Linux Claims 4 of the Top 5 Supercomputer Spots

Anonymous Coward writes to tell us that the November 2005 list of supercomputers has been published. Certainly something to note is that four of the top five use linux. Relatedly Multiflow writes "CNET is reporting that the number of supercomputers on the Top500 list which use Intel Itanium 2 microprocessors has fallen by almost 50% in the past year. While new higher performance Itanium chips are in the pipeline, the article reports that 64 bit Xeons and Opterons have increased their representation on Top500."

Comments

niche market?

[OffTheLip]

It may be a niche market but what a market it is. Rock on Linux!

Re:niche market?

[Anonymous Coward]

Yes, but is it ready for the desktop?

(Hint to the OSS-monkeys: the answer is not "Of course; it has at least 157 different window managers and 476 different toolkits, each prettier than the other, so it must be ready!!!1!!")

Re:niche market?

[miffo.swe]

I use it as a desktop and find it very usable. Thats enough for me. Linux has never been about snagging market share wich is something many Windows jockeys has a hard time understanding. Linux success doesnt stand and fall with the number of users. If it stays at 10% so what? There should really be 10 different OS out there competing and 10% of that is pretty good.

Re:niche market?

[Pantero Blanco]

I'd say Ubuntu's on par with Windows XP on user-friendliness (newbies aren't confused by a seven-item list of window managers to choose from, for example), so the only issues left are compatibility (which has been improving quickly) and exposure.

Re:niche market?

[Pantero Blanco]

The problem is, it's not at 10%. It's at one or two percent of the Desktop market, which means that a corporation who holds 90% of the market can use their muscle to make using Linux difficult. It's hard enough to convince someone to make their application Linux-compatible or bundle their computer with Linux with such a small amount of marketshare, and it gets worse when someone is offering them deals NOT to do it. Ten percent would mean it was established enough that they couldn't do that, because coders and hardware manufacturers would look at it and see a sizable amount of users they could tap into.

Re:niche market?

[nottoogeeky]

Nope. I can install XP easy, but the Ubuntu cd i have here won't install and won't work out the partitions. When i installed mandrake for the first time, took me three weeks just go get my alcatel 330 working. Im an advanced computer user windows/mac......and i still cannot use linux. It's not ready for the desktop.

Re:niche market?

[Anonymous Coward]

I guess it is that bad already, somone who can copy and past with the keyboard is and "advanced computer user windows/mac" [sic].

Re:niche market?

[nottoogeeky]

More Web Design/Development, Coding PHP, MySql, HMTL, XML. Along with Dreamweaver for layout. I would say that's a little more than copying and pasting!!!

Re:niche market?

[Anonymous Coward]

...Not by much, no. Call me jaded, but I've met a lot of "web developers" who don't know nearly as much as they think they do.

Re:niche market?

[nottoogeeky]

It's people like you that slow the progress of Linux desktop with your "everyone is stupid" attitude and everyone should use the "kernal". Geeks & People are two different species im affraid!

Re:niche market?

[LaurenBC]

Ubuntu will be on par with with WindowsXP for user-friendliness when it doesnt take 4 hours to figure out how to make X display the proper monitor resolutions in Gnome (among other random frustrations). Don't get me wrong, I've been enjoying learning / configuring a Linux system, but the average computer user would have given up in the first 30 minutes and booted back up to XP.

Re:niche market?

[nottoogeeky]

damn right!!! I'm all for linux taking over xp/osx. But they have to listen to the desktop market instead of flaming them all. I like fluffy eye candy and movin things!!!

Re:niche market?

[DogDude]

There should really be 10 different OS out there competing and 10% of that is pretty good.

You're obviously too young to remember the OS wars... C64, IMB-compatible, Apple (Mac), Amiga, etc. It used to be a real nightmare to buy and even more of a pain to develop software. As someone who was a geek during those days, I can say that things in the world of PC's are MUCH better today than they were when we actually had a lot of OS competition.

Re:niche market?

[Tanktalus]

I had to move an old RH6.2 box to WinXP just a couple weeks ago. (/usr was missing - so after the last power outage, it wouldn't come back up, and the wife wanted a windows box in the house.) Oh, man, what a pain. I had less trouble getting Gentoo up and running on a fresh box! (Admittedly, the XP install took less time than Gentoo, but I watched a movie while waiting for Gentoo to finish compiling the core, and had a couple days to do other things while X/KDE compiled.)

First, the box couldn't boot off that CD. Had to install NT4 first, just to get Windows at all.

NT4 didn't recognise the network card, so I had no network. XP install wouldn't work without SP3. Back to the gentoo box, download SP3, burn to CD.

Apply SP3.

Insert XP CD again. Every once in a while, XP would stop to ask me more questions (why couldn't they ask these questions all at once?).

Eventually, got to a working system, installed more stuff, all was fine. Lost track of the number of reboots, though. Gentoo only had two: one to run from the LiveCD, and the second to run from the hard disk. (OS/2 ... now that had a pathetic number of reboots - but at least you didn't need to stick around for them after the first reboot - they asked some questions to start, then a bit later it'd reboot, and ask all the rest of the questions, and then reboot a half-dozen or so times, but no more questions.)

I'm not convinced XP is easy to install. I'm betting that if I sat my wife down on a brand new machine with a working CD (thus XP would boot off CD properly), and all the instructions at her fingertips, and I were away for the weekend (so she couldn't ask me questions ;->), she'd actually have an easier time with Gentoo. And THAT is scary.

Mind you, if I did the same to my mother, she'd just freak out without trying anything. There's no such thing as "easy enough" for her. :-)

Re:niche market?

[masklinn]

I use it as a desktop and find it very usable

Asa still disagrees

Re:niche market?

[nottoogeeky]

Get a new box? Sounds like old hardware...

Re:niche market?

[Anonymous Coward]

man this is so not true.. i've installed a few distros, and gentoo was absolutely the worst (i had to read alot just to install the base system (which imho isn't quite user-friendly) the ms installs are easy peasy.. but easiest installer (so far for me) is Ubuntus. and after that comes windows installers..

Re:niche market?

[nottoogeeky]

I've got ubuntu here, and it should be easy. but it's not. Im hoping KDE4 or even 5 will make a difference to the desktop market. If the *GEEKS* listen to the *NOOBS*

Re:niche market?

