World's Fastest Supercomputer To Be Built At ORNL

good stuff by Anonymous Coward · 2004-05-12 01:31 · Score: 4, Interesting

Personally I'm happy to see Cray still making impressive machines. Not every problem can be solved by "divide and conquer" clusters.

Re:good stuff by Anonymous Coward · 2004-05-12 01:41 · Score: 0, Troll

Not every problem can be solved by "divide and conquer" clusters.

Please elaborate, since you seem to know this stuff.
Re:good stuff by Anonymous Coward · 2004-05-12 01:42 · Score: 0, Interesting

Personally I'm happy to see Cray still making impressive machines. Not every problem can be solved by "divide and conquer" clusters.

in reality there is no difference between a multi processor system that uses a motherboard bus or a system that uses ethernet as as a "bus" between processors. It is an artificial distinction when you are talking about things this big.
Re:good stuff by sotonboy · 2004-05-12 01:46 · Score: 4, Insightful

I disagree. There is a huge difference. Bolting a load of boxes together with ethernet and all the associated overheads can never be as efficient as dedicated hardware for connecting, and sharing the processing load.

Obviously there is a lot more that could affect the performance, such as how memory is implemented. In general though, the system will perform best when each processor is performing calculations, rather than looking after ehernet connections.
Re:good stuff by Anonymous Coward · 2004-05-12 01:48 · Score: 0

The question is: does it really matter with a system that large?
Re:good stuff by adam872 · 2004-05-12 01:51 · Score: 4, Informative

Some problems are easily partitioned up and distributed to separate nodes. In particular, code where the nodes do not need to talk to each other much are ripe for clusters, as the interconnect speed is less important. Therefore, you can build a commodity cluster fairly cheaply.

For other problems, where interprocess/node communication is high or very high, you need a high speed interconnect (like NUMAflex in SGI's) to get you the scalability you need, as you increase the number of processors/nodes and the size of the data set increases. The big systems like Crays and the bigger SGI's and IBM Power series have those high speed interconnects and will allow you to scale more efficiently than the clusters. They cost a lot more though :)

A good book to read on the subject of HPC is High Performance Computing by Severance and Dowd (O'Reilly). It's a little old now, but it covers a lot of the concepts you need to know about building a truly HPC system (architecture as well as code).
Re:good stuff by Anonymous Coward · 2004-05-12 01:54 · Score: 0

I disagree. There is a huge difference. Bolting a load of boxes together with ethernet and all the associated overheads can never be as efficient as dedicated hardware for connecting, and sharing the processing load.

Obviously there is a lot more that could affect the performance, such as how memory is implemented. In general though, the system will perform best when each processor is performing calculations, rather than looking after ehernet connections.

"Obviously there is a lot more that could affect the performance"

Try saying "cost", look at the cost of virginia tech's and look at the price of this one. And then tell me you couldn't get a better deal direct from IBM on PPC...

At $50 million you have almost TEN times Virginia Tech... This simply does not compete.
Re:good stuff by Jeremy+Erwin · 2004-05-12 02:47 · Score: 4, Interesting

But Virginia Tech's cluster doesn't use Ethernet as its primary network. It uses Infiniband. As for the cost not scaling linearly, ask yourself whether Big Mac's performance scales linearly.
Re:good stuff by ElGanzoLoco · 2004-05-12 02:58 · Score: 1

Surely you meant "divide and cluster" right? :-D

--
Hello! I'm a disaster waiting to happen!
Re:good stuff by Shinobi · 2004-05-12 03:48 · Score: 1

Wrong. It's a very real distinction. The more nodes you have, the more important it becomes to pay attention to it.
Re:good stuff by Shinobi · 2004-05-12 03:50 · Score: 2, Insightful

The larger the system is, the more it matters.
Re:good stuff by flaming-opus · 2004-05-12 06:58 · Score: 1

Well, that may not be true. The $5 price tag for the VT cluster is only for the hardware. A lot of their people costs were not exactly factored into a lot of the press coverage. Furthermore, ORNL can't pay grad students a $15/year stipend to administer the machine. No matter what they buy they need admins who are capable, well paid, and have security clearance. On big cluster systems, the lifetime labor costs often rival the initial hardware cost in total dollars.

Last I heard, the VT cluster had done little or no real computation, just benchmarks and tinkering. While other sites have put clusters to work on real problems, they tend to have a lot of system down-time for maintenence. Compare that to the cray T3Es (predecessor to the X1) which average better than 90% utilization.

Clusters are cheap, but you get what you pay for.

Wow... by nother_nix_hacker · 2004-05-12 01:32 · Score: 3, Funny

The project involves private companies like Cray, IBM, and SGI, and when complete it will be capable of sustaining 50 trillion calculations per second."

Outlook with no slowdown!

Re:Wow... by FenwayFrank · 2004-05-12 01:38 · Score: 5, Funny

It's so fast, the blue screen shifts to red!
Re:Wow... by 1024x768 · 2004-05-12 01:44 · Score: 1

this has yet to be proven.
Re:Wow... by David+Horn · 2004-05-12 02:20 · Score: 1

"Not only I can tell you your weight, I can compute your personality problems to eight decimal places..."

Guess that applies here more than anywhere else. ;)

*ducks*

--
PocketGamer.org - For the gamer on the go!
Re:Wow... by opus18 · 2004-05-12 03:23 · Score: 1

Or Violet.... in which case, you'd better duck.
Re:Wow... by Anonymous Coward · 2004-05-12 03:26 · Score: 1, Funny

Actually it went to plaid
Re:Wow... by MarvinIsANerd · 2004-05-12 07:38 · Score: 2, Insightful

I can't believe this got modded up to +5 Funny. Any true nerd on Slashdot knows that blue is at a higher frequency than red. So if something blue moves faster (increases in frequency) it is going to shift into ultraviolet and beyond.
Re:Wow... by zmooc · 2004-05-12 08:56 · Score: 1

Ever considered the possibility that it might be traveling towards you and therefore shift the other way?:P

--
0x or or snor perron?!
Re:Wow... by Anonymous Coward · 2004-05-12 09:10 · Score: 0

Something moving quickly away from you, like a galaxy, shifts from blue to red. Something moving quickly towards you shifts from red to blue.
Re:Wow... by Anonymous Coward · 2004-05-12 10:15 · Score: 0

Man, I wonder what the Quake III FPS would be on this baby!
Re:Wow... by forkazoo · 2004-05-12 14:17 · Score: 1

I remember the first time I saw Outlook. I moved away from it so fast that I had to pick up the logo on a ELF radio reciever.

How's that for an anti-MS joke?
Re:Wow... by FrYGuY101 · 2004-05-12 15:51 · Score: 1

Ahem. Moving towards you would result in blueshift. As it is already blue, it would instead result in ultravioletshift, et cetera, et al, yada yada, and so on.

However, if it was moving AWAY at a high rate of speed, it would indeed result in red shift. And since we ALL want to move away from Outlook as fast as we can, this is indeed a grand achievement!

--
"If we let things terrify us, life will not be worth living."

- Seneca
Re:Wow... by FenwayFrank · 2004-05-13 04:49 · Score: 1

Exactly! I have to admit that the OS going away was my first thought, but I like this spin better.

Qualifier by andy666 · 2004-05-12 01:32 · Score: 5, Insightful

As usual, there should be a qualifier as to what is meant by fastest. According to their definition they are, but not according to NEC's, for example.

Re:Qualifier by danormsby · 2004-05-12 03:13 · Score: 1

But the definitions are there.
From the Yahoo article "50 trillion calculations per second" and if the machine is to feature on the top500.org then from top500.org "the best performance on the Linpack benchmark is used as performance measure for ranking the computer systems."

--
Omnis amans amens
Re:Qualifier by pmfp · 2004-05-12 03:25 · Score: 1

and what's that?

--

"So unmerciful is life, that everything afterwards is too late."
Re:Qualifier by fgodfrey · 2004-05-12 03:50 · Score: 1

Yeah, unfortunately, Linpack is not the best bechmark of system performance out there. For instance, it doesn't really stress the memory interconnect between processors. This is why you can bolt together a cluster of machines and get good Linpack numbers.

However, to answer the original poster, our (Cray's) definition of "fast" is probably pretty close to NEC's definition. The goal here is to build a machine that's not just the fastest on benchmarks, but really is the fastest in the world on most real problems. Remember that one of the reasons the NEC machine is fast on real problems is that it is a vector architecture with good memory bandwidth between processors. Seymore Cray invented vector computing and our machines have a *lot* of memory bandwidth :)

--
Go Badgers! -- #include "std/disclaimer.h"
Re:Qualifier by fgodfrey · 2004-05-12 04:19 · Score: 1

Gah! I hate replying to my own posts, but I misspelled "Seymour Cray"!

--
Go Badgers! -- #include "std/disclaimer.h"

50 trillion by Killjoy_NL · 2004-05-12 01:32 · Score: 2, Interesting

50 trillion calculations per second.
Wow, that's darn fast.

I wonder if that processing power could be used for rendering like was done by Weta and how the performance could compare to their renderfarm.

--
This is the sig that says NI (again)

Re:50 trillion by Roger+Keith+Barrett · 2004-05-12 01:34 · Score: 2, Funny

Build me a real time simulation of Morgan Webb PLEASE!

--

Why don't you embrace your slashbotness instead of living in a dreamworld?
Re:50 trillion by WindBourne · 2004-05-12 01:37 · Score: 3, Insightful

I wonder if that processing power could be used for rendering like was done by Weta and how the performance could compare to their renderfarm.
Sure, but the real question is why would you? The cost of this on a per mip basis is sure to be much higher than a renderfarm. In addition, ray tracing lends itself to parellelism. There are many other problems out there that do not that can use this kind of box.

--
I prefer the "u" in honour as it seems to be missing these days.
Re:50 trillion by Anonymous Coward · 2004-05-12 01:47 · Score: 0

Build me a real time simulation of Morgan Webb PLEASE!

I'm sure for the $50M the gov't is throwing at this project, they could probably hire her to give everyone on /. an evening or two of lovin'...
Re:50 trillion by Killjoy_NL · 2004-05-12 01:47 · Score: 1

I know, I was just curious :)
and IMHO being curious is good.

