Slashdot Mirror

← Back to Stories (view on slashdot.org)

NASA Installs Linux Supercomputer

Posted by Cliff on from the penguins-exploring-the-ocean dept.

unassimilatible writes: "Federal Computer Week reports that NASA plans to study the ocean's future with the help of the world's first supercomputer of its kind to run on the Linux operating system. The new supercomputer -- an SGI AltixT 3000 single-system image supercomputer -- has been installed at the space agency's Ames Research Center in California."

37 of 189 comments (clear)

  1. I can almost.. by aaron_ds · · Score: 2, Funny

    Imagine a beo...oh never mind.

  2. Uh-oh... by Hanzie · · Score: 5, Funny

    The article explicitly says they're using NUMA archeticture.

    Obviously, it's running SCO's intellectual property. SGI doesn't really own NUMA, they only wrote it. Deep down, it's really a derivative of vi.

    --
    ********* sig: If you don't like the law, get filthy stinking rich, and buy a better one.
  3. Altix by Preach+the+Good+Word · · Score: 5, Informative

    SGI's Altix handles up to 64 processors on a Linux kernel using the patches they release as opensource. As SGI hacks away at their bigmem and numa patches, they'll be able to handle more and more processors. The plan is to eventually graft enough IRIX technology to support just as many processors on Altix as they do with MIPS processors in Origin with IRIX.

    Even if you aren't a fan of Itanium2, Linux, or NUMA, these patches are bringing some nifty high-end tech to the free software arena.

    1. Re:Altix by rf0 · · Score: 2, Insightful

      The thing is this Altix is 256 CPU machine in a single system image. They are looking to take this to 512 in a single image. That is some serious scaling

      Rus

      --
      Cheap UK and US VPS
    2. Re:Altix by Anonymous Coward · · Score: 5, Interesting

      The only IRIX technology they're grafting is the SCSI layer. With the 2.6 kernel, there really isn't that much they need to do. They're testing their 512 processor systems on linux 2.6 as we speak, btw.

    3. Re:Altix by GooberToo · · Score: 4, Informative

      Tons and tons of work has gone into 2.6 to make it fully NUMA aware and scalable. Last I heard, there were still some minor memory allocation node biases left in the scheduler, however, those are actively being worked on. Furthermore, they were considered corner cases and not general NUMA processing problem domains.

      IIRC, one or more the developers has a 128 (or larger, I forget) CPU NUMA system that 2.6 periodically gets tested against. Many smaller NUMA systems are commonly used by several others. It seems that many, but not all of the NUMA optimizations, also help SMP systems as well. As such, the developers have not been shy about embracing it. When the O(1) scheduler was writen, it was a very short period of time before they started adding HT and NUMA optimizations to it.

    4. Re:Altix by EyeSavant · · Score: 3, Informative
      The technology is fantastic. It will be even more fantastic when it works :-(. At the moment you have to be nuts to buy one of these, as they are complete bleeding edge technology. In Amsterdam they "upgraded" from a 1000 processor mips machine to a 416 processor Altix/Itanium2 machine. On paper an itanium processors should be 5 times faster than the old mpis processors. At the moment we are lucky to see 2 or 3 speedup. And that is AFTER you have tuned the damn things for itaniums.

      It seems to be quite normal for code to run slower on the new machine than the old... So you have a tough porting job to do...

      The problem with the itaniums is that all the hard optimisation stuff has been moved from the hardware to the compiler. Plus the compiler is buggy as hell at the moment... I am currently trying to track down why my code fails on the new machine when I turn the optimization on.. My money is on a compiler bug to go with the many other compiler bugs we have seen before.

      I am sure they will be a nice machine eventually, but if you buy one now you are nuts.

    5. Re:Altix by EyeSavant · · Score: 2, Interesting
      Is this speedup per processor or overall on your application? In any case, any time you move to a new system, you may have to do some tuning of your software for the new machine. For example, the Sun E10K+ has memory latencies very similar across the system. You can write code that works fine there, but when you move to a NUMA machine it may run very poorly because of memory access patterns and memory locality to the processor. You'll need to tune your software to make it run better.

      The speedup I quoted is per processor speedup, but AFTER they had done some tuning. After you just port but before the optimization it seems quite normal to get slower code rather than faster... but from what you said that is fairly common. I am not sure it should be so expected going from one NUMA MIPS machine to a NUMA Itanium2 machine. Apart from the fact that the Itaniums are not very well balanced..

      I do think they will get it right eventually, but it is still bleeding edge stuff unfortunately.

      Note this is what the computer people at the computer centre told me not my personal code. The current state of my personal code is that it runs at 1/4 of the per processor speed on the new computer, (i.e it is running about about 1/10 to 1/20 of the speed you would expect) and gives the wrong answer... Clearly those things could be related... but I am destinctly underwhelmed at the moment.

      So... since it's "easy" to be done in hardware and "hard" to be done in software, hardware engineers are better at software optimization than software guys? ;)

      No that is not quite what I meant. The best way has to be some combination of the two IMNSHO. In the Itanium they started with a blank piece of paper, and asked the hardware guys what they wanted. Like normal engineers they made the tough stuff (the optimisation) someone elses problem. All the itaniums do is do exactly what the compiler tells them to do. So the compiler has to work very very hard to get optimsed code, and you always have more information at runtime than you do at compile time.