[name773]

everyone should use the kernel... otherwise their system won't run.

you need some persistence at first, but once you get good at finding howtos and following directions it gets easier.

Re:niche market?

[nottoogeeky]

That may be fine for you. but what about average joe? no chance!

Re:niche market?

[bersl2]

Ah, but most OSes (with the notable exception of you-know-who's) adhere to certain standards, such as POSIX. By standardizing what can be standardized, and by carefully abstracting, it becomes easier to develop for a wider range of OSes.

Anyway, user share only matters (for us) to the extent that we do not want to be excluded from doing something simply because we haven't enough users to be relevant. The actual number doesn't matter, only the effect that number has on consideration of our OS as a "first-class citizen".

Re:niche market?

[dreadclown]

I grant you that there was a lot of software that was only available on single platforms back in the day. However the parent is discussing operating systems and your examples are hardware platforms.

And interestingly enough, two of those platforms could run UNIX...

Re:niche market?

[budgenator]

I've got a couple of pieces of hardware, such as my DVD drive, that when in their original windows box won't work. Dell is spending a lot of money on warrentee parts that are hopelessly broken in Windows, that just work in Linux; often they tell customer to just chuck, rather than pay for returns.

Re:niche market?

[Anonymous Coward]

Average Joe isn't an advanced user, not in the old sense anyway. I guess advanced in the Windows world changed when most of the truly advanced users migrated to a more advanced system.

Re:niche market?

[Anonymous Coward]

You ever installed windows as the second OS on a computer? What a pain in the ass! Though I found SUSE to be even worse, since it didn't clearly tell me what partitions it was going to do what to.

As far as I'm concerned Slackware is still tops in ease of installation. But I'm sure all the slashbots will be like "it's not easy to install" because you actually need to understand a few things before doing it. With a lot of these other systems (I'm including windows there, but not apple since I've never installed their shit but I'm going to assume it's similar until someone demonstrates that it's not) it's just a pain in the ass to even get the system installed and up and running.

And it seems to me that being able to install easy as a second OS on a system is far more important than having the "click next" style of windows since most people who are planning a switch aren't going to just do it on a new machine.

Re:niche market?

[Anonymous Coward]

Knoppix!

Re:niche market?

[name773]

oh, i don't really care what the average joe uses. i'm just trying to say that it's not so hard once you get the hang of how to get the hang of things.

Re:niche market?

[civilizedINTENSITY]

Well I'd installed a number of windows OSes, and a few versions of several distros, but most of the people I took Operating Systems with were windows users. Everyone else thought that the install labs were going to be easy for windows and hard for linux. Wrong. Getting network printers working, and file servers, and setting up users accounts, etc..., starting with a set of CDs and an unformated hard drive was easier for almost everybody using linux than windows. These were computer science juniors and seniors, not exactly without a clue...

Re:niche market?

[civilizedINTENSITY]

You know, wouldn't it be nice if Microsoft came out with a set of man pages instead of that piece of... code... they call MS Help? It is possible to *quickly* get the necessary information to do basic things with man pages. Not that I think the format of MS Help is so bad (it looks just like gnome help, or java help) although I do think that the phrase "compiled HTML" is at least somehow contradictary. The actual ad hoc assortment of text, and dismal indexing, makes it possible to have done something 6 months ago and not know the secret MS combo of terms to search for to find it again. Rumor has it documentation is so bad for windows to support the third party book market.

Re:niche market?

[NetRAVEN5000]

"You're obviously too young to remember the OS wars... C64, IMB-compatible, Apple (Mac), Amiga, etc. It used to be a real nightmare to buy and even more of a pain to develop software. As someone who was a geek during those days, I can say that things in the world of PC's are MUCH better today than they were when we actually had a lot of OS competition."

First of all, this is not the "OS wars" so much as the machine wars - these are all different machines, even if they do use different OSes - the only reason they have different OSes is because their OSes were written specifically with that machine in mind. This is no different today - you still can't run MS Windows on a Mac.

Second of all, each of these OSes had their own strengths and weaknesses, right?

And third of all, as long as they follow standards, there'd be no problem. I have yet to hear anyone say that they have trouble switching between Opera and Firefox. In fact many of the problems we have with computers today such as vendor lock-in and version incompatibilities are partially due to *certain companies* (AKA MS) not following standards.

Re:niche market?

[dago]

That's what these cluster are used for.

They are running the 74372 variations of window makager / toolkits.

Re:niche market?

[EsbenMoseHansen]

Nope. I can install XP easy, but the Ubuntu cd i have here won't install ...

So? Gentoo installed like a breeze on my new hardware, but windows XP could not. It kept complaining that no hard disk were present, even though there were 2 perfectly find SATA disks present --- I believe SATA is the standard these days, at least I didn't get out of my way to get these disks. It then proceded to ask me for a windows driver, which I presumably had on the a CD, on a diskette, which I did not. As I don't actually own a diskette drive, I gave up after a week or so fighting this.

I have since heard of a few more people having exactly this issue. Also some that did not with a similar setup. *shrug*

I even went as far as borrow a set up service pack 2 (?) install CDs just to try if that would work.. no change.

In conclusion, I won't be playing a game one of my friends helped develop, but other than that, I don't miss windows anyway. Gentoo is much simpler to use, at least for an old mathematician like me. And now that my work laptop is also pure linux (RHEL4xx, a customized version i think) I am happy with my setup

The point is, from anecdotal evidence, windows is just as hard to install as linux. That is, sometimes it works, sometimes it doesn't. Unless someone cares to do a real statistical analysis, that is as close as we will ever get.

well duh

[scenestar]

These aren't off the shelf desktops.

What else would you expect them to run, windows ME?

Re:well duh

[Anonymous Coward]

No, Windows CEMeNT

Re:well duh

[Eightyford]

These aren't off the shelf desktops. What else would you expect them to run, windows ME?

HP-UX, IRIX, Solaris, SCO UNIX, Mac OS X, free/open/netBSD...? Palm OS?

Re:well duh

[Anonymous Coward]

Windows Vista Supercomputer Edition, duh.

Re:well duh

[vhogemann]

Well,

One migth think that Microsoft would pursue a position among the top 100. After all, they have a clustering solution, dont they?