--
This is the sig that says NI (again)

Talking out my ass here, but by SatanicPuppy · 2004-05-12 01:34 · Score: 1, Interesting

I thought the age of the over-priced supercomputer was over, and the age of the cluster had begun?

Sure, I'd love to have one of those things in my house, but as long as the government is spending my money, I think I'd rather see them go for a more cost effective solution, rather than another 1 ton monster that'll be obsolete in two years.

--
ad logicam Claiming a proposition is false because it was presented as the conclusion of a fallacious argument.

Re:Talking out my ass here, but by Waffle+Iron · 2004-05-12 01:48 · Score: 4, Insightful

There are still a few computing problems that can't be efficiently split into a large number of subproblems that can be executed in parallel. For those cases, a cluster of small machines won't help.
Re:Talking out my ass here, but by InsaneGeek · 2004-05-12 01:50 · Score: 1

Insert standard response about how some problems do not parallelize well and that a bunch of nodes in a cluster (no matter how many) wouldn't be able to crunch out the problem in any resonable time.
Re:Talking out my ass here, but by Shisha · 2004-05-12 01:55 · Score: 0, Redundant

There are some problems that lend themselves to be easily divided between thousands of processing nodes and then combined in a final answer / solution. (easily meant in the way mathematicians use the word, i.e easy means it known to be possible, anyone who's ever written a program using more that 2 threads knows what a nightmare it can be)

Then there are problems where such approach is not possible and you just need a very fast pipeline, or a big data troughput or whatever.
...as long as the government is spending my moneyI guess they considered other options before awarding the contract.

Sure, I'd love to have one of those things in my house,... I certainly won't my P4 system is making enough noise already and it's more than fast enough for all the simulation I've run on it in the last three years.
...another 1 ton monster that'll be obsolete in two years... Shame really, I must agree. But obsolete sounds a tad too harsh, perhaps not amongst the top 10 most powerfull but still usefull. And in 10 years time, they can incorporate it into the firewall and use it to run spam detection programs. (if the spam traffic keeps increasing at the current pace, it might be sufficient to service say, 5 mailboxes?)
Re:Talking out my ass here, but by flaming-opus · 2004-05-12 02:01 · Score: 4, Interesting

If you care to, read the pdf on their early impressions of the X1. The Army High Performance Computing Research Center (www.arc.umn.edu) did an analysis of their application and found that the X1 was actually MORE cost effective than a commodity cluster.

Firstly, the X1 was greater per-processor performance by a factor of 4. Then you add an interconnect that has half the latency, and 50 times the bandwidth of myrinet or infiniband. It also has memory and cache bandwidth enough to actually fill the pipelines, unlike a Xeon which can do a ton of math on whatever will fit in the registers. Some problems just don't work real well on clustered PCs, they need this kind of big iron.

Secondly, some problems cannot tollerate a failure in a compute node. IF you cluster together 10,000 PCs, the average failure rate means that one of those nodes will fail about every 4 hours. If your problem takes three days to complete, the cluster is worthless to you. A renderfarm can tolerate this sort of failure rate, just send those frames to another node. Some problems can't handle it.

Oak ridge is very concerned with getting the most bang for the buck.
Re:Talking out my ass here, but by stratjakt · 2004-05-12 02:08 · Score: 2, Informative

Didn't Cray make some comparison about supercomputers vs clusters being like a tractor trailer vs a fleet of honda civics?

The civics might be fine for couriers, but if you need to move - say - an elephant they're useless.

Analogies suck, though, and I'm pretty sure I got that one wrong.

--
I don't need no instructions to know how to rock!!!!
Re:Talking out my ass here, but by Uhlek · 2004-05-12 02:08 · Score: 2, Informative

Clusters are not the be-all end-all of supercomputers. Clusters are really only effective if you have a problem that can be paralellized -- or split into multiple parts that can each be worked independently of one another and then merged into a single result. Factorization, rendering, etc. are all examples of easily paralellized operations.

Certain operations, though, are highly dependant upon each previous result. Physics and chemical simulations are a good example. When you have situations like this, clusters don't do you a lot of good, since only one iteration can be worked on at a time -- leaving most of your cluster sitting there idle.
Re:Talking out my ass here, but by hackstraw · 2004-05-12 02:20 · Score: 1

While we are all talking out our ass here. 1st, the grandparent poster says:

I thought the age of the over-priced supercomputer was over, and the age of the cluster had begun?

Sure, I'd love to have one of those things in my house, but as long as the government is spending my money, I think I'd rather see them go for a more cost effective solution, rather than another 1 ton monster that'll be obsolete in two years.

If you think that $50 mil is overpriced for the fastest computer in the world, then the guys who will soon be in 2nd place that paid $400 mil must feel really stupid.

Then the parent poster says:

There are still a few computing problems that can't be efficiently split into a large number of subproblems that can be executed in parallel. For those cases, a cluster of small machines won't help.

(Score:-10, Wrong)

I'm sorry dude, but this macine is going to have more than 1 CPU in it, and the work will have to be split among the processors and ran in parallel.
Re:Talking out my ass here, but by paitre · 2004-05-12 02:35 · Score: 2, Interesting

Certain operations, though, are highly dependant upon each previous result. Physics and chemical simulations are a good example. When you have situations like this, clusters don't do you a lot of good, since only one iteration can be worked on at a time -- leaving most of your cluster sitting there idle.

Umm, bwah?
It's only going to be sitting there idle if you're not properly scheduling and qeueing jobs. Also, you -CAN- do the kind of simulations (Physics, chemicals) on a cluster *points at clusters at Chrylser and Shell*. The caveat is that you need to write out the result for the appropriate job to handle (in practice - job run 1 contains step 1, job run 2 step 2, etc). And a cluster is perfectly fine for this.

That all said - a supercomputer like this -IS- generally a better tool for the job if you've got the money. Money, in most places, -IS- an object, so we get the best bang for our buck.
*shrug*
Re:Talking out my ass here, but by NineNine · 2004-05-12 02:51 · Score: 1

Not *everything* can be done with a bunch of Linux boxes from local garage sales, as much as the /.'ies want it to be true. You seem to have a serious case of the "one tool fits all" syndrome.
Re:Talking out my ass here, but by Waffle+Iron · 2004-05-12 02:54 · Score: 4, Informative

I'm sorry dude, but this macine is going to have more than 1 CPU in it, and the work will have to be split among the processors and ran in parallel.
The number of processors isn't as important as the memory architecture. Clusters of workstation-class machines have isolated memory spaces connected by I/O channels. Many non-clustered supercomputers have a single unified memory space where all processors have equal access to all of the memory in the system. This can be important for algorithms that heavily use intermediate results from all parts of the problem space.
Even so, for a given number of FLOPS, a vector machine would generally require fewer CPUs than a cluster of general-purpose machines. This reduces the amount of splitting that has to be done to the problem in the first place.
Re:Talking out my ass here, but by bigjocker · 2004-05-12 03:00 · Score: 2, Informative

There are still a few computing problems that can't be efficiently split into a large number of subproblems that can be executed in parallel. For those cases, a cluster of small machines won't help.

(Score:-10, Wrong)

I'm sorry dude, but this macine is going to have more than 1 CPU in it, and the work will have to be split among the processors and ran in parallel.

(Score:-100, Wronger)

Sorry, but you have it all wrong. The parent is right. The parent stated that there are problems that can't be split in smallest problems for being handled by a cluster of computers. A cluster is a set of computers that work independant of each other and have the ability ro comunicate at ethernet speeds (10 - 100 - 1000 Mbits / Sec). There are problems that cant be solved using this approach, for example calculations where all processes must reuse the same data; with really big data sets the network connections become bottle-neck.

For those kinds of problems (the usual example is a simulation of a nuclear explosion, a star system, etc) you need a single machine with loads of processors sharing the same memory space. That's where supercomputers come to play.

--
Life isn't like a box of chocolates. It's more like a jar of jalapenos. What you do today, might burn your ass tomorrow.
Re:Talking out my ass here, but by hackstraw · 2004-05-12 03:32 · Score: 1

From Cray's website:

The system features powerful vector processors combined with an interconnect that scales to peak performances of multiple tens of teraflops.

The Cray X1 programming environments include a powerful and complete set of compilers, libraries, debugger and performance analysis tools that have been designed to exploit its architecture.

The Cray X1 system provides support for a variety of parallel programming models, from traditional distributed memory parallel models, to shared memory parallel models and the latest global distributed memory parallel models.

Trust me. Running an app on a cluster or a big smp box is just about the same. There are interconnects, multiple cpus, etc. A box like the x1 is more tightly integrated and there is only one instance of an OS running across all processors and the OS has hooks so that you can access memory across processors, lock memory across processors, do syncronization across procs, etc.

You said:

A cluster is a set of computers that work independant of each other and have the ability ro comunicate at ethernet speeds (10 - 100 - 1000 Mbits / Sec).

That sounds like an office lan to me, not a cluster. Clusters havn't used 10mbit ethernet in a long, long time. Many utilize interconnect technology like infiniband, myrinet, or dolphin which can go up to 800 MByte/sec.

BTW, I do "supercomputing" for a living.
Re:Talking out my ass here, but by mfago · 2004-05-12 03:49 · Score: 1

Trust me. Running an app on a cluster or a big smp box is just about the same.

No. MPI vs. pthreads can hardly called "about the same."

That sounds like an office lan to me, not a cluster. Clusters havn't used 10mbit ethernet in a long, long time. Many utilize interconnect technology like infiniband, myrinet, or dolphin which can go up to 800 MByte/sec.

I wish you were correct. I know of a new Alpha cluster that uses 100MB ethernet. Specialized interconnects are certainly better, but those in charge do not always wish to pay for them.

Perhaps Big Mac will change this...
Re:Talking out my ass here, but by fgodfrey · 2004-05-12 04:03 · Score: 1

1) I suspect that you will find that once you pay for Infiniband, Myrinet, etc... you will have paid quite a lot of money for your "cheap" cluser". In fact, you may have even gotten into the price range of a Cray.

