Follow Slashdot stories on Twitter

 


Forgot your password?
Close
typodupeerror
×
Hardware

Fastest Commercial Supercomputer To Be Built 106

Posted by Hemos from the not-a-beowulf-cluster dept.
Zeus305 writes: "Today NuTec Sciences, Inc. will be announcing its purchase of the world's fastest commercial supercomputer, second overall only to ASCI White. NuTec will use and lease time on the 1,250 clustered IBM servers to analyse genes decoded by the human genome project to try to better understand the causes of diseases like cancer by running month-long algorithms that analyse the relationships between different areas of the genome. This beast will have 2.5 terabytes of RAM and 50 TB of disk space."
This discussion has been archived. No new comments can be posted.

Fastest Commercial Supercomputer To Be Built More

Fastest Commercial Supercomputer To Be Built

Comments Filter:

  • 1250 CPU's just the beginning (Score:1)

    by Anonymous Coward
    The article Yahoo! [yahoo.com] states Nutec "will install the first quarter of the 5,000-CPU machine this month and the rest during the course of the next year."

  • Re:Windows 2000, Linux or AIX? (Score:1)

    by Anonymous Coward
    The information on IBM's site says that they are using p640's for the nodes in this thing. These run AIX as shipped from IBM, so I would assume that this is what it will be running. You can get more information on the p640 at www-1.ibm.com/servers/eserver/pseries/640.html
  • Unfortunately it's a problem of supply and demand. The 50,000 Sony Playstation 2 nodes they're using in the cluster are backordered.
  • Anything that gets commercials done faster is OK by me. So this is gonna be, like, a really big Tivo?-)
  • how can it be the fastest computer if it's second to ASCI White ?
  • 2.5 terabytes of RAM and 50 TB of disk space.

    Look at that - 2.5 TB RAM, 50TB disk means that RAM is 5% of disk. Compare that to a reasonable system today, where you have 128MB, and 30GB disk, which is somewhere around 0.4% of disk...

    What's the point of virtual memory when you've got RAM like that? :)

  • There would be so much memory to swap with a (relativly) very small disk bandwidth

    Ehmm. Very small disk bandwidth? I expect every node to have its own physical disk should there be a need to add swap. In that case you can expect every node to have about 40Mb per second worth of swapping potential. The machine would have a total of about 50Gb per second of disk throughput.

    (Aren't you happy that your PC can do 800Mb per second to RAM? this thing does 60 times that to DISK!)

    Roger.
  • 2.5 TB /50 TB = 1/20
    For every one byte of memmory you have 20 btyes of harddrive space. Damn could you imagine the amount of data you can hold in memory to work your magic on. I wonder if that puppy needs any swap space.
  • You're missing my point. I'm saying that for many many applications, a hot-swappable RAID array build out of FibreChannel or what have you gets you nothing.

    If all you need is a place to temporarily stash some DNA input files (which are certainly duplicated somewhere else) than cheap-as-shit IDE
    drives are just fine for what you're.

    And I haven't forgotten about controllers, cabinets, powersupplies, etc. Assuming 1,200 machines, you can put 40 gig IDE drives in each machine and have 48 terabytes of disk. Use a 2U or even a 1U case and you don't need anything else, other than the few buck IDE cable.
  • 1. It won't be the second fastest supercomputer ever built until it performs a benchmark that proves it. Both ASCI Blue machines are theoretically *faster* than ASCI Red, but they still haven't beaten Red in Linpack.

    2. 50 Terabytes of disk isn't really that big of a deal. 80 Gig Drives are $250 off pricewatch, so we're talking roughly 3k per terabyte, or only $150,000 for the disk, which in the high-performance community really isn't all that much money. (And don't whine to me about "Well, super-duper-ultra-scsi7 isn't that cheap" because many many apps simply don't need fast disk - they may very well be CPU bound or maybe memory bandwidth bound.)
  • That's why you'd have to do some degree of carry lookahead for this to make any sense. But even then, you're right, you may not get any gain in speed because of transfer times. However, with other dynamic problems where you're working on very large 3 or 4-D tables (ie k(n)=2l(n-1)+(1/2)m(n-1), k(n)=2m(n-4)-l(n-1), m(n)=3k(n-2)+2m(n-2)) this makes much more sense.
  • Imagine a Beowulf cluster of these...
  • Try reading the article.
  • Think of all the neat new commercials this thing can make!!