Re:well duh

[SoloFlyer2]

That must be what they are running on their webservers... SLASHDOTTED! :)

http://www.mirrordot.org/stories/c4d983f435b904443 de279fee8e0ea48/index.html

Re:well duh

[ZakuSage]

Clearly they should be running SCO Unixware.

Re:well duh

[Jake73]

Do you have any idea how much a 1,024-processor version of Windows XP Pro costs??

Re:well duh

[khellendros1984]

I can see the headline now: PalmOS has jumped to number 3 OS in use on multiprocessor supercomputers!....Imagine a Beowulf cluster of Palm V's....

Re:well duh

[Anonymous Coward]

HA HA, SO FUNNAY!

Re:well duh

[Hugonz]

Well that's what you'd normally say. But the MS sales rep would keep telling you that they hace great clustering solutions. See, Windows Server (NT) is a general purpose operating system, just like Linux. Linux is not intended for supercomputers either..

Re:well duh

[ozbird]

What else would you expect them to run, windows ME? Windows XMP.

Re:well duh

[erikharrison]

Interestingly, none of those are the OS used in the only top 5 computer not running Linux.

It's AIX.

Re:well duh

[The_Dougster]

Linux is not intended for supercomputers either.

That's pretty subjective. If you're talking about "the Linux kernel" then while it was originally designed to just be a monolithic unix kernel clone for x86 machines, over the recent years, big boys like IBM have made some significant contributions to the source tree to make Linux an effective kernel on big iron.

If you're talking about Linux as in "GNU Userland" there's no reason it can't be run on big iron. You can probably build quite a lot of the GNU utilities against a non-Linux kernel and generic libc without serious problems. Personally, I like the GNU versions better than proprietary ones just because of the more consistent command line switch syntax and usually the extra features.

I know about zero concerning supercomputers though. Presumably you would access them using a graphics terminal of some type. Whatever the underlying OS might be, the most important thing for a supercomputer seems to be the ability to run massively parallel programs to solve huge matrix problems and otherwise process massive amounts of data quickly. No matter what OS they run, as long as you have your special FORTRAN compiler (or whatever it is they use on those things) to write super-programs, I don't see why Linux isn't at least as good a choice as any others.

Re:well duh

[borsi]

These aren't off the shelf desktops. What else would you expect them to run, windows ME?


HP-UX, IRIX, Solaris, SCO UNIX, Mac OS X, free/open/netBSD...? Palm OS?


Hmmm... OS/400, perhaps... everything for user friendliness :)

Re:well duh

[Anonymous Coward]

We're not talking about big iron. We're talking about compute clusters.

Typically, the individual nodes have 2-4 CPUs, so large SMP scalability is a minor point. They typically run exactly one thread per CPU, so the scheduler is mostly irrelevant. Most of these jobs allocate all their memory up front (in large pages, if they can find enough contiguous memory), so the memory management code is mostly irrelevant. They typically communicate using MPI backed by a custom lightweight network protocol (often running directly over infiniband), so the TCP/IP stack is mostly irrelevant.

The real reason Linux is so popular for this task is that it is easy to strip down to the bare essentials. I suspect only Microsoft (and perhaps the sysinternals guys) have sufficient windows knowledge to strip down Win2k3 sufficiently; hence Windows 2k3 Cluster Edition.

Hooray for Linux!

[Elrac]

While this will not do much to encourage the Unwashed Masses to embrace Linux, it certainly shows that Linux is a serious operating system suited to high-powered computing (or at least to hosting high-powered computing applications). I hope at least a few Fortune 500 CIOs will take notice.

Re:Hooray for Linux!

[Anonymous Coward]

I hate to break it to you, but the term "Unwashed masses" couldn't describe the Linux user base more accurately. =)

Re:Hooray for Linux!

[Decaff]

While this will not do much to encourage the Unwashed Masses to embrace Linux, it certainly shows that Linux is a serious operating system suited to high-powered computing (or at least to hosting high-powered computing applications). I hope at least a few Fortune 500 CIOs will take notice.

Actually, it doesn't show that at all. Supercomputing is a very specialised niche use of hardware. Generally, this sort of software wants the operating system to get out of the way as much as possible and allow the fastest possible access to memory and processors and (depending on the situation) I/O systems. In the past major supercomputer applications have required very little operating system functionality to back them up.

There is little comparison between specialised numerical supercomputing and general multi-processor mainframe use, which requires concurrent multiuser access to app servers, general filesystems, databases etc. This is where older OSes such as IBM operating systems and Solaris work very well, and where Linux is now making inroads.

It is rather like comparing a formula one racing car to a truck. I agree that Linux is suited to both purposes, but working well in one environment does not indicate usefulness in another.

Re:Hooray for Linux!

[Anonymous Coward]

Well, except for the "masses". Unless you mean their weight.

Re:Hooray for Linux!

[Anonymous Coward]

"Unless you mean their weight."

Astonishing deduction Mr. Holmes!

Re:Hooray for Linux!

[networkBoy]

In fact, IIRC earlier supercomputing apps did away with the OS concept entirly. The application was written for the particular target hardware and ran on the hardware directly, without the abstraction an OS provides. While this would eek out every possible drop of performance, it sure is hell for portability, which is why the current applications expect at least a minimal os to field the basic functionality.

Come to think of it, what most apps want is a souped up BIOS with some smarts and a filesystem.
-nB

Re:Hooray for Linux!

[Anonymous Coward]

Uh, sorry to burst your bubble, but mmm Linux has been 4/5 of the top 5 for (several years now), and also about 300 out of the 500 machines on the list. On the last list, there was 1 (ONE) system running a microsoft OS. According to Microsoft Research (earlier this week), WindowsXP-SP2 needs 5,376,000 cpu cycles to start a new process, Linux only needs 719,000 (roughly 2/15 as much CPU). In this instance Linux is approximately 7.5 times as efficient as XP (at least according to Microsoft Research). Apparently the silly ding-a-lings that run supercomputers for biomedical research, computational fluid dynamics, and nuclear weapons testing figured that out too. The US navy has been using Linux for a while now (Tactical nuclear sub sonar applications, Naval research office, etc), so apparently the ding-a-lings over at the US navy figured out that Linux is a good, weapons-grade bet for them too. They even help in securing it (see the Fort Knox for Linux product sponsored over at the Space and Surface Warfare command in San Diego, or Security Enhanced Linux over at the U.S. National Security Agency (NSA). In short, Linux has been dominant in any area where computing is done by people who know better for a long time (Top500 too). I welcome your new-found enthusiasm though!