2) It's not just the bandwidth, it's the latency between nodes that is helpful. If you have a large application that is sitting on a barrier waiting to proceed, you don't need much bandwidth to tell everyone to go, but you sure as heck want to be able to tell them quickly! Not having to go through the Northbridge and over a PCI bus then through the network then back through the PCI bus and the Northbridge on the far side is a big win.

3) 800MB/s is "nothing" for a Cray. Each node (16 processors) in a Cray X1 has links that do 1.6GB/sec. And we have 16 of them per node...

4) You say "Running an app on a cluster or a big smp box is just about the same". This would be disputed by anyone who has had to optimize an application for both. The OS on the X1 doesn't just have "hooks" to access/lock/sync memory across processors, the memory is simply mapped into your address space and you access it. That means no syscalls, another thing that can slow you down. You simply manipulate the memory whether it's local or remote.

5) A Cray (and an SGI) is not a "big SMP box". It's a NUMA system.

6) I also do supercomputing for a living.

--
Go Badgers! -- #include "std/disclaimer.h"
Re:Talking out my ass here, but by fgodfrey · 2004-05-12 04:08 · Score: 1

Seymour Cray on Clusters: "If you had to plow a field, which would you rather have, one strong oxen or 1024 chickens?"

--
Go Badgers! -- #include "std/disclaimer.h"

Hmm by LaserLyte · 2004-05-12 01:34 · Score: 5, Funny

> ...capable of sustaining 50 trillion calculations per second.

Hmm...I wonder if I could borrow it for a few days to give my dnet stats a boost :D

Re:Hmm by Anonymous Coward · 2004-05-12 01:43 · Score: 0

Answer: NO

Next stupid question please?
Re:Hmm by scotch · 2004-05-12 01:53 · Score: 2, Funny

Will you be here all week?

--
XML causes global warming.
Re:Hmm by Anonymous Coward · 2004-05-12 03:17 · Score: 0

A friend of mine got a little "burn-in" time on a 2048 processor cluster for a few days.

He got like 90,000 seti work units done. :)
Re:Hmm by Anonymous Coward · 2004-05-12 04:43 · Score: 0

Answer: NO

Next stupid question please?

Shamelessly plagerized by Anonymous Coward · 2004-05-12 01:35 · Score: 3, Funny

Wow, 50 trillion calculations per second. Thats almost fast enough to finish an infinite loop in under ten hours.

Yeah... by Stile+65 · 2004-05-12 01:36 · Score: 1

And then VT will add more nodes to their G5 cluster. :P

--
I claim first use of "Error No. 0B" - or "No. 0B error." It'll be the new ID 10T!

Re:Yeah... by word+munger · 2004-05-12 01:48 · Score: 2, Interesting

Unfortunately we haven't heard much from them lately (Notice the "last updated" date). I suspect they're still waiting on their G5 xServes.

--
ScienceSeeker.org
Re:Yeah... by paitre · 2004-05-12 01:56 · Score: 1

Which is forcing me to continue waiting for the one I ordered the day the fucking things were announced.
They've gone from giving me a mid to late April ship date to "Sometime in June".

Screw that. Apple is screwing the pooch if they're at all serious about getting into enterprise computing. It's one thing to slip one or two months, but now they're at four, and I wouldn't be suprised to see it go to 6 at this point.

Fartknockers.
Re:Yeah... by birukun · 2004-05-12 02:06 · Score: 2

Remember, DOE is a tax-payer funded agency. For my money, the G5 solutions looks better!

--
Self Defense - A Human Right www.a-human-right.com
Re:Yeah... by raalynthslair · 2004-05-12 02:21 · Score: 0

well if IBM could get them chips faster they could get the machines out faster, VT would get theirs X Servers, and you get your machine...

--
-- "You must be the change you desire to see in the world." Mahatma Gandhi --
Re:Yeah... by Anonymous Coward · 2004-05-12 03:38 · Score: 1

Don't blame Apple, blame IBM. The supply shortage is due to a lack of 970fx processors.

How many Apples would it take? by Anonymous Coward · 2004-05-12 01:36 · Score: 0

How many Apple servers would it take to reach this capacity and what would the cost be? VPI was able to build theirs for a LOT less than $50mil.

Re:How many Apples would it take? by Jeremy+Erwin · 2004-05-12 03:05 · Score: 2, Informative

Big Mac was tested in a small 128 node configuration as a prelude to the full 1100 nodes.

The 128 node cluster was benchmarked at ~80% efficiency, or ~1.6 Teraflops. The final cluster achieved a RMax of 10.28 TFlops, ~60% of the 17.6 TFLOP theoretical peak.

A 6000 node cluster would be very difficult to manage.
Re:How many Apples would it take? by Anonymous Coward · 2004-05-12 03:34 · Score: 0

Big Mac was tested in a small 128 node configuration as a prelude to the full 1100 nodes.
So the 128 node was tested at a McDonald's restaurant, and the the 1100 node system is used here: McDonald's right?
Re:How many Apples would it take? by TennesseeVic · 2004-05-12 05:48 · Score: 1

I'm excited that it's going to be a Cray, as they have the best memory-to-processor architecture. Scientific problems can be subdivided, but there will always be heavy communication between the processors. In particular global communication (a global sum for instance) is a killer. The more processors, the slower this operation will run, so for a big enough machine, this can actually dominate the cost. You get to a point where adding more processors does not make you any faster! The Cray has a beautiful architecture, where one processor can put data straight _into_register_ on another processor. No cache and network delays. This is freakin' awesome. In Cray's presentation about the X1 they have an ocean simulation code that keeps scaling way beyond IBM, HP, &c machines, precisely because of the efficiency in global operations. My disappointment is that they are only aiming for 50Tflop. The Earth Simulator hit 37 two years ago! This is no progress. Victor.
Re:How many Apples would it take? by Jeremy+Erwin · 2004-05-12 07:48 · Score: 1

No. Big Mac was one of the early nicknames for the virginia tech cluster.
Re:How many Apples would it take? by renderdude · 2004-05-12 08:05 · Score: 1

If you look at the Cray press release, they state it will be 20 Tflops in '04, 100 in '06, and 250 in '07. So the answer is a whole lotta apples.

It must be said: by Recoil_42 · 2004-05-12 01:38 · Score: 0, Funny

Imagine a beowulf cluster of those.

--

Newsie, Moderator, www.tauniverse.com

Re:It must be said: by stephenisu · 2004-05-12 02:11 · Score: 2, Funny

Wow, that's the first Beowulf cluster comment I would mod as interesting.

--
Sigs? We don't need no stinking sigs!

Doom III by MrRuslan · 2004-05-12 01:39 · Score: 4, Funny

at an Impresive 67fps on this baby...

Re:Doom III by jabex · 2004-05-12 02:10 · Score: 1

Unfortunately it will be the only computer capable of playing Doom III (outside of the Japanese Earth Simulator... which Doom III was originally designed for).

--
Like Teddy with an elephant gun.
Re:Doom III by The_Mr_Flibble · 2004-05-12 02:51 · Score: 1

But is it good enough to run the next version of windows ?

World's Fastest Supercomputer! Wow. by cableshaft · 2004-05-12 01:41 · Score: 0, Troll

Considering the almost exponential rate that technology increases, this computer won't be any faster than a personal computer in 6-8 years, and will undoubtably have its rank supplanted within probably 4 years. I fail to see why this is all that interesting, especially since I know its status will be ousted sometime in the near future, barring some sort of crazy worldwide nuclear or biochemical war.

It's not quite as "prestigious" as the guy with the world's longest fingernails or anything, that's all I'm trying to say.

--
Creator of the popular web game Proximity

Re:World's Fastest Supercomputer! Wow. by Anonymous Coward · 2004-05-12 01:55 · Score: 0

Yes. And no one should buy a new console for the first four years it is out, because it will just be supplanted by new consoles anyway. As a matter of fact, no one should make any technology purchases, because it's not like you get to use it or anything. It's all a waste in the long run.

Of course, your post isn't as "prestigious" as the guy with the world's biggest asshole, that's all I'm trying to say.
Re:World's Fastest Supercomputer! Wow. by TheGavster · 2004-05-12 02:01 · Score: 1

Sure, personal computers will be faster in a few years, but sheer speed isn't the only thing that makes a supercomputer. What divides massive clusters of $200 Wal-Mart boxes from a mid 80's Cray (those Crays, by the way, still go for ~$20K) is that the Cray uses much faster, wider pipelines between components. With the current trend of lagging the various buses behind the processor almost by orders of magnitude, the desktop PC won't approach current-day supercomputers for a long time.

--
"Because Science" is one step from "Because old book". Try "Because of my experiment testing my falsifiable assertion".

I GUESS that's fast but... by Anonymous Coward · 2004-05-12 01:42 · Score: 0, Redundant

How many FPS does it get in Q3? :P

They better hurry ... by realSpiderman · 2004-05-12 01:42 · Score: 5, Interesting

... or this is going to beat them hard.

Still a whole year until they have a full machine, but the 512-way prototype reached 1.4 TFlops (LinPack). The complete machine will have 128 times the nodes and 50% higher frequency. So even with pessimistic scalability, this will be more than twice as fast.

Re:They better hurry ... by Anonymous Coward · 2004-05-12 02:07 · Score: 0

Blue gene is a parallel processing computer. Paralell processing and clustering is only effective for problems which can effictively be broken down into orthogonal parts. Some problems such as fractals where the next calculation is dependant on the previous one cannot be thus broken down. If Blue gene tried to tackle this kind of problem it could only make effective use of a single processessor and it would progress only at the speed of a single processer. Single power chips are fast but not nearly good enough for a super computer.
Re:They better hurry ... by flaming-opus · 2004-05-12 02:08 · Score: 4, Informative

Two radically different designs, will probably solve very different sorts of problems. Linpack is extremely good at giving a computer an impressive number. It's the sort of problem that fills up execution piplines to their maximum. Blue Gene was origionally designed to do protein-folding calculations. While many other tasks will work well on that machine, others will work very poorly.