    _joshua_
  • This annoys me. Slashdot are really happy to pander to the PR that these sorts of companies have but consistently turn down interesting stories about how we are trying make the human genome open and accessible for all, in projects like Ensembl [ensembl.org]. What are these guys really going do with this? Probably nothing. They don't look like they know what they are doing. And yet they get posted to slashdot.

    I wish Slashdot was more interested in the real science of the genome and less PR orientated. Slashdot aint what it used to be...

  • Check out
    http://www.pcworld.com/news/article.asp?aid=1415 1

    Basically, it's a million processor giant that will hopefully crank out a petaflop. I'm sure they're doing reasearch somewhere, but the only thing I've seen is bad marketing crap. First, "Blue Gene", which is just an obvious, stupid joke. Then the article says it will use a new type of architecure called SMASH...simple, many, and self-healing. How long did them spend coming up with words that fit "SMASH"?

    Anyway, it will probably be rad when the build it. But then again, any supercomputer built four years from now will be pretty amazing by today's standards.

    -B
    1. Something of this size would require FibreChannel disk arrays.
    2. A 360GB FibreChannel array is roughly $100,000 (and that's w/o hot-spare drives)
    3. That's $277,000/TB which equals out to $13.8 million for disk, and I haven't even added any redundency.

    Bottom line is that just becasue you can go to Best Buy and buy a 80GB drive doesn't mean that you can/should use that drive in server.

  • We need somebody to implement the street thing from snow crash on this baby, to hell with cancer, i want VR!
  • www.watchtower.org
  • I saw that article too a year or so ago. I think it was in one of the astrophysics journals (can't find a link, anyone else out there?), though I'd have thought it would have been on slashdot too.

    It really is true (if you can't upgrade the equipment whilst the calculation is underway). Say you've got a problem that would take 10 years to solve on todays computers, and computers double in power every 18 months. If you wait 36 months and then start you could finish the problem in only 5.5 years total! (the optimum time to wait with these figures is a little more, just under 40 months)
  • Fastest commercial supercomputer. ASCI White is owned by the US military, afaik.
  • yes, you can solve for x - i was not trying to say that the fibonacci problem can't be solved better, i was just trying to come up with a trivial example of a problem that must be solved in a linear/sequential fashion. i probably could have done better if i'd thought about it, but this is slashdot... i try to play by the rules... post first, think later.

    (-;

    cool page though! can you come up with a problem that must be solved in an entirely linear fashion? (that can't be memoized or divided into a search space)?

    i'm sitting here trying to remember one from algorithms and i can't, and it's starting to drive me nuts. (:
  • i claim that distributing the adding gains you nothing but delay, since each adding node must wait for another node across the network to complete before it begins its operation.

    or maybe i missed something?

  • It was "447" (not 437) years of computation time, but only on "traditional computers". Of course, now define "traditional computers".

    I mean, to me, a "traditional computer" is the C64, so I am really not talking about a lot of speed...
  • Unless cables cost 4 figures, that's a pretty negligible cost. You can go to radio shack and get random cables for $30 tops.

    -Egon
  • how many you can feed doesn't matter. how many could you teach to fish?
  • 2 years after release there's finally a system that might run Trespasser smoothly.

    Now we can replay all those rigidly linear levels spent stacking rubber boxes in rubber trenches - except now we can do it with a frame rate >20!

  • How about the exchange rate for an Amercian dog?

    Actually, I think the more realistic exchange rate for people/people is 16:1.

    Which means, Americans are contributing more to global population problem than africans.