Google = the world's biggest supercomputer

[thisisauniqueid]

When will Google add themselves to the list? (I know, they don't qualify... and they don't want to tell people how many beige boxes they have...)

Re:Google = the world's biggest supercomputer

[Tenareth]

They regularly publish how many beige boxes they have...

Re:Google = the world's biggest supercomputer

[Anonymous Coward]

And where can we exactly find these publishements?

Re:Google = the world's biggest supercomputer

[UnknowingFool]

I'm sure Google is up there in terms of size, but this list is about performance. Google's computers do searching which is different in terms of processing requirements than raw number crunching like simulating a nuclear explosion. Google also probably wouldn't release any specifications anyway.

Re:Google = the world's biggest supercomputer

[thisisauniqueid]

Ask a real Google employee, and they'll get edgy as soon as you ask questions like that. Also try asking them how many Ph.D.s they hire... There are things they are not supposed to talk about. The mysteriousness of their company is part of their corporate image.

Re:Google = the world's biggest supercomputer

[thisisauniqueid]

Doesn't matter, check out MapReduce: http://labs.google.com/papers/ They have a general-purpose computing framework for just about anything. Google engineers regularly submit MapReduce jobs to do all sorts of things other than search (whatever they happen to be working on at the time).

Re:Google = the world's biggest supercomputer

[Anonymous Coward]

they don't have beige boxes at all.

They have cabinets that hold ~100 motherboards feed off a huge DC power supply. The harddrives litterally sit ontop of the memory sticks and they are sandwiched together with maybe 10MM of space between each shelf.

Atleast thats what their servers in mae-west mci/uunet pop look like...

Re:Google = the world's biggest supercomputer

[rmallico]

heh... my back bedroom looks like that...

Re:Google = the world's biggest supercomputer

[Anonymous Coward]

Usually when I have a question like this, I just Google it.

Re:Google = the world's biggest supercomputer

[diegocgteleline.es]

They regularly lie about those numbers, too. They don't want to say it, it's just that. The google cluster is the real heart of google.

Re:Google = the world's biggest supercomputer

[Ironsides]

Imagine if google started running SETI@HOME, Protien Folding or Distributed.net on their spare CPU cycles.

Re:Google = the world's biggest supercomputer

[Anonymous Coward]

They do run protein folding.

Itanic, we hardly knew ye

[drinkypoo]

Itanium already has no popularity whatsoever. If it can't even be successful in the supercomputer market, it can't succeed anywhere (last I looked, itanium had truly awe-inspiring FP but was slow at everything else.)

MY HEART WILL GO ONNNNNNNN!

Re:Itanic, we hardly knew ye

[Hal_Porter]

Yup, it's an interesting architecture, but the sales are unbelievably bad

http://www.theregister.co.uk/2004/08/30/opteron_it anium_sales_q2/

      The Itanium ecosystem is as unhealthy as ever with HP totally dominating
      sales. HP moved 4,789 of the 5,665 boxes shipped in the second quarter,
      earning $250m in revenue. That total is roughly equivalent to the RISC
      server business done by IBM or Sun in one week.

Ouch. I thought it would at least manage to sell as many chips as POWER or SPARC, but it's an order of magnitude worse. The sad thing is that the people marketing it managed to kill off Alpha and Mips before this happened - both Compaq and SGI are too far along that process to revive them when Itanium finally dies. Looks like the whole world will end up on x86-64 after all.

I expect to see...

[JrbM689]

A resurgence in Intel-based machines after their push for increased performance per watt. Power efficiency is right on the list of those who plan the building of supercomputers, right aside performance per dollar.

Re:I expect to see...

[drinkypoo]

I expect to see AMD continue to make lower-power-consumption processors than intel does, especially since even the normal desktop Athlon 64 consumes about the same amount of power as a Pentium M - even before you add in the chipset consumption. You have to do that to get a good comparison, however, since Athlon 64's memory controller is onboard, and Pentium M's isn't.

Re:I expect to see...

[diegocgteleline.es]

A resurgence in Intel-based machines after their push for increased performance per watt

Yeah. Intel still has to show one non-laptop (ie: servers and desktops) CPU which is not the n 1 power-hungry CPU in its class; and other companies are also (obviously) working in making their CPUs less power hungry (see: future versions of the amd cpus, the low-power PPCs), still Intel/Apple fanboys are already sure that Intel will be the best at performance per watt numbers

It's amazing what the Jobs' intel switch has been able to do to the Intel's public "face". They're WAY behind of amd and IBM still some people will think that that will change just because the intel ceo and Jobs said it. Maybe it will be true, but from the facts I (and anyone who looks at the facts) would say that it's AMD who is going to top the supercomputers list not intel.

Re:I expect to see...

[Anonymous Coward]

333 of those systems are Intel-based, and 55 are AMD-based. These are people who know what they are doing choosing Intel over AMD.

AMD's CPU quality is way behind Intel. I've seen plenty of AMD CPUs which were either DOA or proceeded to fail after a few months. While they may be popular to play Doom 3, outside of /., AMDs really aren't suitable for a production system except possibly without extensive testing by a 3rd party.

Now watch me get modded down for this.

Re:I expect to see...

[AnotherBlackHat]

Athlon 64 consumes about the same amount of power as a Pentium M - even before you add in the chipset consumption.

I'd like to see the tech. specs for that claim.
Last I checked, the power consumption of the Pentium M was much lower than the Athlon 64.
Even with the extra chips, it's on the order of 30 watts vs. 80 watts, but a "fair" comparison isn't really possible since the M isn't a 64 bit processor.

-- Should you believe authority without question?

Re:I expect to see...

[Anonymous Coward]

You seem to misread the current trend: Amd has been a no-go for a long time - due to the lack of decent mainboards mainly. It's increasing its share. Intel has lots of systems because of the system they've already installed - not because they're more of them right now.

Facts? November 2004:
Intel 63.6%, AMD 6.2%.

November 2005:
Intel 66.6%, AMD 11%

IOW: AMD almost doubles its share, wins 4.8%; Intel wins 3%

Re:I expect to see...