It's a mesh of a LOT of microcontroller-class processors. The theory being that these processors give you the best performance per transistor. Thus you can run them at a moderate clock, get decent performance out of them, and cram a whole hell of a lot of them into a cabinet. It's a cool design, I'm interested to see what it will be able to do, once deployed. However, for the problems they have at ORNL, I'm sure the X1 was a better machine. Otherwise they would have bought IBM. They already have a farm of p690s, so they have a working relationship.
Re:They better hurry ... by realSpiderman · 2004-05-12 02:34 · Score: 1

Of course, I was referring to the Top500. Vector Processor Computers like the NEC SX and the Cray's provide far better performance (efficiency) for some problems. (After all, the ORNL people probably know where to stick the 50M.)
But Bluegene also has a SIMD FPU and a very, very low latency interconnect with two topologies (one torus and one tree). This network is the strength of Bluegene. They reached 70% of the theoretical performance of the whole system for a 512-way system (that is 1024 processors).
Re:They better hurry ... by Anonymous Coward · 2004-05-12 03:00 · Score: 0

IBM news page estimates that it will have a peak performance of 360 trillion operations per second when completed.

I don't see it holding #1 place for very long (if at all)...

50 trillion calcs/sec...how fast really? by Debian+Troll's+Best · 2004-05-12 01:43 · Score: 4, Insightful

I love reading about these kinds of large supercomputer projects...this is really cutting edge stuff, and in a way acts as a kind of 'crystal ball' for the types of high performance technologies that we might expect to see in more common server and workstation class machines in the next 10 years or so.

The article mentions that the new supercomputer will be used for non-classified projects. Does anyone have more exact details of what these projects may involve? Will it be a specific application, or more of a 'gun for hire' computing facility, with CPU cycles open to all comers for their own projects? It would be interesting to know what types of applications are planned for the supercomputer, as it may be possible to translate a raw measure of speed like the quoted '50 trillion calculations per second' into something more meaningful, like 'DNA base pairs compared per second', or 'weather cells simulated per hour'. Are there any specialists in these kinds of HPC applications who would like to comment? How fast do people think this supercomputer would run apt-get for instance? Would 50 trillion calculations per second equate to 50 trillion package installs per second? How long would it take to install all of Debian on this thing? Could the performance of the system actually be measured in Debian installs per second? I look forward to the community's response!

Re:50 trillion calcs/sec...how fast really? by Anonymous Coward · 2004-05-12 01:46 · Score: 1, Informative

Intallation is dependent on disk speed not mips. This computer lends itself more towards computional problems like solving RSA keys or finding new primes.
Re:50 trillion calcs/sec...how fast really? by Anonymous Coward · 2004-05-12 02:10 · Score: 1, Interesting

Fractal iteration is also a very good use for this machine.
Re:50 trillion calcs/sec...how fast really? by Anonymous Coward · 2004-05-12 02:11 · Score: 0

Thats obvious... 200 libraries of congress
Re:50 trillion calcs/sec...how fast really? by MtViewGuy · 2004-05-12 02:35 · Score: 1

Well, besides weather simulation (which is among the most CPU-intensive work around), they could use this new computer to do computational fluid dynamics analysis--perfect for studying the aerodynamics of airplanes, shaping the aerodynamics of an automobile, and possibly studying how to reduce noise on a maglev train travelling at over 250 mph.
Re:50 trillion calcs/sec...how fast really? by joib · 2004-05-12 02:47 · Score: 1

Quantum chemistry, or ab initio, calculations tend to be a biggie. I wouldn't be surprised if ab initio alone would account for > 50 % of all supercomputer cpu cycles in the world.

Other big things are weather prediction, fluid dynamics, classical (i.e. "Newtonian") molecular dynamics with some kind of empirical potentials (e.g. protein folding and stuff can be thought of as MD).
Re:50 trillion calcs/sec...how fast really? by Anonymous Coward · 2004-05-12 03:16 · Score: 0

... weather simulation IS computational fluid dynamics, along with some thermo and heat transfer...
Re:50 trillion calcs/sec...how fast really? by Analogy+Man · 2004-05-12 03:23 · Score: 1

The fellow children posts elude to it, but more directly, these sorts of hardware implementation (like NASA Ames big iron etc) are allocated out to various research projects. Some may go to companies doing governement funded research, a lot will be farmed out to PhD candidates and post-doc researchers in academia.
Typically these folks will submit proposals for what they will be doing, and some review board will dole out the cycles. Queues are set up with different processor, run duration and memory allocations.
Back in the mid 90's I was running Computational Fluid Dynamics code for sonic boom propocation/mitigation research on what was then the largest super computer in the unclassified world. They had low priority queues that ran if all the regular ones that burned up your allocation cleared. Over the Easter holiday in '94 I had every processer on that Cray running full tilt on my stuff for about 12 hours. I had trouble getting my data off disk (they limited that too) in CA to Seattle fast enough!
Give researches more horsepower and they will run bigger problems. At the time, boundary layer researchers would set up problems will millions of grid points for flow over a flat plate. This sort if research is a true "build it and they will come" scenario.

--
When the people fear their government, there is tyranny; when the government fears the people, there is liberty.
Re:50 trillion calcs/sec...how fast really? by meadowsp · 2004-05-12 03:46 · Score: 1

Insightful???

Troll more like.
Re:50 trillion calcs/sec...how fast really? by Anonymous Coward · 2004-05-12 09:09 · Score: 0

The article mentions that the new supercomputer will be used for non-classified projects. Does anyone have more exact details of what these projects may involve?

It involves counting votes in Florida elections. And recounting. And recounting.
Re:50 trillion calcs/sec...how fast really? by smooth1 · 2004-05-12 10:37 · Score: 1

Measure it in seti@home packets. My XP2700 quotes something like 6 (or it is 10) hours per packet! So I could relate to this. Would this be packets/second or seconds/packet?
Re:50 trillion calcs/sec...how fast really? by Goldsmith · 2004-05-12 16:06 · Score: 1

Some people here have mentioned nanotechnology simulations, I don't know that the label needs to be thrown around so much but I have written a few models of the type which will be used. I make, every once in a while, a computer model to explain my experimental results (usually for myself, when I don't believe something).

Generally, these are voltage and force relaxations, with some areas of well defined voltage, some point charges thrown around, and very complex geometry. Basically, that means I set up a system which would be very difficult to solve analytically, describe Laplace's equation on a grid, and iterate until it's satisfied everywhere on my grid. It takes a long time, and is probably hugely inefficient, but it will give me all the fringing fields and strange interactios I need.

I'm not a computational physicist, so none of my stuff will end up there, but that's the kind of thing that will be running. Laplace's equation describes a huge number of phenomena, and is easy to describe computationally in a grid. What would be cool to know is how many X by Y by Z grid iterations can be done a second. Is it 50 trillion/(XYZ)? My God, that would be awsome! Unfortunately, I don't know the right answer.

Maybe it's me. by jacoplane · 2004-05-12 01:46 · Score: 1

Can anyone explain what "DOE" is? I'm assuming it's some american govt thing like department of energy. is that correct?

Re:Maybe it's me. by jacoplane · 2004-05-12 01:48 · Score: 1

RTFA
Re:Maybe it's me. by Killjoy_NL · 2004-05-12 01:49 · Score: 2

That's correct, it's the Department of Energy.

I don't know why they would need it, but that's just because I don't know anything about the work of the DOE (not being an american and all that)

--
This is the sig that says NI (again)
Re:Maybe it's me. by henryhbk · 2004-05-12 01:51 · Score: 4, Informative

Yes, DOE is the Federal Government's Department of Energy. Oak Ridge is a large federal govt. lab.
Re:Maybe it's me. by Anonymous Coward · 2004-05-12 01:59 · Score: 0

Yes. It's the Department Of Erections. They're going to build the Internet's fastest porn server.

But the computer's record will be short-lived... by BrentRJones · 2004-05-12 01:48 · Score: 2, Funny

...because a day later Palm users will massively interconnect to form the World Fastest Clustered Computer Environment. The OS? Linux, of course. .}

--
Help end the use of Sigs. Tomorrow

It's Longhorn compatible then ? by Anonymous Coward · 2004-05-12 01:52 · Score: 2, Funny

or it certainly seems like it (reading the specs of the thing)

First Ninnle Question... by Anonymous Coward · 2004-05-12 01:53 · Score: 0

Will they be smart enough to install the version of Ninnle Linux already ported to the Cray XI architecture?

Huh? by NegativeK · 2004-05-12 01:54 · Score: 1

. . . it will be capable of sustaining 50 trillion calculations per second.

Does anyone else not gain anything from that statement? 50 trillion calculations means very little if all it can do is flop a bit back and forth 50 trillion times. Perhaps someone could come up with a better benchmark, like the largest number it can factor in a minute, on average. Even then we may be talking about huge exponents in scientific notation. =p

It could just be the fact that it is ten in the morning, but 5*10^13 seems incomprehensbily large.

--
This statement is false.

Re:Huh? by flaming-opus · 2004-05-12 03:16 · Score: 2, Informative

The important part of the statement is "Sustaining". There are a lot of computers out there on the top500 list that get peak numbers way ahead of their sustained numbers. An Army reseach center (www.arc.umn.edu) published a comparison of a xeon cluster and the X1. For their codes (weather simulation, material sciences, air flow, etc) the Xeons sustained performance was 5% of peak. The Cray was about 30% of peak. (this is probably due to the really awesome memory bandwidth of the cray)

You're correct that these are just numbers so lets talk about a real problem. The AHPCRC reported that a 32 processor cray X1 (peak 400 Gigaflops, 66 gflops realized) was able to simulate a weather model of the entire US with 33 vertical levels at 5Kilometer resolution in just under 2 hours. Today these models are done at 10KM resolution with 20 levels. IF you take this theoretical ornl system and assume (peak 60-80TF, 40 sustained on easy codes, 15 sustained on hard codes) then they might do a 2KM simulation with 45 layers in 1 hour.
Re:Huh? by Anonymous Coward · 2004-05-12 03:22 · Score: 0

In answer to your question it could factor a number up to about 1.8*10^24 in 1 second the relationship is nonlinear and for two seconds its about 9*10^24. These numbers assume that mod is a single calculation.
Re:Huh? by heydrick · 2004-05-12 10:10 · Score: 1

For ORNL's performance evaluation of the X1 vs. the Altix, SP, and Alpha, see:
ORNL Cray X1 Evaluation.
The X1 comes out on top on nearly every benchmark.