    Or more bluntly, 1 american baby born, 16 (or 3714) african babies must die to keep the planet in equilibrium. (as baby-specific political spin practiced by so many American politicians.)

    It's ugly outta here.
  • It would be interesting if IBM would donate some time as a form of charity
  • Using Moores Law I get about 2024 for this amount of RAM sitting in front of me, based on how long it takes leading edge to make it to off the shelf. ie (2500GB/128MB) = 1.5 ^n, n is about 24. And for others in the know, what would the complexity of the leading edge be then and how does this compare to human neural complexity?
  • What do you have against Madagascar? I don't think that they have done anything to deserve such a harsh sentence... :-)
  • According to IBM's press release [ibm.com], this thing is made up of 1,250 p640's. Check out the specs [ibm.com] of one here.

    Since the thing has 2.5 TB of RAM, my best guess is that each p640 has 2 Power 3's and 2 GB of RAM (1 GB per CPU). It is possible that each p640 has 4 cpus (and 500 MB RAM per cpu), but that seems like an odd RAM/CPU ratio for a hefty IBM machine.

    dopp

  • I wonder if they will do anything interesting with their unleased time on this beast (assuming, of course, there is any)?

  • You can look ahead as far as you like in a fibonacci sequence by multiplying your current value by whatever power of Phi (approximately 1.618033988749894848204586834365638117720309179805 76, check this page [surrey.ac.uk] for 2000 digits worth) you like, then rounding to the nearest integer that would match the sequence's odd-odd-even pattern. For example in the 1 1 2 3 5 8 13 21 34 55 89 144 233 377 610 987 sequence, to look ahead 5 places from the 987, do 987*(Phi^5). That comes out to 28656.994, not that far off from the correct answer 28657. The answers' accuracy reflects the number you're starting with (larger integers = more accuracy).

    Distributing this across a dozen or so processes wouldn't be that hard.

  • "The next big issue will be how fast will they be churning out cures/treatments? If this helps speed this up, there will certainly be a great number of lives saved."

    Curing of diseases is basically not a function of "speed of investigation" but of politics. We have known how to cure a simple ulcer for ~30 years and yet doctors STILL don't utilize this knowledge. Infact Dr's still prescribe antibiotics for viral based colds which has created huge problems with resistance.

    Western medicine is great for acute conditions but totally rots for chronic conditions. After years of study we still can't cure much. How could this be? Looking in the wrong place or in the wrong way. Why can this go on? Human nature. We wont see what's going on until we want to (or have it rammed down our throat).

    Answers come from asking a wise question with open ears and eyes.
  • Damn it, inquiring trolls want to know.

    How can I post some tasty bait w/o this essential information?

  • Technically the newer versions of AIX (I believe starting with 4.3.3) do no require swap space. As long as you have the physical RAM available. I've seen S80's (one of IBM's bigger AIX boxes) humming along quite happily with no swap space at all.
  • Disclaimer: I'm just looking for insights from those who know more about this subject.

    The msnbc article glances over this a bit, but I'm hazarding a guess that what they're trying to do is statistical analysis based on protein folding relating to the entire active genome. From what I understand, this is a lot like the traveling sales person problem and the whole genre of NP v. P problems that are more suited to quantum computing.

    So as our society's tech progress in the quantum computing realm to the point where we can actually build these things, does that mean we get faster answers to questions like those posed of the genome map?

    If anyone has any informed insights (not my bs guesswork) I'd like to hear them.

    Thanks, -- RLJ

  • If you want protein folding, check out Folding@Home: http://foldingathome.stanford.edu.
  • I hate to feed the trolls, but, it is running AIX 4.3.2 the last I knew
  • It would be interesting if IBM would donate some time as a form of charity

    I'm sure that they would do it if the following conditions were met:

    * There were no paying projects waiting.

    * There were to be some sweet publicity for them.

    * They could claim some huge tax deduction.




    How every version of MICROS~1 Windows(TM) comes to exist.
  • And don't whine to me about "Well, super-duper-ultra-scsi7 isn't that cheap" because many many apps simply don't need fast disk - they may very well be CPU bound or maybe memory bandwidth bound.