[drinkypoo]

Well, I went back and looked and now I can't find what I was previously reading. Maybe it was the original Pentium-M core? When I looked, the A64 ate a few more watts than the P-M but again, there was the memory controller, and when you added in the chipset, they were about the same.

Re:I expect to see...

[masklinn]

Duh, idle Pentium-M feature a 27W TDP while A64 are all well above 60W TDP...

On the other hand, low-power Opterons and Turion64 (AMD's mobile A64s) get as low as 25W TDP.

Oh, and BTW some A64/socket939 chipsets get hot as fucks, check the running temperature of a nForce4 chip, mine currently checks 47C while my A64 3000+ runs a cool 35C (and I'm not using SLI, which makes the nF4 get even hotter)...

Re:I expect to see...

[Bloater]

> Duh, idle Pentium-M feature a 27W TDP while A64 are all well above 60W TDP...

Not to say that AMD64 are as low as Pentium-M, TDP is meaningless, it is the manufacturer's guide to system integrators what cooling to provide so that the CPU doesn't have to be throttled too much, but it does not say how much power the CPU uses. Turion64 ML CPUs have a TDP of 35W but measure (including chipsets - which is necessary to compare apples to apples) lower than Pentium M.

Re:I expect to see...

[The_Dougster]

A resurgence in Intel-based machines...

Intel sure seems like they have shot themselves in the foot. They worked for years making sure that Intel X86 was king of the hill, then they tried to switch to Itanium, and AMD blew past them in the meantime. Intel has been still selling well just because of inertia, but it looks like the market is starting to jump over to AMD now finally.

Personally, I don't care. In fact I'd rather be using something with a sensible machine programming scheme and more grokkable architecture. PowerPC, ARM and Sparc are basically the three remaining competitors to X86. Mips, PA-RISC, Alpha, and others are still around, but I don't forsee them making much of a comeback, maybe Mips.

You're probably right if you consider X86-64 to be "Intel-based." It would be cool to see a supercomputer based on something like ARM if they could add FPU's and Vector processors, but by then it probably wouldn't be an ARM anymore.

Sun and IBM probably have the best hope of making non-x86 processors which might be viable for supercomputers. Especially IBM.

Universal deployment.

[Anonymous Coward]

With all the tiny inperfections and debates about the developtment processes that are wildly talked about in the tech media it's still remarkable that a system that can do embedded to desktop to server to supercluster systems ( and all variations in between ) in all areas and still keeps on growing and growing on all fronts.
It is not a niche product, It's the only OS that has any momentum at all at this time.

Happy happy !.

Retep Vosnul.

Re:Universal deployment.

[VENONA]

And so far as I know, Linux is the only OS that runs in both the embedded market (can it still fit in, and run a wristwatch?) and has multiple Top 10 technical computing rankings. Surely that's *some* reason for newcomers to computing to see that computing != MS.

Re:Universal deployment.

[supermank17]

No, its not. Why should a newcomer care if the OS runs on a wristwatch? They want something to work well on their Desktop.

Re:Universal deployment.

[VENONA]

I just thought the article in Linux Journal from a few years back was cool, relevant to a universal deployment thread, and likely hadn't gotten much popular press recently. How did newcomers and/or desktops become the foci of the thread, anyway? Some of us aren't evangelists. I couldn't care less about getting everyone to use what I use. To each their own. And as far as those newcomers go, a case could be made that more Linux newcomers are exposed to it on the server than on the desktop. That's certainly been the case at a couple of gigs I've done. Not that there's any way of proving either case, given the myriad ways Linux gets distributed. There are a lot more ways to be exposed to Linux (or the BSDs, etc.) than via the desktop. You sound like yet another one of those people who think their environment constitutes all of computing. That's ugly enough in Windows people. BTW, all this caused me to doubt the accuracy of your statement that it wasn't small enough to fit on a wristwatch. A good thing, because I just googled up http://www.research.ibm.com/WearableComputing/linu xwatch/linuxwatch.html Last entry in the timeline there is "May 2004: IBM demonstrates Session Initiation Protocol (SIP) on the watch. Sounds as if the Linux watch project may still be getting sporadic work done on it.

Source?

[PMoonlite]

Does someone have a source that tells what OS these things run? I'm not seeing it in either article.

Re:Source?

[jbellows_20]

Well, if Sony hadn't went off and got caught they would have the largest supercomputer and it would be running Microsft Windows!

Re:Source?

[cnkeller]

In the case of Columbia, it's listed right on the systems page.

Re:Source?

[Anonymous Coward]

Click on the name of the supercomputer and then click on "System Details". The OS is listed there. Like so: http://www.top500.org/system/details/7747

Re:Source?

[NetRAVEN5000]

"Well, if Sony hadn't went off and got caught they would have the largest supercomputer and it would be running Microsft Windows!"

In order for it to be considered a supercomputer all the boxes generally have to be in the same building and owned by the same company. A large network of computers that were overtaken by a Skynet-like, we-control-your-PC-now piece of malware such as Sony's rootkit doesn't count for a "supercomputer".

Re:Source?

[Sique]

Says who?

yeah well would you want to pay for m$

[mikek3332002]

It makes perfect sense!!

With windows licenes costing about $300 for a couple of processors

With the total cost it would be more powerful to get linux and spend the left over increasing its performance.

Re:yeah well would you want to pay for m$

[ajlitt]

Of course. Because, you know, the only two operating systems that a supercomputer could possibly run are Linux and Windows.

Re:yeah well would you want to pay for m$

[jtolds]

The concern isn't whether or not these supercomputers run Microsoft software, but whether or not Linux is a worthy alternative to top-end *nixes.

Of course, it is, and that's all this shows.

Re:yeah well would you want to pay for m$

[jtolds]

I'm sorry, that comment should have been in response to two parents up.

Re:yeah well would you want to pay for m$

[jmv]

While Windows probably wouldn't make sense for many other reasons, price is probably the least of the issues. Given how expensive the interconnects are in these machines, 300$ CPU is negligeable.

Re:yeah well would you want to pay for m$

[ZakuSage]

Well yes, but they could easilly go for other free options other then Linux. FreeBSD, for example.

Re:yeah well would you want to pay for m$

[mikefe]

Well yes, but they could easilly go for other free options other then Linux. FreeBSD, for example.