So... by Anonymous Coward · 2004-05-12 01:54 · Score: 0

you'd be moving away from Outlook then?

Re:So... by Anonymous Coward · 2004-05-12 03:32 · Score: 0

you'd be moving away from Outlook then?

wouldn't you?
Re:So... by suso · 2004-05-12 07:23 · Score: 1

Or would it be moving away from you?

Someone mod the parent up, it's funny.

DOE = DOD by Anonymous Coward · 2004-05-12 01:58 · Score: 0

...capable of sustaining 50 trillion calculations per second.

Seems like a lot of computing power for non-classified science

I suspect it'll just run 'computations' to work out how to blow up Muslims quicker...........

2 Years? by XMyth · 2004-05-12 01:58 · Score: 3, Informative

I don't think Crays that were build 5 years ago are considered obsolete by anyone's standards.

Clusters solve different jobs than supercomputers. Sometimes they bleed into one another, but there are some things supercomputers will always be better at (because of higher memory bandwidth for one thing).

No bombs? by stevesliva · 2004-05-12 01:59 · Score: 1

I couldn't find the source for the "non-classified" bit... These things are often not used for simulating new bombs but for, "evaluating the stability of the nuclear stockpile." Does research into whether the yield of our cold war nukes is down or up a few kilotons qualify as non-classified?

--
Who do you get to be an expert to tell you something's not obvious? The least insightful person you can find? -J Roberts

Re:No bombs? by Anonymous Coward · 2004-05-12 02:07 · Score: 0

Oh that's very classified.

But even so, they find time to squeeze in a little fusion research, and in at least one instance virtually crashing a mars sized planet into earth to see where our moon might have come from. But yes bombs bad. If it wasn't for the cold warriors we'd all be enjoying our lives in a gruel loving communist utopia. Boo US why aren't tax dollars spent filling everyone's lives with prettier flowers. Boo.
Re:No bombs? by Patrik_AKA_RedX · 2004-05-12 02:23 · Score: 1

"evaluating the stability of the nuclear stockpile."
Why don't they just print a best-use-before-date on those nukes?
Re:No bombs? by Bahumat · 2004-05-12 02:49 · Score: 1

Oh sure, a best before date on nuclear half-lives?

"Best Before 1.7684^18^2 seconds after manufacture."

--
"To pass through the jungle; silence, courtesy, ferocity, as the occasion demands." -- Kamau, "Proper Passage"
Re:No bombs? by Anonymous Coward · 2004-05-12 07:48 · Score: 0

ORNL is not a nuclear weapons facility (those are Los Alamos, Sandia, and my own Livermore). On top of that, if the computer is not physically located in a classified area, it cannot be used for classified applications. Since locating it in a classified building means it cannot be connected to the internet, it is really a binary thing: either the machine is open to academics and others via the internet and then can do no classified computing, or it is a classified machine with no external access via the net.

Cray X1.. What role do IBM and SGI have? by freelunch · 2004-05-12 02:01 · Score: 2, Informative

They were listed as part of the solution.

Oak Ridge has done extensive evaluations of recent IBM, SGI and Cray technology. Though I am still looking forward to data on IBM's Power5.

Cray X1 Eval
SGI Altix Eval

Re:Cray X1.. What role do IBM and SGI have? by flaming-opus · 2004-05-12 02:39 · Score: 2, Informative

ORNL already has a 256 processor X1, a large IBM SP made of p690s, as well as a large SGI altix. I imagine the 50Tflops number will be a combined system with upgraded systems of all three types. They are obviously impressed with both the X1 and the Altix. The IBMs are no slouch though, and they are upgrading the interconnect, and IBM is just getting ready to launch a power5 update.

It's probably just spin to call the project "A computer", rather than "several computers". Deep in one of those ORNL whitepapers you see that they are planning to cluster together these three machine's with a cluster filesystem. You throw in a clustered batch control system and you can kinda call it "A" supercomputer. Really it's a cluster, except each of the nodes may have a thousand processors. We'll have to wait and see what it really looks like.
Re:Cray X1.. What role do IBM and SGI have? by Anonymous Coward · 2004-05-12 02:45 · Score: 0

I have the same question. Nothing in ANY of the links in the story back up the claim that it will be based on Cray's arch.

Being Snide Here by Seanasy · 2004-05-12 02:02 · Score: 4, Insightful

I think ORNL and PSC know a lot more about supercomputing than you (or Internet rag pundits) do. As others have noted, there are real reasons for Big Iron.

Clusters are great for certain problems but for heavy computation -- think simulating two galaxies colliding or earthquake modeling -- off the shelf clusters don't cut it.

They're not wasting tax-payer money unless you consider basic researcher a waste.

Grab that cash with both hands and make a stash by HarveyBirdman · 2004-05-12 02:04 · Score: 1, Offtopic

Since it's funded by federal grants, how much time, as a taxpayer, do I get on it?

And I'm still waiting for my turn to drive one of the Mars rovers.

--
--- Ban humanity.

Re:Grab that cash with both hands and make a stash by sean23007 · 2004-05-12 04:12 · Score: 0, Troll

You don't seem to understand what government means.

The government has spent billions fighting a war overseas; when is it your turn to go fight?

--

Lack of eloquence does not denote lack of intelligence, though they often coincide.

3D torus topology by elwinc · 2004-05-12 02:05 · Score: 4, Informative

I checked out the topology of the Cray X1; they call it an "enhanced 3D torus." A 3D torus would be if you made an NxNxN cube of nodes, connected all ajacent nodes (top, bottom, left, right, front, back), and then connected all the processors on one face thru to the opposite face. I can't tell what an "enhanced" torus is. (Each X1 node, by the way, has four 12.8 gflop MSPs, and each MSP has eight 32-stage, 64 bit floating point pipelines.)

So each node is directly connected to six ajacent nodes. Contrast this with the Thinking Machines Connection Machine CM2 topology, which had 2^N nodes connected in an N dimensional hypercube. So each node in a 16384 node CM2 was directly connected to 16 other nodes. There's a theorem that you can always embed a lower dimensional torus in an N dimensional hypercube, so the CM2 had all the benefits of a torus and more. This topology was criticized because you never needed as much connectivity as you got in the higher node-count machines, to CM2 was in effect selling you too much wiring.

Thinking Machines changed the topology to fat trees in the CM5. One of the cool things about the fat tree is it allows you to buy as much connectivity as you need. I'm really surprised that it seems to have died when Thinking Machines collapsed. On the other hand, any kind of 3D mesh is probably pretty good for simulating physics in 3D. You can have each node model a block of atmosphere for a weather simulation, or a little wedge of hydrogen for an H-bomb simulation. But it might be useful to have one more dimension of connection for distributing global results to the nodes.

--
--- Often in error; never in doubt!

Re:3D torus topology by PetoskeyGuy · 2004-05-12 02:46 · Score: 2, Funny

So each node is directly connected to six ajacent nodes.

Excellent. We can finally solve the Optimal Dungeon Theorem on hex tile games.
Re:3D torus topology by hawkstone · 2004-05-12 03:20 · Score: 1

I don't think the fat tree died with Thinking Machines. For example, MCR at LLNL uses a Quadrics fat tree. I imagine many sizeable clusters (way more than 64 nodes) use one. There's one link here, and the MCR link here but you can probably google for quadrics and fat tree to find some more. I'd be surprised if fat trees didn't show up in Myrinet / other interconnects, but you typically need to have a sizeable cluster before there's any point in calling it a fat tree.

(Oh, and if you meant something else entirely by fat tree, I apologize. I'm not too familiar with the particulars of the CM5 fat tree, so the Quadrics one is the only usage I'm aware of.)
Re:3D torus topology by natet · 2004-05-12 04:19 · Score: 1

Fat trees are still alive and well. It appears to be Quadrics topology of choice, as it is applied at LLNL and PNNL, which both user thier interconnect. I'm not sure the folks at ORNL would have specifed a Torus, unless they believed that they could make use of it. I know those guys, and they are some very smart people. I don't recall hearing a reason for the topology decision though.

--
IANAL... But I play one on /.
Re:3D torus topology by elwinc · 2004-05-12 05:23 · Score: 1

thanks for the quadrics pointer!

--
--- Often in error; never in doubt!

Is competition good? by mr_tap · 2004-05-12 02:08 · Score: 1

Man I hope Virginia Tech buys some more Xserve G5s - they are slipping down the ranks :)

Numbers... meaningful numbers? by hkfczrqj · 2004-05-12 02:08 · Score: 1

50 trillion of calculations per second. Is that a synonym of flop (floating-point operation)? ...

How does this computer compares with the BlueGene/L (131,072 cpus, 0.5 Petaflops -estimated)? Don't be mislead by the name (*Gene)... this will be a computer for classified simulations (it will have a 1-2 year long "science run", for testing purposes with non-classified simulations).

Cheers...

Re:Numbers... meaningful numbers? by Anonymous Coward · 2004-05-12 06:58 · Score: 0

How does this computer compares with the BlueGene/L?

Part of this machine already exists and was used for evaluation of the technology. This is also a vector supercomputer, which is vastly different.

Outlook? by Anonymous Coward · 2004-05-12 02:09 · Score: 0

Will Doom 3 run on it?

Hyphenation Troll by Anonymous Coward · 2004-05-12 02:14 · Score: 0

I've noticed the incorrect use of hyphens a lot on Slashdot, so I thought I'd submit a hyphenation troll when I saw a misuse to try and improve things. :)

In the story text it says, "dedicated exclusively to non-classified scientific research."

1) Latin prefixes (such as "non", "un", "anti", "pre", "nano", etc.) almost never require hyphens. Thus it should be written as "nonclassified" (although I believe "unclassified" is more common).

Hyphenation Troll.

Re:Hyphenation Troll by Archangel+Michael · 2004-05-12 02:50 · Score: 1

Nonclassified implies that it was never classified, while unclassified implies that it was once classified, but now is not.

I don't know which is more accurate in this case being the typical slashdotter, and not actually reading the article.