    Perhaps not, but they'll probably be using Fibre Channel anyway, if not solid-state storage.
  • I don't know about you guys, but this whole "fastest supercomputer in world" bit is getting kind of old. It seems like every week someone is releasing a new world's fastest supercomputer. Think about it, it's nothing new. String a whole bunch of processors together, put a ton of memory in it and build a huge RAID or SCSI disks and I have the world's fastest supercomputer. In another 5 - 10 years we will have desktop machines that rival this multi-million dollar supercomputer and yes, it will run Linux.
  • Counting ballots?
  • B12. (Niacin)

    Best thing in the world for a hangover. You can buy little packets of pills in gas stations that advertise increased energy.. etc.
    Look on the back of the package to ensure that it is indeed B12 you're getting. Pop'em and wait half an hour.

    I've heard of a lot of things to do to treat hangovers.. this is the only thing I've ever encountered that cures one. (Along with a glass or two of water.)

    -Pip's drunken uncle

    --

  • I imagine they'll have a lot of problems with protestors outside the facility, whinging about how disenfranchised inner-city kids don't get an equitable share of supercomputer time.

    Jesse Jackson will be there, claiming that the very *nature* of the computer is discriminatory: "Ones and Zeros? That's just another way of saying Black and White! We demand that the computer *always* have an equitable number of ones and zeros stored in memory! And if there's a register that has too many ones, by God, we'll send some zeroes *to* that register. On the *bus*!"

    --
  • ...i'm glad to see supercomputers used for fighting disease, rather than predicting weather or simulating nuclear explosions.

    ---
  • think of what all that processing power can do. is this really what we need it to do? raw power like that, available only to those who can construct it. oh well.
  • The only OS that runs on the IBM SP is AIX. > I was curious - it had been asked - what OS are these beasts running?

  • Slaps forehead!

    Of course, DB2... I really don't care too much about what OS is running - Linux would be nice to extend The Cause(tm), but one thing I've always been curious about is the partitioning of the data in particular.

    What really constitutes the data for this job - the entire DNA for some subject? Or is the data collected from several people with and without the disease(s) to determine which genes are likely candidates for leading to the disorder?

    Just curious. If anyone has a link to such answers, it would be muchly appreciated.

    Thanks

  • so that they can restart if there are any problems. That's a hell of a lot easier than
    trying to make some super fault tolerant system.

    Flat5
  • Sure, Red has a good linpack mark. But ask any scientist which machine they would rather run
    their actual codes on. The answer is the blue machines.

    Flat5
  • Sure. That's why I said "for the kinds of problems they're running."

    Flat5
  • That paperclip has raised the minimum requrements a LOT! 8-)
  • Actually I once read that for problems over a certain size you could finish earlier by waiting longer to start, because you could use a faster computer. Not sure if that's true anymore, but I imagine (a beowulf cluster of these) that it is..
    ---
    Posting at -1 means never losing karma.

    • Surfing the Brittney Spears chatrooms on AOL
    • Calculating the precise air speed velocity of an unlaiden swallow (both Afrian and European)
    • Counting votes in Florida
    • OSX (you need that kind of power to push those pretty windows around)

    Are there Voodoo3 drivers for this thing yet?

    Ridiculopathic Zombies From Outer Space [ridiculopathy.com]

  • With Usenet traffic over 200gigs/day this baby oughta have the power to keep up without the meltdowns that periodically plague most news providers. Not to mention some sweet retention in the process.
  • <sarcasm>

    There's no mention of NT? What's wrong with these people? NT is the single best OS out there, powerful and simple to use. Obviously for a top-notch supercomputer, NT is the only obvious choice for the system. I read about it in PC Magazine, just last week, and as everyone knows they are the defacto standard for everything related to computers.

    I just can't believe they didn't mention NT.

    </sarcasm>

  • D'uh, they'll use it for SETI of course, once they've mapped/understood our genome, they'll want to look at the genome of another "advanced" species!