Please point me to the papers showing better scaling for SMP and NUMA in FreeBSD when compared to Linux 2.6

Re:yeah well would you want to pay for m$

[NetRAVEN5000]

"Given how expensive the interconnects are in these machines, 300$ CPU is negligeable."

Given that many of these machines have over 4,000 CPUs. . . it's really not all that negligible. Saving over $1,000,000 isn't what I'd call negligible - it'd buy you quite a few extra CPUs. Or quite a few extra gigabytes of memory, or whatever else you might need/want in there. Sure, maybe there's other areas that are more costly but saving that much money by ditching MS is something to consider.

Re:yeah well would you want to pay for m$

[jmv]

Oh, I'm not saying anyone in the right mind would ever consider Windows for a supercomputer. Just that in this particular case, the money itself isn't that much of an issue. If an OS costs you 300$/CPU and improved the performance by just 10%, it would most likely be worth it. Of course, Windows would probably do the opposite, but that's another thing.

One Supercomputer?

[msbsod]

Are these individual supercomputers? BlueGene comes with 131072 processors. Is this one (1) computer? Or 131072? If this is not one computer, then what does list tell us? That 131072 processors are faster than 1 processor?

(The top 6 are all from the US - followed by Japan and Europe.)

Re:One Supercomputer?

[Wesley Felter]

The definition of "supercomputer" these days seems to be "a collection of hardware that can run an MPI job". So BlueGene/L is a cluster of 64K computers, but it counts as one supercomputer.

Re:One Supercomputer?

[Anonymous Coward]

You're free to build your "supercomputer" any way you want: one processor overclocked in the THz range, 1 million Vic-20s, or 1 billion smart hamsters.

What matters is the Linpack performance that you get out of your machine. The Linpack benchmark is parallelizable but not "embarassingly" parallelizable. Gluing two boxes together with crazy glue does not allow you to run the Linpack benchmark twice as fast.

Re:One Supercomputer?

[Kadin2048]

Fairly good question. I'm not sure where you start calling something a "computer" and where you fall off into the grey area of "computational network" or "cluster" or "grid computing system." After all, isn't SETI@Home a pretty massive computer? By some (very loose) definition it should be.

I think most people consider a computer to be something that, at some level, runs a single operating system (which then can abstract other OSes on top of itself), or perhaps is capable of addressing a single logical range of main memory (although this might not be a good definition either).

I haven't read the article yet to see if they give their definition, but it does seem as if the line between 'this is a computer' and 'this is a bunch of computers working together' is fairly blurry, and perhaps where one draws it is completely arbitrary.

Re:One Supercomputer?

[msbsod]

So all I have to do is run MPI and Linpack (or better Lapack) on a bunch of computers and that makes them altogether a supercomputer?

Re:One Supercomputer?

[imsabbel]

It is more or less arbitrary.
For example, Seti at home _could_ qualify as a supercomputer... But it would HORRIBLY suck at 99.999% of all problems because of limited node to node bandwith and latency.

Re:One Supercomputer?

[Anonymous Coward]

Depends on how fast it is. A supercomputer is anything that can achieve at least 1 GFlop/s on the Linpack benchmark.

Re:One Supercomputer?

[Wesley Felter]

As far as the Top500 is concerned, I think the answer is yes. (And this explains why SETI@HOME does not count as a supercomputer for the purpose of compiling the Top500 list.)

Re:One Supercomputer?

[Muad]

The definition of supercomputer is " a computer that costs more than $1,000,000" - you will find it in any textbook.

Re:One Supercomputer?

[imsabbel]

If the endresult is FAST, yes.
You dont need more than a few 100k Of computer equipment to get into the top500 list.

Re:One Supercomputer?

[slavemowgli]

Actually, I think there is a difference that goes beyond the fact that nobody ever ran an MPI benchmark on seti@home. It's hard to pinpoint exactly, but suffice to say that I think it's the same reason why the Internet is not a LAN, if you catch my drift.

Re:One Supercomputer?

[photon317]

The big difference is that the types of multiprocessing you can typically achieve with MPI on a tight-knit cluster is different than what you can achieve with a loose-knit cluster like seti@home. Seti@home, in terms of clusters, has very low bandwidth between nodes, and a very high interconnect latency, making certain tasks infeasibly slow on such a layout. Contrasted against a distributed setup like Set@home, the current gen of Top500 linux clusters tend to have very high bandwidth and very low latency between nodes.

But for problems which can be reasonably broken up into large chunks with low inter-node communication, obviously, the distributed approach is capable of winning by brute force if enough people throw their PCs at it.

Re:One Supercomputer?

[slavemowgli]

Oh, that definitely. But it'll still be a network of individual machines, not one supercomputer.

Re:One Supercomputer?

[NetRAVEN5000]

Seti@Home generally isn't considered a "supercomputer" because each of the client machines work independently. They aren't working together to analyze the same piece of data, each is crunching away at its own record, and the other computers go on without it if one computer isn't done yet. If you turn off the Seti screensaver mid-process the rest of the network keeps on crunching data. However, in a supercomputer the main computer waits for every computer to finish up before it goes on - so if one were to freeze or lose power the whole system would go down.

At least that's the way I understand it.

Re:One Supercomputer?

[__aanekd3853]

The definition of "supercomputer" these days seems to be "a collection of hardware that can run an MPI job". So BlueGene/L is a cluster of 64K computers, but it counts as one supercomputer.

Top500 classifies supercomputers as clusters, constellations, and MPPs (Massively Parallel Processors). BG/L is an MPP. It is obviously not a cluster since a single compute node is not a general-purpose machine. BG/L is specifically intended for massively parallel applications.

The difference between clusters and constellations is the ratio between the number of nodes (that can either work independently or be clustered) and the number of processors in a node.

Go to the Sublist generator of Top500 to see what machines belong to which category.

See, e.g., definitions here

But Virginia Tech is No. 1 in Education!!

[acercanto]

But our supercomputer, System X is the top supercomputer in academia! :-D (Man, it's already been slashdotted!) -Acercanto You can have only two of the following three qualities when developing a product: cheap, fast or good. You can produce something cheap and fast, but it won't be good, good and fast, but it won't be cheap, good and cheap, but it won't be fast.