      What we are seeing from this is exactly what you would expect, something with a VERY good peak performance that performs well on some benchmarks, but royaly sucks for real world applications..

      Buggy compilers are always a pain, but ignoring that, there's no reason why the compilers can't get better over time to produce better code. I know that doesn't help you right now though :)

      I am sure they will get better... Intel are patching their compiler about every month, and have a huge number of people working on it. Whether the whole EPIC idea will work on the other hand I don't know.... They are going to struggle very hard to get anything like the peak numbers out of it for normal applications..

    6. Re:Altix by bdrago · · Score: 2, Informative
      This is incorrect.

      The Altix at NASA is a true 512 processor single system image - it is not a cluster of smaller nodes.

      Disclaimer: I work for SGI.

  4. All research welcome by fr0m · · Score: 2, Offtopic
    It is fairly obvious that changes in our environment has increased rapidly during the past years.
    Last winter has been one of the coldest in a few hundred years in Sweden. I was there (south part) during christmas and the warmest temperature we had was -24. The same goes for the summer here in Europe. So damn hot. Here in Paris we've had thousands of deaths due to the heat.

    Something strange is happening. All research about our "new" environment is welcome. Ocean or otherwise. What are your thoughts?

  5. Re:Ocean? NASA? by javiercero · · Score: 4, Informative

    Fluid dynamics and environmental studies are also part of NASA's research mission.

  6. NASA was going to roll this out earlier... by AvantLegion · · Score: 5, Funny
    ... but their copy of Mandrake 9.2 broke their supercomputer's LG CD-ROM drive.

  7. Re:Patches ? by BJH · · Score: 5, Informative

    Don't know if this is what SGI is using, but the status of NUMA in the kernel and associated patches for it is shown here.

  8. Re:Ocean? NASA? by Stubtify · · Score: 3, Funny
    Didn't you see the movie "The Abyss"?

    or the tv show "Seaquest DSV"?

    the ocean seems to be a gateway to the stars...

  9. In other news..... by micaiah · · Score: 5, Funny

    SCO sues Nasa for using Linux.

    Darl McBride stated yesturday, "Since Nasa is using Linux we now own the entire universe and are claiming our rightful ownership."

    1. Re:In other news..... by klafhat · · Score: 3, Funny

      SCO sues Nasa for using Linux. Darl McBride ...

      In yet other news, Darl McBride is now so afraid how NASA is going to respond, that he have hired a couple of bodyguards.

      --

      Tell me more, tell me more

  10. they also have. by rf0 · · Score: 4, Funny

    A 512 CPU and 1024 CPU IRIX system. The 512 one is referred to as the small one :)

    Rus

    --
    Cheap UK and US VPS
  11. Not quite "Supercomputing" by quigonn · · Score: 4, Interesting

    The SGI Altix 3000 is not quite a supercomputer. Our local university got the very first model for production use of the Altix 3000's successor, the Altix 3700, in last April or so, and it made it in the TOP 500 supercomputer list in last June, but it fell out of the current list. And the 3700 is even faster than the 3700, so what's so special about it?

    --
    A monkey is doing the real work for me.
    1. Re:Not quite "Supercomputing" by Realistic_Dragon · · Score: 2, Funny

      And the 3700 is even faster than the 3700, so what's so special about it?

      If it's faster than itself, then with some handy infinite recursion you can prove it's infinatly fast!

      Sounds pretty special to me.

      --
      Beep beep.
    2. Re:Not quite "Supercomputing" by martingunnarsson · · Score: 2, Funny

      "And the 3700 is even faster than the 3700, so what's so special about it?"

      It's obviously faster than itself, that sounds pretty special to me!

      --
      Martin
    3. Re:Not quite "Supercomputing" by Error27 · · Score: 5, Informative

      The thing that is special about the NASA computer is that it is a single image system and not really some cluster type thing. Mostly people say that Linux 2.4 scales well up to 8 processors, but this system has 256 processors.

      SGI is working on scaling the kernel to even more processors. For example, Erik Jacobson from SGI recently noticed that 'cat /proc/interrupts' doesn't work if you have 512 CPUs in your system. Frankly when I saw that I thought it was a joke, but I guess it must be real if they already have paying customers.

  12. how does this compare? by kjba · · Score: 3, Insightful
    "NASA plans to study the ocean's future with the help of the world's first supercomputer of its kind to run on the Linux operating system."

    Nice to know that it is the fastest Linux supercomputer, but how does this compare to the other top-ranked supercomputers in the world?

  13. Bad Influence by Anonymous Coward · · Score: 5, Funny

    Isn't using a supercomputer that has hundreds of very hot processors to simulate climatic change going to directly cause a change in the climet be ejecting large quantities of hot air?