Which brings me to my point. Half the time I don't even bother trying to read the article and the other half the time it is slashdotted, which is about the same result. :-/

--
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.
Re:Hyphenation Troll by stanmann · 2004-05-12 02:58 · Score: 1

Ok, nonclassified means that there is no classification applied.

Unclassified means that the material has been assigned the classification of Unclassified... there are other classifications too.

For example personal e-mail at home is generally nonclassified. e-mail on a government computer would generally be unclassified.

--
Food not Bombs is a nice platitude but it breaks down when you notice that the Bombees are usually well fed
Re:Hyphenation Troll by Archangel+Michael · 2004-05-12 03:09 · Score: 1

You are speaking about Government definitions, regarding specifically intellegence classification system. I was using a more broad application in my definition, that would be more applicable outside of the intellegence community.

In fact, the intellegence definition is the typical oxymoron. Classified as "unclassified" is typical government stupidity. Think about it.

--
Agent K: A *person* is smart. People are dumb, stupid, panicky animals, and you know it.

Fighting the temptation ... by Rico_za · 2004-05-12 02:16 · Score: 1

... to post the usual jokes, I've got to ask: What runs on these kind of machines? What OS do they use, and what kind of software? Can you buy software for supercomputers, or will the customer/new owner have to write all the software to run on it themselves? Anyone out there working on something similar have interesting facts about the software?

Re:Fighting the temptation ... by compupc1 · 2004-05-12 02:36 · Score: 2, Informative

Supercomputers usually run some flavor of UNIX -- Unicos, IRIX, I think even Linux. In any case, they are specially built and designed for the supercomputer. Supercomputers are used for highly specialized scientific applications, and as such the programs would be specially written in Fortran, C, or Assembly, and often specially optomized for the architecture.

--
-James
Re:Fighting the temptation ... by flaming-opus · 2004-05-12 02:49 · Score: 3, Informative

The SGI altix runs a hacked up version of linux that's part 2.4 with a lot of backported 2.6 stuff as well as the Irix scsi layer. They are migrating to a pure 2.6 OS soon. The IBM system runs AIX 5.2. The Cray runs Unicos, which is a derivative of Irix 6.5, though they seem to be moving to Linux also. I'm gonna geuss that they run totalview as their debugger. They use DFS as their network filesystem. They have published plans to hook all these systems up to the Stornext filesystem which does Heirchical Storage Management. MPI and PVM are likely important libraries for a lot of their apps.

For these sorts of machines, one can by utilities for data migration, backup, debugging, etc. However, the production code is written in-house, and that's the way they want it. Weather forcasting, for example, uses software called MM5, which has been evolving since the Cray-2 days, at least. A lot of this code is passed around between research facilities. It's not open source exactly, but the DOD plays nice with the DOE, etc.

The basic algorithms have been around for a long time. In the early 90's, when MPPs and then clusters came onto the schene, a lot of work was done in structuring the codes to run on a large number of processors. Sometimes this works better than other times. Most of the work isn't in writing the code, but rather in optomising it. Trying to minimize the synchronous communication between nodes is of great importance.
Re:Fighting the temptation ... by Seanasy · 2004-05-12 03:51 · Score: 1

While much of software will be custom applications, there are common packages that you'll find for simulatiing molecular interactions, doing sequence analysis, etc.

You can check out a list of software available on a CRAY T3E to get an idea.

Imagine... by EduardoFonseca · 2004-05-12 02:17 · Score: 0, Redundant

a beowulf cluster of these!

as a former DOE employee by bsDaemon · 2004-05-12 02:18 · Score: 5, Interesting

I worked in Instrumention and Control for the Free Electron Laser project at the Thomas Jefferson National Accelerator Facility. We also host the CEBAF (Concentrated Electron Beam Accelerator Facility), which is a huge ass particle accelerator.
the DOE does a lot of basic research in nuclear physics, quantam physics, et cetera. the FEL was used to galvanize power rods for VPCO (now Dominion Power) and made them last 3 times as long. Some William & Mary people use it for doing protein research, splicing molecules and stuff.
The DOE does a lot of very useful things that need high amounts of computing power, not just simulating nuclear bombs (although Oak Ridge does taht sort of stuff, as does Los Alamos). We only had a lame Beowulf cluster at TJNAF. I wish we would have had something like this beast.
I want to know how it stacks up to the Earth Simulator.

Re:as a former DOE employee by Killjoy_NL · 2004-05-12 02:25 · Score: 1

Cool, thanks for the info, it is interesting :)

--
This is the sig that says NI (again)
Re:as a former DOE employee by mgoulding · 2004-05-12 04:40 · Score: 1

You may have already found out by now... in any event, it will be able perform 14 trillion more calculations/second than ES (ES can do 36 trillion).

Cray X1 OS is.. by freelunch · 2004-05-12 02:24 · Score: 1

SGI's IRIX.

That detail is kept under pretty tight wraps by Cray. It is licensed from SGI and is discolosed as a business risk in their regulatory filings.

IRIX has always been my favorite UNIX.

Re:Cray X1 OS is.. by flaming-opus · 2004-05-12 02:57 · Score: 1

I geuss. It's right in the user manual, which is published on the website. They don't run around with Irix pom-poms and wave little irix flags, but they aren't shy about it.

Cray is a company that sells to huge research labs, and fortune 500 companies. Just because they don't appear on TomsHardware, or do interviews for /. doesn't mean they aren't saying anything about it. "know your audience", and all that.

Cool video on Cray X1 by Anonymous Coward · 2004-05-12 02:29 · Score: 0

Great video with a fair bit of detail on the X1's cooling system, chip density, etc.

View it with mplayer:

Cray X1

Posted Anon to not karma whore.

NOT the fastest! by VernonNemitz · 2004-05-12 02:30 · Score: 4, Interesting

It seems to me that as long as multiprocessor machines qualify as supercomputers, then the Google cluster counts as the fastest right now, and will still count as the fastest long after this new DOE computer is built.

Re:NOT the fastest! by compupc1 · 2004-05-12 02:32 · Score: 2, Insightful

Clusters and supercomputers are totally different things, by definition. They are used for different types of problems, and as such cannot really be compared.

--
-James
Re:NOT the fastest! by tony_gardner · 2004-05-12 03:15 · Score: 1

Depends what you're doing. Something like google or SETI or frame rendering scales very well to a cluster, because the amount of internode communication required is very low.

Something like CFD or FEM is about in the middle, which is to say that clusters and SCs do about as well as each other. This is because, although there is a requirement that nodes communicate, the amount of communication is relatively low compared to the amount of internal computation. ie each cell is mostly affected by the cells directly neighbouring it and the previous status of the cell.

Something like plasma or MHD modelling just doesn't work on a cluster because at each iteration, every node needs to hear from every other node, so the speed of the computation is basically dependant only on the rate of communication between nodes.

What that means is that one google PC can calculate at a significant fraction of the speed of the whole cluster. Which is why you need a supercomputer.

More sinister intentions... by slim_jimmy · 2004-05-12 02:31 · Score: 1

(Tin foil hat on) The labs at Oak Ridge got their start during the 1940's and were the source for the enriched uranium used in the first atomic bombs used against Japan and in the deserts of New Mexico. Perhaps the DOE has ulterior motives when it comes to using those extra floating point operations...

Comment removed by account_deleted · 2004-05-12 02:41 · Score: 0, Redundant

Comment removed based on user account deletion

Slashdot Stock Effect by imsirovic5 · 2004-05-12 02:48 · Score: 0, Offtopic

Nice going guys! This story gives a whole new meaning to Slashdotting. I own cray stock (CRAY on Nasdaq) and since this article got posted stock got quite a boost. Now if I can only get you guys to post something positive about my gambling penny stocks that I hold position in....

failure rate .... one of those nodes will fail by Anonymous Coward · 2004-05-12 02:53 · Score: 0

The same thing will happen in the X1.

The major difference is that the rate may not be quite as high (past experience indicates 1/2 failures per week), AND the system does have a checkpoint/restart capability.

That usually allows for a recovery of all but the last 5 minutes-2 hours of computations (the range is both system and user defined).

SkyNet by TheReal_BarkMan · 2004-05-12 02:55 · Score: 1

This is the beginning of the end... Marry this supercomputer with AI and BitKeeper access to the kernel and who knows what will happen next.

Handheld version? by Anonymous Coward · 2004-05-12 02:57 · Score: 0

I'd like mine as a hand held. Then again battery life might be a bit short.

Not quite... by chudmung · 2004-05-12 02:59 · Score: 1, Informative

ASCI Purple (IBM Power5) is capable of 100 teraflops. The Blue Gene/L machine is capable of 367 teraflops.

This press release is almost 18 months old, btw...
http://www-1.ibm.com/servers/eserver/pseri es/news/ pressreleases/2002/nov/asci_purple.html

Maybe the headline "fastest -unclassified- supercomputer" would be more fitting.

--
...

Re:Not quite... by Anonymous Coward · 2004-05-12 07:53 · Score: 0

BG/L will be entirely unclassified.
Re:Not quite... by Anonymous Coward · 2004-05-12 14:10 · Score: 0

are you sure? We're talking about Livermore here (you know, nukes, lasers). BTW, a guy from LLNL told me they will be using BG/L for classified research (after the science run, of course).

And, as a side note, BG/L it's NOT capable of 367 TFlops... yet (the tests are encouraging... will reach probably ~500 TFlops).

DOE "user facility" by Richard+Mills · 2004-05-12 02:59 · Score: 1

"Will it be a specific application, or more of a 'gun for hire' computing facility, with CPU cycles open to all comers for their own projects?"

This will be what is known as a "user facility" at DOE. CPU time will be doled out on a competetive basis, i.e., if someone has a project they would like to use it for, they will submit a proposal which will then be reviewed against others.

Re:Cray X1 OS is.. UNICOS by Anonymous Coward · 2004-05-12 03:04 · Score: 0

Sorry - it is UNICOS.

IRIX cannot support the torus, and the OS is not designed to high speed operation with dedicated application CPUs.

The way UNICOS works is to assign certain processors that are closely connected with the IO subsystem as device nodes. These are the only CPUs that can perform I/O.