  • It's using IBM's DB2 Universal Database.. which runs on....... various ***x's, Windows, and OS/2. So we still don't know. Damn.
  • Try reading the article.

    It isn't a matter of reading the article, the bloody twit couldn't even understand the title.

  • Now I know what I want for Christams !
    My current UT server is just not fast enough....

    --
  • Heck, I ran a single Oracle instance once that was 2.5 TB on a Sun box. No big deal with the disk space.

  • Take this babe to firingsquad so wa can see how it does QUAKE3!!! YAY!!
  • a beowulf cluster of hot domineering redheads?? sounds like someone's dream come true...
  • well the simulation of nuclear explosions does have some benifit.. from what i remember it was intended to help keep the old stockpiles safe so that we don't all start glowing in the advent that they go off underground .)

    btw- anyone know if this thing will do quake3 better than a p4?

  • It seems like every time I pick up a newspaper or turn on the tv, there's some idiot telling me that genetic researchers have found the gene that "causes" cancer or some other dreaded disease.

    We already know the cause of cancer: carcinogens. We eat, drink, and breathe them, and have since the rise of consumer culture. The same marketroids who have pushed all of this crap on us are the same marketroids who are going to sell us the high-priced cures.

    There is a much better use for a machine like this--use it to model human population negative growth from cancer, etc... Maybe then we can get all of the marketroids and GM researchers/investors into cattle cars and ship them off to Madagascar or somewhere...



    I'd rather be a unix freak than a freaky eunuch
  • Even though I don't agree with the project is is a much better use of an IBM mainframe than a chess game.
  • I'm sure it wont be solitaire. Although a quake deathmatch would be very interesting.
  • One problem that hopefully the engineers have designed the system to prevent would be a program that crashes may destabilize part (or maybe in whole) the network that the computers communicate over. One bad program corrupting something and crashing, lets say, could wreck months worth of work. And it would be months, too.. it mentions that many of the computations can take that long to finish.

    Reminds me of the Hitchhiker's guide.. instead of the Vogons, wouldn't it have been pathetic if the Earth (the most powerful supercomputer in the universe) had an Exception 0E error instead?

  • Re:Ho-hum (Score:2)

    by Anonymous Coward
    Actually, there are major disk I/O problems with supercomputing in many instances not because it needs to move data to and from the disk during compute time but rather because the files are frequently incredibly large and they need to be moved to archival storage/visualization/whatever. ASCI white, for example, is slated to be producing data sets that will take up an average of 5-8TB of space per session and all that data needs to be piped through a network for stoage/viz. They have fat network pipes (lots of fiber for those things), but disks often have a hard time keeping up in those situations. How many hours are we talking about for a single 8TB chunk of data to be pulled from the drive arrays? Spending the extra on faster disks doesn't necessarily speed computation, but it makes a lot of other work more efficient when you're talking about such a massive scale.

    Of course, I'm not sure what type of storage requirements they're going to have in that field or if they're planning on keeping the majority of the archival storage on the main system's 50TB or if that's just a "scratch pad" for huge jobs that will eventually be moved to storage arrays.

    :::shrug:::
  • I don't think they are looking for fast drives either. But... for storage of something this important I would bet that a lot of money did go in storage for some of those fancy scsi drives to go into a RAID system.
    If the data is really important, a redundant system is needed and that ofcourse can double the price. I have looked at good 10TB RAID systems and they are not cheap.
  • Last i heard, wasn't IBM working on a supercomputer to be used for calculating protein folding? If I recall, that was supposed to be the fastest, when complete. Of course, I could be wrong.
  • "Compute the Fibbonaci sequence (without solving it for x) and race your PIII with this computer - and you might win."

