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

 



Forgot your password?
typodupeerror
Intel Supercomputing Government Hardware

US Pens $200 Million Deal For Massive Nuclear Security-Focused Supercomputer 74

An anonymous reader writes For the first time in over twenty years of supercomputing history, a chipmaker [Intel] has been awarded the contract to build a leading-edge national computing resource. This machine, expected to reach a peak performance of 180 petaflops, will provide massive compute power to Argonne National Laboratory, which will receive the HPC gear in 2018. Supercomputer maker Cray, which itself has had a remarkable couple of years contract-wise in government and commercial spheres, will be the integrator and manufacturer of the "Aurora" super. This machine will be a next-generation variant of its "Shasta" supercomputer line. The new $200 million supercomputer is set to be installed at Argonne's Leadership Computing Facility in 2018, rounding out a trio of systems aimed at bolstering nuclear security initiatives as well as pushing the performance of key technical computing applications valued by the Department of Energy and other agencies.
This discussion has been archived. No new comments can be posted.

US Pens $200 Million Deal For Massive Nuclear Security-Focused Supercomputer

Comments Filter:
  • by pla ( 258480 )
    Wow, just imagine a Beowulf cluster of these!

    / Dating myself
    // Hurry up, Rosie, or we'll miss the movie!
  • Remember IBM supercomputer chips like the BlueGene family, Cell processor and various other POWER processors? IBM has been building supercomputers for the U.S. Government since forever and they only recently stopped making their own chips when they sold off their fabbing business to GloFo.
    • by tomhath ( 637240 )
      The Power processors in BlueGene were described as having "modest" performance. The graphics coprocessors were what made it fast.
      • No BlueGene supercomputer used GPUs as accelerators. Their point was to joind a very large ammount of cpus off, as you say, "moderate performance", but with excellent performace per watt into unparallelled computeing power density.
  • ... before they crank out the 180 petaflop score on Linpack (which officially would put them at the top of the Top 500 Supercomputer list), they're going to mine all the remaining Bitcoins. :-) Not sure that will pay for the cluster though.

    • Someone should benchmark this against Mechanical Turk. It would be interesting to see how things stack up.
    • Every 1400 blocks or something like that, the difficulty adjusts so they'd get burned as soon as they hit the next difficulty adjustment.
    • Actually the bitcoin network reached over 1 exaflop (equiv.) some time ago and has since passed that mark. As slashmydots says, the network self-adjusts.

    • Except the Bitcoin Network already runs at 324 Petahash/second, and each hash computation requires many floating point operations - 128 rounds of applying a complex hash function on several hundred bytes of data. Aurora competing for bitcoins won't make a significant difference in the network hash rate, it is too puny. The network already runs at ~1 million petaflops by dint of custom designed mining chips that perform the necessary calculations in hardware, massively parallel in each chip. Then you aggr

  • 20 years (Score:5, Informative)

    by ShakaUVM ( 157947 ) on Thursday April 09, 2015 @04:09PM (#49441665) Homepage Journal

    >>For the first time in over twenty years of supercomputing history, a chipmaker [Intel] has been awarded the contract to build a leading-edge national computing resource.

    That's bullshit. Multiple supercomputers were built for nuclear security that were constructed after 1995.

    I worked at the San Diego Super Computer Center during this time period, and could get access to them to run computations occasionally. Kinda neat.

    ASCI Red (1.3 teraflops) was built by Intel in 1997 at Sandia, upgraded to 2.4tf in 1999:
    http://en.wikipedia.org/wiki/A... [wikipedia.org]

    ASCI Blue Pacific (3.9 teraflops) was built by IBM in 1998 at LLNL:
    http://en.wikipedia.org/wiki/A... [wikipedia.org]

    ASCI Blue Mountain (3.1 teraflops) was built by SGI in 1998 at Los Alamos:
    http://en.wikipedia.org/wiki/A... [wikipedia.org]

    ASCI Q (7.7 teraflops) replaced it in 2003:
    http://en.wikipedia.org/wiki/A... [wikipedia.org]

    ASCI White (12 teraflops) was an IBM box built in 2001 at LLNL:
    http://en.wikipedia.org/wiki/A... [wikipedia.org]

    ASCI Purple / ASC Purple (100 teraflops) replaced it in 2005:
    http://en.wikipedia.org/wiki/A... [wikipedia.org]

    Red Storm (36 teraflops in 2005, 101 teraflops in 2006, 204 teraflops in 2008) was built by Cray at Sandia in 2005:
    http://en.wikipedia.org/wiki/R... [wikipedia.org]

    Blue Gene (which are a whole line of supercomputers since the 90s continuing to the present day) have been built in different places, including Argonne and have hit 17 pflops and hold half the top10 list of supercomputers:
    http://en.wikipedia.org/wiki/B... [wikipedia.org]

    I did some of my Master's thesis on the SDSC Blue Gene supercomputer. Good times.

    But yeah, anyway, the article is factually wrong.