Other CPUs are assigned the general duty of the OS - these support the usual scheduling activities, others are assigned to interactive functions: command interpreters, compilers, base utilities (though some utilities are flagged for parallel use). ALL of the remaining processors are designated for computation.

There are usually 4 processors associated with a single IO channel, and though all 4 won't be using it, the IO capability should failover to another if an IO cpu fails. The actual number of IO cpus can vary. (a 1024 cpu unit might have 12 I/O cpus, 16/24 OS cpus, and 8-12 command cpus, with the remaining available for pure computation.

IRIX just can't cut the mustard.

Re:But the computer's record will be short-lived.. by Anonymous Coward · 2004-05-12 03:04 · Score: 0

what interconnect do you propose they use? IR? Bluetooth?

The pinnacle of modern technology by Anonymous Coward · 2004-05-12 03:12 · Score: 0

and i bet all you wieners care about is whether or not it runs linux

classified nonsense by vandenberg5 · 2004-05-12 03:20 · Score: 1

I love how the poster writes that all classified government work deals with bombs. There are just a few more programs than "bombs".

Re:classified nonsense by Little+Brother · 2004-05-12 04:46 · Score: 1

The poster never says anything of the sort. All he says is that the computer will not be used for military projects. The "not bombs" bit does not appear to be all-inclusive. Considering the computer is being developed in a city which traces its origions back to "the bomb" it is a rational consideration that it might be used in bomb/missle research, so it is appropriate to point out that it will not be.

--

Little Brother, watching the watchers

One word: NS-Equations by vogon+jeltz · 2004-05-12 03:21 · Score: 1

Run a test case of a Navier-Stokes problem (just one of many which spring to mind; solving the Maxwell equations isn't actually fun either) on both a shared memory versus a clustered machine with poor IO bandwidth. Compare.
The shared menory machine will beat the shit out of those nowadays (admittedly) cheap clusters. Comparing bang for bucks is an entirely different issue though ...

Just my 0.01Euro.

I find it interesting by Eagle5596 · 2004-05-12 03:22 · Score: 0, Flamebait

that when one looks at their available graduate research programs, they only have summer research opportunities for blacks and other minorities. Why is this legal? Should we be allowing a government facility to participate in such extreme racism? Why should someone who is more qualified for the same position be denied the position solely on the basis of the color of his/her skin? I've got news for you, if said minority is in grad school, and already has the skills necessary to be useful at a government labs graduate research, they don't need affirmative action.

Re:I find it interesting by Anonymous Coward · 2004-05-12 14:25 · Score: 0

Look around a bit more and you'll find more positions. Check www.orau.edu.

I grew up in Oak Ridge by Walker · 2004-05-12 03:23 · Score: 1

And I remember what was happening when the Cold War was declared "over": the lab funding started to dry up. K-25 was shut down while X-10 (aka ORNL) and Y-12 were scaled back a lot. So these labs were forced to re-examine what they do for better funding opportunities.

X-10 (ORNL) has branched out into a lot of helpful areas. Some of its projects include environmental cleanup and alternative energy production. It also spends a lot of resources on testing how to safely store and transport dangerous waste (a friend's dad was one of those people that drops containers all day). Any of these could be candidates for this computer.

Any tin-foil hats should be directed at Y-12. That's the DOD plant; X-10 is just DOE.

Re:I grew up in Oak Ridge by cot · 2004-05-12 03:42 · Score: 1

"X-10 is just DOE"

And since DOE does our nuclear weapons work, i don't know if assuming the DOE is benign relative to the DOD is prudent.

--
Re:I grew up in Oak Ridge by Walker · 2004-05-12 03:46 · Score: 1

The security for X-10 is substantially less than Y-12. Y-12 is where all the serious work of that type is done.
Re:I grew up in Oak Ridge by Rostin · 2004-05-12 03:50 · Score: 1

I'm not sure that automatically assuming nuclear weapons work isn't benign is especially prudent, either.
Re:I grew up in Oak Ridge by cot · 2004-05-12 04:01 · Score: 1

I wouldn't consider any kind of weapons work to be benign, and that certainly applies to nuclear weapons.

Pick one

--

EHEM by Anonymous Coward · 2004-05-12 03:24 · Score: 0

The fastest supercomputer is to be built in Spain over the next two years, financed by IBM

Un-classified research uses by bradbury · 2004-05-12 03:25 · Score: 3, Interesting

One of the major un-classified research uses is for molecular modeling for the study of nanotechnology. This really consumes a lot of computer time because one is dealing with atomic motion over pico-to-nano-second time scales. An example is the work done by Goddard's group at CALTECH on simulating rotations of the Drexler/Merkle Neon Pump. If I recall properly they found that when you cranked the rotational rate up to about a GHz it flew apart. (For reference macro-scale parts like turbochargers or jet engines don't even come close...)

In the long run one would like to be able to get such simulations from the 10,000 atom level up to the billion-to-trillion (or more) atom level so you could simulate significant fractions of the volume of cells. Between now and then molecular biologists, geneticists, bioinformaticians, etc. would be happy if we could just get to the level of accurate folding (Folding@Home is working on this from a distributed standpoint) and eventually to be able to model protein-protein interactions so we can figure out how things like DNA repair -- which involves 130+ proteins cooperating in very complex ways -- operate so we can better understand the causes of cancer and aging.

Doom3 by Anonymous Coward · 2004-05-12 03:28 · Score: 0

The first computer that will be able to run Doom3 at a decent 40fps!

Bomb usable life by Mark+of+THE+CITY · 2004-05-12 03:50 · Score: 1

No, there are other components that go bad over time.

Also, in nukes, the short-lived component is the initiator, which is based on an alpha emitter with a half-life of a few months. They have to be changed out regularly.

--
The clearance system sounds logical. It is not. It is completely arbitrary. -- John Bolton

Actually, after auditing, it looks like you owe us by pastafazou · 2004-05-12 03:52 · Score: 2, Funny

As a direct percentage of total taxpayers, your time would be equal to under one second. However, when calculated as a percentage of your tax contributions in relation to all tax revenues collected, it looks like you still owe us 23 days, 17 hours, and 54 minutes of processing time on your computer. You can drop your computer off at the closest IRS office to you.
Thank you for your understanding in this matter,

Your friendly neighbourhood IRS agent.

Folding@Home URL by bradbury · 2004-05-12 03:55 · Score: 2, Informative

Sorry, it looks like the URL has changed. The home page for Folding@Home is here.

What makes for fastest? by nick_davison · 2004-05-12 04:11 · Score: 1

As usual, there should be a qualifier as to what is meant by fastest.

When complete it will be capable of sustaining 50 trillion calculations per second.

Screw that. How many fps can it manage in Quake III?

Re:What makes for fastest? by madamimadam · 2004-05-12 15:30 · Score: 1

Most gamers would tell you "not enough"

It would give great bragging rights but you always have to worry that your mate will get something better next week.

Also, where do you put the neon lights?

I Guess the Real Question is... by TimTheFoolMan · 2004-05-12 04:24 · Score: 1

...will it be able to run Longhorn?

Tim

Re:I Guess the Real Question is... by dotslasher_sri · 2004-05-12 10:38 · Score: 1

By the time longhorn is released, the supercomputer will be outdated :(

Thinking Ahead by crashnbur · 2004-05-12 04:32 · Score: 1

Warning: abstract thoughts ahead.

Considering the whole of spacetime as a single unit, with our perception limited to only one piece of it at a time, it occurs to me that perhaps everything in both our future and past exists all at once; we're just sliding down a scale as the next section is revealed to us.

That said, wouldn't it make sense that the world's fastest computer is among the very last "super" computers built, many years (centuries? millennia?) in our future (if you want to call it that)? No computer we build today could possibly ever be the world's fastest unless the world cease's to be.

Besides, according to Douglas Adams, Earth itself is the world's fastest computer, designed by Deep Thought.

Is AMD behind this ? by billcopc · 2004-05-12 04:41 · Score: 1

Why do I have the feeling this is going to be a 64-way nitrous-cooled Athlon64 beast ? And you can bet it will be running SETI.

--
-Billco, Fnarg.com

Wow! by rspress · 2004-05-12 04:53 · Score: 1

At that speed, if it were running Windows XP, the whole internet could be infected with a virus in mere nanoseconds.

cray and fast computing by spamchang · 2004-05-12 04:55 · Score: 1

as a relative n00b to digital systems (i'm taking part 2 of an intro course), it's my understanding that roundoff in larger digits was one of the major contributing factors to the early cray supercomputers' speed. has cray moved on from that design philosophy?

Re:World fattest people are USian GAYS ! by dnahelix · 2004-05-12 05:21 · Score: 1

We're not fat, we're drought and famine resistant.

--
Slashdot Eds Link Anonymous Posts With Logged Posts
They Are Vermin Feeding On Each Other's Feces.
I Hate \.

Doom3 by Anonymous Coward · 2004-05-12 05:31 · Score: 0

it will be capable of sustaining 50 trillion calculations per second

This is 50 terahertz right? Is this machine being built to play Doom 3?

Re:cray and fast computing -- I don't think so by elwinc · 2004-05-12 05:43 · Score: 2, Informative

I believe the early crays implemented 64 bit floating point. Not IEEE floating point; no NaN or Inf codes, but still full precision.

I believe the speed was due to many factors. Here are a few.

(0) 64 bit word and and a ton of registers including eight 64 word vector registers.
(1) very fast memory - at a time when many folks were using magnetic cores, Cray was using multi-transistor static RAM (like in the on-board caches of today's CPUs).
(2) load - store instruction set. Many of the ideas that became popular in 1990s era RISC computers were present in the Cray 1 instruction set. One of the key ones is to separate instructions that read and write main memory from those that operate on data. That way, a program can start fetching data several cycles before the data is needed, and hide the fetch delay.
(3) 16 banks of memory - each bank can handle a fetch independently; another way of overcoming memory latency.
(4) Freon cooling! - does this make Seymour the first overclocker?!

--
--- Often in error; never in doubt!

Awesome! by esdjco · 2004-05-12 06:09 · Score: 0

Ah good another cold war. This is all we need. "No I have the fastest box...No no trust me. I can do 4 billion calculations per nano second, therefore I am a superior country...Didn't we beat you in WWII?"