    Or, transform the generic Fibbonaci recurrence into a dynamic programming problem an solve it in linear time :) -- this assumes of course unit time for computation so solving fib for small numbers is not very interesting. However, for arbitrarly large numbers (say 1 meg integers) you would be much better off distributing the dynamic fibbonaci program (which operates on an array: for(array[0]=1,array[1]=1,i=2; 1; printf("%i ",array[i++])) array[i]=array[i-1]+array[i-2];) and having one master computer which manages the array and the other computers would act as adders (each computer in a group would act as a sub adder and there would have to be some degree of carry lookahead).

    I guess my point is that many problems that appear to be "0% verctor-optimized" actually are not.
  • The ASCI series are owned by the National Energy
    Labs. The ASCI series are sub-commercial proof-of-concept computers. That is, the mainstream makers are always bragging they can configure a teraflop computer, but no customer can afford them. So Uncle Sam kicks in a few bucks to call them at their word. Everyone wins. The government gets something really fast. The computer companies get an R&D test at government expense. The second customer, a commercial site, gets a more affordable computer.
  • As discussed in the book Cracking DES [oreilly.com] chips specifically made to handle the DES algorithm are much faster than Alpha or X86 chips at cracking (brute force) the DES code. I don't know if this method applies to DNA, but my guess is that it might.

    Using a similar method as outlined in the book, I suspect that chips that are custom built to understand DNA and how it acts (or at least the inherant algorithms that will be used in studying DNA)are possible to build. Ironically, they probably will not be built until we understand genes and DNA better than we currently do. Once machines like this start to be built, how far off will we be from the machines described in Michael Chritons book Jurassic Park? Some scientists have purposed that it will take 100 years to understand the human genome (or other genomes) perhaps, but I think we are closer than that.
  • IBM claims this computer will do 7.5teraflops.
    Compare this to Seti@home's 26.11 TeraFLOPs/sec.

    Why wouldn't NuTec develope the software so every joe blow and there handheld could run a distributed client that does this. I personally have a hard time justifying time spent installing distributed.net or seti@home clients on all the machines I have access to, as I know my boss wouldn't understand the importance of cracking encryption, or searching for aliens on company time. ;)

    However searching for cures to human illnesses, who wouldn't want to do this? With a good piece of software and some proper advertising, theres no doubt they would surpass 30 or even 50 teraflops.

    Though this may not be a possibility if huge amounts of data are required for the calculations. Anyone have some ideas about this?
  • And in fact, ASCI White uses 375 Mhz Power3 SMP processors, somewhat slow in Mhz than what most people have on their desks.

    Here are the specs from a web page I found at LLNL [llnl.gov]:

    • 512 16-CPU nodes (8192 processing units)
    • 375 Mhz Power3 SMP processors
    • Will have 6 TB total of system memory
      • 8 or 16 GB per node
    • Will have 484 batch and 8 debug nodes
    Data from http://www.llnl.gov/computing/tutorials/jw.lcres/t sld014.htm [llnl.gov].
  • "think of what all that processing power can do. is this really what we need it to do? raw power like that, available only to those who can construct it. oh well."

    What do you want, to give the results of this effort to those who don't need it and can't use it? "Here are your food stamps and your array processing ration card."

  • Gene matching is very data intensive -- basically one of the things that they're going to be doing is matching each part against every other part to see where things match. This means having to have a copy of all of the data available to each client. Seti/distributed/etc are really just compute-intensive apps -- they don't need to see a huge data set to do their work.
  • 2.5 Terabytes? Piff. That's not enough ram. Tell those boys to call back when they're ready to play.
    Sig:
  • The next big issue will be how fast will they be churning out cures/treatments?
    It probably won't be anywhere near as fast as they will be churning out patents, unfortunatly.
  • What's the point of virtual memory when you've got RAM like that

    That's the thing: I'm guessing that this machine doesn't use swap at all. There would be so much memory to swap with a (relativly) very small disk bandwidth. For that reason, if you want any performance, you need to put almost everything in memory. The disk is only there when you need permanant storage.
  • wasn't IBM working on a supercomputer to be used for calculating protein folding?

    You're thinking of Blue Gene, which is supposed to hit 1 petaflop.