    • The point is, it's a chip company, not a supercomputer company, that got the contract. All of your examples are of computer companies (IBM, Cray, Digital, etc.) getting the contract. In this case it's a chip company (Intel) that doesn't usually build the actual computers.

      Sure, Cray is involved. But the contract went to Intel from what I understand of the article.

      • IBM makes chips!

      • >The point is, it's a chip company, not a supercomputer company, that got the contract. All of your examples are of computer companies (IBM, Cray, Digital, etc.) getting the contract. In this case it's a chip company (Intel) that doesn't usually build the actual computers.

        Re-read my list. Intel built one of the ASCI Machines. IBM is also a chip manufacturer (they did, in fact, create the chips for some of their supercomputers). Cray and Digital were also chip makers, though IIRC Cray was out of the busin

    • Those are only the unclassified ones.

  • The computers are probably part of an effort to make "safer" nuclear bombs without nuclear testing. Our warheads are now decades old. They need to be rebuilt and redesigned to institute safer technologies. Many warheads do not have inert explosives, which means that the warheads may become dirty bombs during a fire. We have created insensitive munitions that will not explode even when dropped or burnt. These newer explosives have different properties that require testing with computers to simulate.

    • Actually we have a lot of nukes which could detonate (i.e. a full yield nuclear detonation) accidentally. It's a common misconception that nukes can't accidentally detonate. This is only true with designs that use multi-point explosive lenses. But to make warheads small enough to fit inside a missile, two-point detonation schemes were developed, and these have a very significant likelihood of accidentally going off in just the right way to initiate a nuclear chain reaction. A large part of current nuclear s

      • by gtall ( 79522 )

        The only sane option is mutual disarmament and it is chimera, it will never be realized because you'd have to get India and Pakistan to agree. And even if you did, Putin would keep several back and threaten the West with them. He's recently made such threats. Also, every two-bit Islamic My-Dick's-Too-Small Head-O-State would realize that with a secret program, they can break out and threaten the West with Convert or Die. Allah is a cruel, unfeeling god.

        Just kidding about Allah, he's so Other that he never c

        • The only sane option is mutual disarmament

          Even if this were an option, the end result is still that more nukes will be built.

          If nobody in the world has nukes, an obvious good strategy is for your side to start building them. This is just a high stakes multi-way prisoners dilemma. In multi--way prisoners dilemma there are too many players for there to be a reasonable expectation of full cooperation.

          The best result is if all of the nuke owners don't use them, because there being nuke owners are inevitable. Thats what we've got right now, No sens

          • I envy your simplistic view of the world. People like you try to present a false and dichotomous view. That we have to choose between nukes and annihilation, because "prisoner's dilemma doncha know."

            There are two main reasons that you're wrong. 1. Real life isn't as psychopathic as you make it out to be. During the cold war, the USA had several very good opportunities to take out the USSR with a nuclear strike but it never did. Or do you think only Americans are human?

            Regardless, even if you completely lack

            • I envy your simplistic view of the world.

              its not simplistic. its just that you don't see or understand the complexity, for instance everything else you wrote had nothing to do with what I said. You didnt understand what I said at all, obviously. So since you require simplicity, let me lay it straight:

              If everyone dis-arms, then players will eventually re-arm because its the right strategy. Its the right strategy because its inevitable that someone will and there are advantages to being the first ones to do it (if you don't believe this, ask Iran

              • No, I understood your point, and my second point was a direct response to it. You have a simplistic view of the world because you heard about game theory once and think that your childish interpretation of the prisoner's dilemma can be used as a direct substitute to understand something as complicated as global politics and nuclear deterrence!

                I'm not even going to reply to you again, just advise you to re-read what I wrote.

                • No, I understood your point, and my second point was a direct response to it.

                  No it wasn't. Your second point just hopes that only a few, and I quote, "dinky nukes" will be built and that the the rest of the world will strike down this country before any more are built because the only countries that can build them post-disarmament are, and I quote, "rogue states."

                  Pure fantasy.

                  (A) No country that has built nukes has ever built "just a few dinky nukes" - so you are imagining a world that doesnt even fit objective reality.
                  (B) Every country has friends, even "rogue states", and it

                  • I think I figured out what your sticking point is. You completely lack knowledge of what nuclear weapons are and how they work. You seem to have this idea of nukes as some mythical all-destroying force. You seem to think that once a country develops nukes it instantly has the power to flatten Washington DC or New York, and can start dictating terms to everyone.

                    > No country that has built nukes has ever built "just a few dinky nukes" - so you are imagining a world that doesnt even fit objective reality.

                    Ac

                  • In addition to my reply above, I'm actually going to say that I agree with you that disarmament and non-proliferation is a fantasy that probably isn't going to ever happen (at least not until there's a full-blown nuclear war, which is inevitable). But that's only because of the existence and influence of insane RWAs such as yourself.

  • The eternal question: Does it run Linux?

  • More proof that Obama is into bitcoin mining.

To err is human, to moo bovine.

Working...