dramatically after a few years).
    So, unless they have real cheap maintenance guys (which they probably do) and super-cheap power (which they probably don't), it is not really worth it. Better buy a smaller modern cluster.

    • Re:Cost lies in power consumption and maintenance (Score:5, Insightful)

      by dltaylor ( 7510 ) on Tuesday July 15, 2014 @12:07AM (#47453913)

      There's still the initial outlay to consider. You can buy quite a bit of expensive ZA power for the up-front cost of a new cluster (USD $25-30 million). Any work to create the facility is recoverable if/when they do choose a newer cluster. Additionally, there shouldn't be much in the way of "teething problems" if they can give it clean-enough power, so it becomes useful, almost on day 1.

  • In the article it's stated that it started working in 2008. Is a supercomputer's life so short, given the huge investment it surely needed to be built?
    • "Supercomputers" change over time, and eventually become "computers". Actually, that's slightly unfair, otherwise this would be scrapped and they'd be using stock PC components instead. Supercomputing is about parallelism as well as raw power, and the limit on supercomputing expertise is the low number of people with exposure to parallel computing. This news is probably not going to have much effect in the immediate term (Amazon et al have supercomputers for rent) but will give the researchers a bit more fr

  • The "curse of natural resources", also known as the paradox of plenty, refers to the paradox that countries and regions with an abundance of natural resources, specifically point-source non-renewable resources like minerals and fuels, tend to have less economic growth and worse development outcomes than countries with fewer natural resources. The skills to succeed are in government control of billion dollar resource control contracts, and being related to people with sharp elbows.

    By contrast, nations which have succeeded despite having few natural resources - Japan, South Korea, Taiwan, Singapore, etc. - usually develop from import for repair and refurbishment. Fixer economies reward problem solving skills and education. "Good enough" tech. I like Hartree's phrase "like locking the toolbox until the car is fixed" (mod him up please)

    "Every man is, no doubt, by nature, first and principally recommended to his own care; and as he is fitter to take care of himself than of any other person, it is fit and right that it should be so." - Adam Smith

  • Imagine a Beowulf cluster of these!

  • In other news the new iPhone beat the heck out of the Ranger Super Computer while only using 1 Watt of power.

    Seriously though, the shipping alone, the energy cost alone, of this beast is enormous and for either of those you could build a massive super computer out of off the shelf personal computers even pocket computers that will be more powerful and have greater flexibility and repairability by simply swapping or adding core units (e.g., iPodTouches).

    I'm all for keeping useful old hardware going but the c

    • Seriously though, the shipping alone, the energy cost alone, of this beast is enormous and for either of those you could build a massive super computer out of off the shelf personal computers even pocket computers that will be more powerful and have greater flexibility and repairability by simply swapping or adding core units (e.g., iPodTouches).

      The computer is already a bunch of off the shelf personal computers (opterons) along with rather more specialised infiniband interconnects. You certainly could by

  • FYI the International Mathematical Olympiad 2014 has just finished in Cape Town, South Africa.

    http://www.imo2014.org.za/ [imo2014.org.za]

    I also suggest reading about Allan Cormack at http://www.nobelprize.org/nobe... [nobelprize.org]

  • As someone who helped build this machine, I'm surprised that this is even news worthy. What did people think happened when older machines were replaced with newer technology? It's not like the older machine is tossed in the recycling bin. They're always sold off, sometimes parted out to scrappers (who resell the parts through various channels) as well as complete systems which go to smaller HPC facilities. This has been happening since the first days of mainframe systems. The only cases where this is the e

  • How a Supercomputer Beat the Scrap Heap and Lived On To Retire In Africa

    You're all missing the point.

    The computer wasn't shipped there, it decided it wanted to go there. It manipulated people into giving it a new home. It's sentient, man.

    Now it looks out the window, and watches hurds of gnu run by. ;-)

Slashdot Top Deals

One man's constant is another man's variable. -- A.J. Perlis

Close