    Unfortunately it isn't supposed to be operational until 2005.
  • Regardless of whether they need "super-duper-ultra-scsi7" or not, you've still forgotten about controllers, cabinets, power supplies, cables, and stuff. And my guess is, with something like this you are talking hot-swappable RAID arrays, and that stuff isn't quite at spitting-distance-from-free yet.

  • I'm sick of all the money being spent on supercomputers that simply make sure that dollars and cents add up (for instance, VISA's supercomputers). It's heartening to see how motivated people (or companies) are willing spend the doe to solve cancer. My father narrowly escaped it a few years ago, and I wish a cheaper and not so nasty treatment could be developed to prevent it. Others have said, but it needs repeating, that hopefully this will spurn competition in the medical market as pharmucitical companies try to beat each other to better cures.
  • I have a question about these clusters, is it posable to add extra nodes after it is up and running? It would look to me as though you could just keep adding nodes year after year to keep it up to date. Did I miss something ?


    ________

  • You could, but the key to the power of this type of machine is not the number of nodes, its the switch. The switch is the bottleneck, and adding nodes to a machine this size doesn't buy you much for the kinds of problems they're running.

    Flat5
  • Uhh, yeah. This sort of thing is precisely why patents exist. This isn't Linux - biotech venture capitalists aren't stupid enough to sink billions into harebrained schemes to recoup costs by selling hard drive space or airplane tickets.
  • The ASCI series are owned by the National Energy Labs.

    They're actually leased by the DoE, not owned, according to this CNet story [cnet.com].

    ASCI = Accelerated Strategic Computing Initiative

    Their site is here [llnl.gov]. More on ASCI White, including a picture, is here [llnl.gov].

  • They keep this kind of thing up and soon we'll be finishing stuff before we start it, and who knows what sorts of embarassment that will cause.
  • It would take a traditional computer 447 years to solve the first equations that were constructed, according to Michael Mott, a spokesperson for NuTec Sciences, Inc. That's a long time to wait for the health miracles promised by the decoding DNA.

    I'm sure than in only 47 years not 447, that there will be pcs that can put ascii white to shame.
  • At last a decent machine for playing Quake 3 on!

  • Re:But could they.... (Score:3)

    by tolldog ( 1571 ) on Monday December 18, 2000 @08:28AM (#551424) Homepage Journal
    Not always.
    The switch could be a bottleneck. It doesn't have to be. It all depends on how much data has to be transfered.
    I know that with a render farm, which is a NOW (network of workstations) the switch is not the limiting factor. The machine pulls in data, thinks and spits out data, in small chuncks (a couple K at a time). Any switch should be able to handle this.

    Also, if more of an interaction is needed between machines, they can be networked in hypercube configs by adding a few more switches.

    The backbone can get in the way, but if that is what the limiting factor is, then maybe smaller groups should be used, each working on different segments. Or, more stuff should be stored locally with some sort of smart push script.

  • Re:Parallel computing & computer science... (Score:3)

    by acacia ( 101223 ) on Monday December 18, 2000 @09:34AM (#551425)
    The advances in parallel computing today are driven by three factors, as I see it. Marketing, homogeneous (relatively) environments, and Mathematics & Methodology.

    1.) Marketing - Computer manufacturers and systems integrators/consultancies purport to be able to solve bigger/more ambitious problems. Moreover, it makes good business sense to be able to do so. Business got a hit from the crack pipe of information and they got hooked. These problems now fall outside of the realm of national security. Furthermore, government work can be precarious for many companies, and by diversifying their wares and selling to public corporations, vendors spread the risk around.

    2.) Homogeneous (relatively) dev/prod environments - Not too many people can claim knowing how to program for a Cray or Thinking Machines box, but a lot of intelligent people can move around in/administrate a UNIX environment, and some of them can code to a messaging interface. For that matter, some know tools like Ab Initio or Orchestrate and can create parallel applications very easily.

    3.) Mathematics and methodology have changed - People now recognize the conceptual and practical challenges of parallel computing, and can tailor the algorithms, hardware, and OS to accommodate the challenges of that paradigm.