  14. Re:Ocean? NASA? by IvyMike · · Score: 4, Informative
    To quote one of the articles:
    "Using the vantage point of space, NASA gains an understanding of our home planet that we could never achieve were we bound to the Earth's surface," notes Dr. Ghassem R. Asrar, associate administrator of NASA's Earth Science Enterprise. NASA's remarkable 45-year history and vast scientific and engineering resources have helped the agency launch numerous research missions to understand and protect planet Earth.
    Sounds pretty reasonable to me.
  15. 3000 Procs @ $699 each. by kidgenius · · Score: 2, Funny
    3000 Processors @ $699 ea ~ $2.1 million
    3000 Processirs @ $1399 ~ $4.2 million

    So that's where SCO was planning on getting its money for Linux. It all makes sense now

  16. Re:Ocean? NASA? by quigonn · · Score: 2, Interesting

    Well, they pay lots of people for doing other stuff than space research. Just think of Larry Wall, who was working on Perl during his time at the NASA JPL.

    --
    A monkey is doing the real work for me.
  17. Nice to see SGI still making sales by Space+cowboy · · Score: 4, Insightful

    ... I have something of a soft-spot for SGI, and it's nice to see them still making high-profile sales - it'll do their government profile no end of good :-)

    512 processors running a single image is pretty cool :-))

    Simon

    --
    Physicists get Hadrons!
  18. Almost but not quite... by Powercntrl · · Score: 4, Informative

    Don't hit the post button if your joke requires a life support system such as:

    "oh, never mind"
    Printed backspace symbols^H^H^H^H^H^H^Hcharacters
    ...pause (Think about it longer, you'll find a new way to make an old joke funny!)
    A comment relating to the moderation system or karma
    Rehashing all your old Slashdot memes are belong to Natalie Portman's hot grits in Soviet Russia goatse.cx posts YOU!
    Using any form of Slashdot cliche as an attempt at humor
    Ending your post with @^T#G@#YHB^#@$NO CARRIER

    --

    ---
    DRM is like antifreeze, to the MPAA/RIAA it's sweet, to the consumers it's poison.
  19. How is this "News" for nerds? by ODD97 · · Score: 3, Informative

    If I'm not mistaken, NASA *invented* the bewulf cluster. And it ran Linux then, too.
    Clicky

    --
    The emperor is naked.
    1. Re:How is this "News" for nerds? by Styx · · Score: 4, Informative

      It isn't a beowulf cluster.

      It uses a single system image for all processors, as opposed to a beowulf, which has separate system images for all cluster nodes.

      --
      /Styx
  20. Heads are going to roll by dcordeiro · · Score: 2, Funny

    Linus is going to ask the responsible for the /proc/interrupts code how on hell didn't he tested that at home:
    " Why didn't you test /proc/interrupts with 512 CPU in ?"

    DUH!!

  21. This new supercomputer has paid off already... by Anonymous Coward · · Score: 5, Funny

    ... as it has completed it's simulation of the ocean in order to predict it's future:

    cold and wet.

  22. Japanese govement ordered 4 x 64 CPU Altix 3700. by zzztkf · · Score: 2, Informative

    According to SGI Japan, Ministory of Education and Science
    has orderd 4 Altix 3700 computer to make up 4 node
    super computer, November 18th.

    Each node, which altix3700 is equipped with 64 cpu. Total
    main memory has reached 1.9TB.

    It's also said that hardwares will be installed and in
    operation in the early half of 2004.

  23. /proc/cpuinfo anyone? by prestwich · · Score: 2, Interesting

    So anyone got a copy of /proc/cpuinfo from this mother?

    (Oh and do you do something special to 'top' so it doesn't give you 512 lines of CPU state?)

  24. Single System Image is Nice by 4of12 · · Score: 3, Insightful

    The thing that is special about the NASA computer is that it is a single image system

    I did parallel code development on Sun SMP boxes. Starting up jobs, seeing what was going on, killing zombies, debugging was all easier on one system than through different boxes you'd have to ssh over to see.

    Even though I was using MPI and getting ready for a distributed memory architecture for the really big runs, the development was easier on the SMP box that showed a single system image.

    I haven't used things like OpenMOSIX, and Don Becker, early pioneer of Linux ethernet drivers (not many other folks can claim a complete decade of experience with Linux networking), founded a company called Scyld that sells Linux clusters with single system image.

    Sometimes it's convenient to see the whole box as if it were one, even though efficient programming dictates that you become aware of the different costs of data access (network, main memory, cache, disk). Practically speaking, developing and running parallel jobs is a higher user productivity proposition on a single system image.

    --
    "Provided by the management for your protection."
  25. Re:Ocean? NASA? by BuilderBob · · Score: 2, Informative

    The research is being done at the AMES labs in Moffett field, California. Home of the climate model for Mars and Solar system modelling and numerical modelling in general. As well as some X-projects (as in X15 and X33).

    NASA doesn't just send things into space anymore.

  26. 256 Procesors, that's not parallel computing!!! by GNUALMAFUERTE · · Score: 2, Funny

    I am at work just at email reading and reloading Slashdot, the radio at high volume, Watching an Iron Maiden '92 live video, while i'm on the phone with yet another client that is mad 'cause we deactivated http upload for php, Now, *That* is _real_ Multitasking!!!!

    --
    WTF am I doing replying to an AC at 5 A.M on a Friday night?