Re:But Virginia Tech is No. 1 in Education!!

[ShadowFlyP]

What does it mean for a product to be "good"? I would say there are four qualities you're looking for in computers: cheap, fast, low-powered/low-heat, high avaliability. "Good" is a product of those four values. Likely, when designing a product you can excel at one of those qualities and perform decently at two others.

Re:But Virginia Tech is No. 1 in Education!!

[emlprime]

You seem to have left out reliable.

Re:But Virginia Tech is No. 1 in Education!!

[Anonymous Coward]

You forgot "artistic."

Re:But Virginia Tech is No. 1 in Education!!

[ShadowFlyP]

You're right I should have been more general when I said "avaliablity". What I really meant was RAS.

linux? Not exactly.

[daknapp]

Where, exactly, did you get the information that these systems "run linux?"

In the Blue/Gene system, for example, the user front-end nodes use linux, but the OS for the system itself is very definitely NOT linux. So acting as if the system runs off a linux kernel is misleading, to say the very least!

Fastest Computers?

[nnnnnnnn]

We can rule out top500.org's web server...

Re:Fastest Computers?

[Anonymous Coward]

30 comments to slashdotting!
That's got to be some kind of a record... since when do we read articles?

Linux for SuperComputers!

[Neptune0z]

I wouldn't want to be the tech guy trying to explain to the project admins that they would have to fork out 2.5 million dollars in license fee's to microsoft!... Let's also not forget that having the source code allows them to make changes to fit any particular harware or software methodology...

Re:Linux for SuperComputers!

[Anonymous Coward]

I know you guys are incapable of grasping that everyone doesn't buy software at Best Buy like you do, but if you're running a supercomputer, you a) don't pay retail for the licenses and b) get access to the source code.

Re:Linux for SuperComputers!

[budgenator]

Does source code access include fix, recompile and install? Can a mere mortal even read that mangled mess of spaghetti?

Re:Linux for SuperComputers!

[NetRAVEN5000]

"Can a mere mortal even read that mangled mess of spaghetti?"

Why would you want to read the kernel code? Maybe certain parts of it to optimize it, but not the whole thing. You'd split it up between the people on the team (if you wanted the whole thing read).

Re:Linux for SuperComputers!

[Anonymous Coward]

You do know that the basics of the UNIX operating system fits in a relatively thin book called Lions' commentary on UNIX 6th edition? Right?

One person can learn the entire theory of UNIX operating systems by reading this 254 page book. UNIX is clear, UNIX is concise, UNIX is a beautiful elegant masterwork. Much of the framework used in every modern OS is present in UNIX 6th edition.

I read and comprehended the entire book in just over a week.

Re:Linux for SuperComputers!

[budgenator]

The parent was talking about discounted windows licensing and access to the source code in Windows for large scale clusters/super-computers. My understanding is, the windows access to source code is on a "look but don't touch and don't talk" basis. Also I understand that Microsoft is putting conciderable effort into straightening out their code and bring it up to professional standards, de-spaghettizing it in other words.

Re:Linux for SuperComputers!

[NetRAVEN5000]

"One person can learn the entire theory of UNIX operating systems by reading this 254 page book. UNIX is clear, UNIX is concise, UNIX is a beautiful elegant masterwork. Much of the framework used in every modern OS is present in UNIX 6th edition."

I'm not the one who said the kernel code was a tangled mess. I'm sure it's very clear and easy to read. But I don't know many people who'd want to sit there and read all of it, unless they're either trying to fix part of it or are trying to learn from it - and I doubt they'd even read the entire thing, they'd probably just pick specific parts having to do with what they're trying to learn/fix.

IOW I wasn't saying that you can't read or can't understand the kernel code - I was just saying that I, for one, would rather read something else.

Re:Linux for SuperComputers!

[NetRAVEN5000]

Oops. It sounded to me like it was talking about the Linux kernel code - AFAIK you're not allowed to fix or recompile the Windows kernel code.

Re:Nothing unexpected

[Anonymous Coward]

Well this dosen't come as much of a suprise to me. Obviously if you're looking for the best computer you will use linux. The only reason people use Microsoft is because they are more familiar with it.

welcome to slashdot, idiot. stop trying to be cool.

Re:I see that...

[Anonymous Coward]

Apple's customers are like no others -- a rich blend of the most sociologically elite with those seeking elegant, simple computing. ... Unlike users of Intel/Windows computers, a significant portion of Apple's users are active, exploratory, avant-garde and early adopters . The activities they enjoy are unique in the the way that they more often incorporate rich media such as video and music as well as more active prosumer behavior than many more passive Windows [and Linux] users.

-- MetaFacts, Inc.

With above-average household income and education levels, the Mac population [is] very attractive [intellectually as well as physically.]

-- Nielsen/NetRatings (as quoted by C|NET)

the scoop

[SebNukem]

Looking at this chart http://www.top500.org/lists/2005/11/l/Operating_Sy stem it actually appears that the OS ran on all system are:
- Linux: 72.2%
- Max OS: 1.0%
- Others 4.4%
- UNIX and Linux: everything else (~22%)

So it appears that Linux/UNIX* runs on about 95% of all super computers. The Story headline should have been:
Linux Claims Almost All Supercomputers Spots

What a scoop.

*Linux,UNIX, what's the difference really?

Re:the scoop

[drinkypoo]

Linux,UNIX, what's the difference really?

Freedom. Especially if you specify capital-UNIX and not just Unix.

Re:the scoop

[SebNukem]

Of course. I meant Technology-wise.

Re:the scoop

[Flwyd]

MacOS, BSD, Unix... what's the difference?

Re:the scoop

[Anonymous Coward]

The diff is that shortly you wont see MacOS in the list, after all Apple offers this market nothing once they are assimilated.

Re:the scoop

[JPriest]

Linux,UNIX, what's the difference really

About 10 million line of code.

Re:the scoop

[drinkypoo]

Linux has the kitchen sink, and the others don't? It's pretty hard to find a feature in some UNIX that Linux doesn't have; dtrace is the last example I can really think of. On the other hand, you can find numerous features in Linux that aren't in other Unixes.

Re:the scoop

[Fordiman]

Well, it's not POSIX compliance, nor is it performance or stability. And considering the BSDs, it's not price, so long as you're an individual.