    Seeing random poster's on Slashdot recognize that compartmentalized data and code is necessary for distributed computing to be effective is a tribute to how far this field has come. The engineering has come a long way, as has the marketing, and overall level of conciousness.

    As for the new adjective, I would say that wider is o.k., but you have to recognize that a machine does not have to be uniformly wide. I think of parallel programming as a stream metaphore, with speed (CPU), width, (#of CPU's or units of work/data ways parallel) and depth (depth of queue/instructions between checkpoints). How about liters? :-p

  • Re:I'm glad to see... (Score:3)

    by Verteiron ( 224042 ) on Monday December 18, 2000 @07:48AM (#551426) Homepage
    Well, here's a link [ibm.com] to IBM's story on the thing. It delves a LITTLE more into the technical side of it but not much more than the CNN article.

  • I was surprised to read the responses and not see any discussion of these increasingly "super" computers' ability to crack strong encryption.

    Believe me, I'm not an Area-51-head, but in the few short years after strong ecryption has been widely available to concerned citizens and terrorists alike there seems to be many more huge supercomputers getting built, each with a greater altruistic purpose attached. "It will allow us to test nuclear weapons without building them! It will cure cancer!"

    It's wholly reasonable to assume that there are military initiatives to ensure that we can't be snuck up on by PGP-wielding bad guys. As someone not wanting to be blown up, I hope there are, anyway. It's also wholly reasonable to assume that the military couldn't amass the kind of hardware necessary to do this without lighting up some analyst's bat-computer.

    But does anyone feel that the initiative could survive being entirely in the light of day? What would the /. response be to an announcement that says "Military announces super-computer initiative break strong encryption in real time, promises to leave private citizens alone!"

    Of course I'm not saying that every computer faster than a Pentium 4 is part of an arms program, there are serious economic incentives to making progress in cancer treatment. I just think that we would expect to see the military arming itself with the weapons du jour, and my guess is that a few are probably sitting in plain view.

  • I'm glad to see... (Score:4)

    by glebite ( 206150 ) on Monday December 18, 2000 @07:24AM (#551428)

    That very powerfull distributed systems are starting to become more mainstream. It's about flipping time that companies made use of computing resources beyond their previously wildest dreams.

    Estimates of 437 years compressed to 1 month timeframes are awesome! The next big issue will be how fast will they be churning out cures/treatments? If this helps speed this up, there will certainly be a great number of lives saved.

    Hopefully though more companies will jumpt to the forefront, and try to outdo each other ( you know they will) and come up with more radical applications and solutions.

    I was curious - it had been asked - what OS are these beasts running?

  • Parallel computing & computer science... (Score:5)

    by mwalker ( 66677 ) on Monday December 18, 2000 @08:54AM (#551429) Homepage
    Isn't this trend towards faster supercomputers being driven by advances in Computer Science, rather than Engineering?

    Remember the Cray Y-MP? Used to be the world's fastest computers were designed to be extremely fast CPU's, built as a sphere to shorten contact length and liquid-cooled. Parallel computing was possible then - the problem was that we couldn't break down the problems we wanted to solve into parallel events.

    Today's brand of parallel supercomputers exist to solve a different kind of problem - a problem in which the "search space" can be compartmentalized and distributed- like the RSA challenge, fluid dynamics, chess, and -of course- the human genome.

    The thing to remember when we read about ever-faster parallel computers is that, for all intents and purposes, when you have to solve a truly sequential problem (what the cray folks would have called a 0% vector-optimized problem) - today's supercomputers usually aren't any faster than the desktop computer you're sitting in front of. Compute the Fibbonaci sequence (without solving it for x) and race your PIII with this computer - and you might win.

    Just something I wish they'd point out. We need a better adjective than faster for parallel computers. They're something else. Maybe... wider.

    Suggestions?

Slashdot Top Deals

"Who alone has reason to *lie himself out* of actuality? He who *suffers* from it." -- Friedrich Nietzsche

Close