Re:But the computer's record will be short-lived.. by Anonymous Coward · 2004-05-12 06:34 · Score: 0

Dude, were you literally masturbating when you wrote this?

Ok, now do a M$ joke!

Why not other topologies? by Anonymous Coward · 2004-05-12 06:53 · Score: 0

The hypercube isn't used much because it costs too much in unused interconnect wiring and fat trees are limited in popularity due to poor worst case crosssectional bandwidth, which means it can't handle a certain class of common problems.

CM machines also had other problems with their handling of parallelism for problems with conditionals inside loops. Every processor executed the loop at the same speed, in spite of branch prediction and even if the code was dead for that iteration. CM machines experimented with some hot new ideas, but I'd hardly recommend emulating them exactly.

UPDATE: Cray Press Release Gives New Numbers by Anonymous Coward · 2004-05-12 07:15 · Score: 0

A 100-teraflop (trillions of calculations per second) Cray system at Oak Ridge is planned for 2006, with the potential to grow to 250 teraflops in 2007. Near-term plans call for increasing the capacity of the current Cray X1? supercomputer at ORNL to 20 teraflops in 2004, with a 20-teraflop Red Storm-based system from Cray added in 2005. The systems will be housed in ORNL's new National Leadership Computing Facility in Oak Ridge, Tennessee.

Read the entire Cray and ORNL releases and X1 info at http://www.cray.com

HA!! yea OK!!!! by comet69 · 2004-05-12 07:18 · Score: 0

I somehow DOUBT this machine can top my diskless cluster of 386's...

--
- Hi I'm Linus Torvalds and I pronounce Linux, Lih-nix..

Re:cray and fast computing -- I don't think so by flaming-opus · 2004-05-12 07:23 · Score: 2, Informative

well, 0-4 are all true.

comparing this to early crays is a little difficut though. For the early crays one advantage was vectors and the other was pipelines.

vector processors are cool, because they tend to be much more tolerant of the latency. You issue a load command, and it does loads until the vector-register is full. Equivalent to dozens of loads (and dozens of round trip latency to memory) on a scalar architecture. The same thing applies to the execution units. You tell the CPU ADD R1 R2 R3, and it pumps the first elements of R2 and R3 registers through the ALUs and into R1 and keeps working until it gets through all of the elements in the vector. Later models supported chaining, which allowed the output from one of these operations to feed into the input of another operation. Vector CPUs are very good at keeping the ALUs busy.

The other advantage of the early crays was pipelining. YMP designs, for example, had multiple integer, FP, load/store, and reciprical devide units. All of these (and the dispatch unit) were pipelined, allowing a munch higher clock rate than traditional designs. Multi-pipeline designs are now the norm, (powerPC, Pentium, MIPS, etc.) but were pretty amazing at the time.

The cooling, incidently, was necessary at any clock rate. Early Crays. (well right on through to the T90) used bipolar transistors, rather than CMOS. In this sort of logic you switch current rather than switching voltage. The net result is that the early crays used a TON of electricity and needed massive cooling systems.

Open to all scientists by ggwood · 2004-05-12 07:46 · Score: 2, Insightful

This project claims many big improvements. First, programmers will be available to help parallalize code of scientists, who may be experts at, say, weather or protein folding but may not be experts at parallel code. Further, the facility is supposed to be open to all scientists from all countries and funded by any agnecy. CPU cycles are to be distributed on a merit-only basis, and not kept witin DOE for DOE grantees to use, as apparently has happened within various agencies in the past.

The idea is to make it more like other national labs where - for example in neutron scattering - you don't have to be an expert on neutron scattering to use the facility. They have staff available to help and you may have a grant from NSF or NIH but you can use a facility run by DOE if that's the best one for the job.

I attended this session at the American Physical Society meeting this March and I'm assuming this is the project referred to in the talks - I apologize if I'm wrong there, but this is at least what is being discussed by people within DOE. I'm essentially just summarizing what I heard at the meeting so although it sounds like the obvious list of things to do, apparently it has not been done before.

The prospect of opening such facilities to all scientists from all nations is refreshing during a time where so many problems have arisen from lack of mobility of scientists. For example, many DOE facilities such as neutron scattering at Los Alamos (LANL) have historically relied on a fraction of foreign scientists to come and use the facility and this helps pay to maintain it. Much of this income has been lost and is not being compensated from other sources. Further, many legal immegrants working within the Physics community have had very serious visa problems preventing them from leaving the country to attend foreign conferences. The APS was held in Canada this year and the rate of people who could not show up to attend and speak was perhaps ten times greater then the APS conferences I attended previously. Although moving it to Canada helped many foreign scientists attend, it prevented a great deal of foreign scientists living within the US from going. Even with a visa to live and work within the US, they were not allowed to return to the US without additional paperwork which many people had difficulty getting.

Obviously, security is heightened after 9/11, as it should be. I'm bringing up the detrimental sides to such policies not to argue no such policies should have been implemented, but to suggest the benefits be weighed against the costs - and the obvious costs such as to certain facilities should either be compensated directly or we should be honest and realize we are (indirectly) cutting funding to facilities which are (partly) used for defence in order to increase security.

I mention LANL despite it's dubious history of retaining secrets because I have heard talks by people working there (this is after 9/11) on ways to detect various WMD crossing US boarders. Even though they personally are (probably) well funded, if they facilities they need to use don't operate any more this is a huge net loss. My understanding is that all national labs (in the US) have had similar losses from lost foreign use.
____________________________________________ ___

--
a war on terrorism? How can we end a war on a method?

It has to be said... by SageMadHatter · 2004-05-12 07:47 · Score: 1

...can you image a beowulf clusters of those? :)

can't believe noone said it it.... by laugau · 2004-05-12 07:56 · Score: 1

Yeah, but does it run linux?

More cynicism is needed by Anonymous Coward · 2004-05-12 08:06 · Score: 0

There's nothing wrong with supercomputing.

What is wrong here is the incredible hyperbole used to justify what amounts to nothing more than welfare for computer scientists.

This project is *not* about doing great new science as we are told. This computer is not going to give "great advances in medicine, and improve our quality of life." *Perhaps* you can make that claim about the general field of supercomputing, but what is missing in the comments so far is the separation of this machine from the general field of supercomputing.

Quite frankly, the supercomputing research community has used the success of the Japanese machine to scare congress into funding a new supercomputer project, so that they could have a new sandbox in which to explore architectural issues.

There is no way on god's little green earth that the additional computing capabilities of this machine are going to advance science $50M worth (and count yourself lucky if that's really the final cost). Most of the top end machines are rarely used for actual calculations; they are instead research vessels for thinking about the next generation of machine. They spend a good deal of their cycles being tested for scalability, having new OS support installed and tested, running benchmarks, exploring new algorithms, etc. By the time these issues are worked out, by the time application software has been written that runs on them, they are 2-3 years old and are no longer the "top machine".

The real workhorse machines of science are always those 1-2 generations behind the machine at the top. These are the machines with stable OSs, with software implemented and optimized for them, with usable I/O, etc.

You as taxpayers are imho being ripped off by my community (I am a supercomputer scientist). It's hard to sit and watch the machinery of government be perverted and twisted by my own peers for their own personal gain, but that's what I see. Programs that have been funding big iron like ASCI are starting to dry up, and the community has been flailing for several years to find a new pill to sell to congress... A good "we might not be #1" was well timed for them.

WOO HOO by Krizhek · 2004-05-12 08:56 · Score: 0, Redundant

Finally we can have a system that can show the full potential of DOOM 3!!

NNSA vs. Office of Science by Richard+Mills · 2004-05-12 09:24 · Score: 2, Informative

"Any tin-foil hats should be directed at Y-12. That's the DOD plant; X-10 is just DOE."

You're right, but let me clarify something:

The biggest weapons labs in the country are DOE, not DOD facilities. These are the "tri-labs": Los Alamos, Lawrence Livermore, and Sandia. They are operated by the DOE's NNSA (National Nuclear Security Administration).

The other major DOE labs (including ORNL) are operated by the DOE's Office of Science. These are non-weapons labs. For you conspiracy theorists out there, its pretty obvious that these are non-weapons labs. No guys standing around with M-16's etc., as you would find at a place like Los Alamos. Much, much less security.

But... by JohnDoe.Slashed · 2004-05-12 11:00 · Score: 0

... imagine a Beowulf of these ...

Commander Data beats super computer by Anonymous Coward · 2004-05-12 11:27 · Score: 0

50 trillion calculations per second

Well. According to "The measure of the man" (Star Trek The Next Generation), commander Data is capable to 60 trillion calculations per second. :-D

50 trillion! by trillion · 2004-05-12 12:13 · Score: 1

the world always has needed more of me.

Cray and IBM by Impy+the+Impiuos+Imp · 2004-05-12 12:27 · Score: 1

Well, from that top 500 list, I'm impressed my desktop PC (3200 MHz PIV, not actually listed, 2.5 is fastest).

It smokes the earliest few dozen Crays, not to mention the IBM RS6000 series, and smokes the holy hell out of the IBM 3090 I used at U-Mich in the mid-late '80's.

Of course, if it were to process a ton of data, I dare say the I/O might make it slow down a lot from those machines...

--
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.

Re:Cray and IBM by green+pizza · 2004-05-12 20:16 · Score: 1

Well, from that top 500 list, I'm impressed my desktop PC (3200 MHz PIV, not actually listed, 2.5 is fastest).

It smokes the earliest few dozen Crays, not to mention the IBM RS6000 series, and smokes the holy hell out of the IBM 3090 I used at U-Mich in the mid-late '80's.

Too bad you don't have access to a time machine, that Pentium would be the cat's meow back in 1985!

Hmm, y'know, someone should devise a standard unit to show just how well a newer machine can smoke an older machine. Liters of Smoke perhaps?

UNICOS/mp is based on IRIX 6.5 by green+pizza · 2004-05-12 15:51 · Score: 1

Dude, it's no secret:
http://www.cray.com/craydoc/manuals/S-2346-23/html -S-2346-23/z1019077484.html

Slashdot Mirror

World's Fastest Supercomputer To Be Built At ORNL

230 comments