It ain't ease of use, as they can run the same software. It used to be binary format, but Linux likes ELF now.

Portability? You can run linux on most anything with a processor, but BSD has certain... requirements.

Re:the scoop

[MagnusDredd]

shouldn't that be:

*NIX runs on over 95% of all supercomputers?

Re:the scoop

[NetRAVEN5000]

Wonder why Windows isn't on there. . .

"Welcome to the unveiling of the unveiling of the new WTFBBQ Supercomputer! The fastest in the. . ." (whispers) 'Dammit Jim, we've got the BSOD again! Go hit Ctrl-Alt-Del!"

*yawn*

[davmoo]

I swear I'm not trying to be a troll here, but am I the only one who is not really impressed by speed figures of multi-cpu systems? All you have to do to beat one is build an identical system but add a new processor or two, and all that takes is more money, not more invention or innovation.

Re:*yawn*

[Anonymous Coward]

You need to get the infrastructure to have those additional processors operate symmetrically. Gettting the bus and OS to handle the new processor or two efficiently is the hard part.

Re:*yawn*

[FooAtWFU]

Oh, give them some credit. There's a lot of effort and design that goes into organizing such a big mess of supercomputing, keeping everything streamlined, keeping processes on thousands of different processors talking to each other, deciding what to do if one processor decides to fail, et cetera et cetera. There is real work and real innovation present- perhaps not as glamorous or even as useful as faster general-purpose microprocessor cores, but don't sell them short, either.

Re:*yawn*

[greg_barton]

All you have to do to beat one is build an identical system but add a new processor or two...

Um...that's the impressive part...

Re:*yawn*

[Anonymous Coward]

While this is true at the low end, building a large system (that is actually useful) is a pretty hard problem. The issues tend to be software dominated; any monkey can string more machines together. That said, IBM has done their homework, and getting a faster BG/L system is just a matter of cash (for everyone but LLNL at least)

Re:*yawn*

[nbert]

On the one hand it takes a lot of effort and expertise to build such systems, on the other hand both can and have to be bought usually, so I definitely see your point.

So you could argue that they could slap each other with their wallets instead of building those supercomputers. But first of all it's quite impressive that some entities are willing to invest so much money into computing power. And afterall they are not doing this because it's some kind of game - they believe that they need the processing power. Furthermore money is the dominating aspect in so many fields nowadays (especially those most people assume to be innovation driven): Cars, airplanes, watches or even sports. Just about anything related to reputation is strictly connected to a factor called money. One might argue that a Veyron is just an ordinary car powered by two upper level engines combined to a beast plus some added technology being in use since decades in other fields. What factor makes it possible to assemble such a car? Money! But still many people (even though most of them could never afford it) are fascinated by the concept. Apart from that (right here I'm not comparing cars to supercomputers anymore) they don't even question the idea of haveing a car being able to drive multiple times as fast as your average highway speed limit. They don't even care that they would run out of fuel in minutes at maximum speed. I guess that's just human nature...

Re:*yawn*

[Anonymous Coward]

Strangely enough, the top machines always seem to be "custom" hardware. Right now that would be the IBM BlueGene/L systems with 128k CPUs and 280T Rmax. Yes, the CPU inside these is PowerPC based (an embedded PPC core with the FPU from a G5 bolted on to make it fast) but this is ultimately a custom chip designed specifically for parallel computing applications.

You will not find this CPU in commodity PCs nor will you find any other machine on the list with an equal number of commodity CPUs. (The machines built with "commodity" CPUs top out around 10k or about 1/10th the number of CPUs. There are practical reasons for this: can you pump enough electricity into the building to run this thing? Can you pump enough heat out of the building to keep it from melting down?)

Before BGL, the top machine was the Earth Simulator. Again, custom CPUs / custom hardware designed specifically for the application. (For ES it was vector processors. So there it is, a "real" supercomputer just like the old Crays)

Re:*yawn*

[decepetion]

Adding "a processor or two" is not always as simple as it sounds. Can you say Scalability? Sometimes that 1 processor or 2 falls out of the realm that the compiler, interconnect, or management system has yet to work in. Just upping the size does not always work.

Re:*yawn*

[wild_berry]

The thing that I'm struck by is the efficiency of the systems: that the BlueGene/L system gets around 2GFlops/processor; 6.2 GFlops/proc for ASC/Purple; 5.1 for Columbia; 4.8 for Thunderbird; Red Storm gets 3.6; Earth Simulator, 7.7; MareNostrum, 5.8; Thunder (the highest-ranked Itanium cluster), 4.86; Korean Meteorological Administration's Cray X1E gets an outstanding 15.4GFlops/processor (with due respect to the people at Cray as was); and VTech's XServe's provide around 6GFlops/processor. I suspect that the bandwidth between nodes and the network topology to ensure that data flows smoothly is the hinderance for getting high performance per node in the larger systems.

Re:*yawn*

[raxx7]

Linpack is not very interconnect sensitive. Despite all the bragging companies make on being number 1, the top500.org are the first to recognize that Linpack is not a proper benchmark, just a quick way to compile a list of the biggest machines on the planet.

But FLOPS/CPU numbers come primarly from cpu design, clock speed and memory bandwidth.
For example, among the scalar CPUs, some can issue 2 double precision floating point ops per clock cycle, some can issue 4. And they run at different frequencies, from 700 MHz to 2.2GHz+.
The vector based machines can perform a lot of operations per cycle, but they tend to run at low clock frequencies.
For example, the Earth Simulator's processor (based on NEC SX-6) can perform 16 but it runs at 500 MHz.

still not ready for prime time

[Anonymous Coward]

that may be so... but 9 out of 10 microsoft surveys clearly point to linux not being ready for prime time.

Re:still not ready for prime time

[WindBourne]

Since they are MS surveys, is it not 11 out of 10 surveys that prefer Windows?

Slashdotted a SUPERCOMPUTER ranking site!

[pcnetworx1]

Linux > Windows

Slashdot > Linux

??? > Slashdot

Lemme reiterate, A SUPERCOMPUTER RANKING SITE WAS SLASHDOTTED!!!!! w00t!

Perhaps they should consider using one...

[Anonymous Coward]

Given that their server has been slashdotted.

Itanium quote of the day

[whitehatlurker]

This is from the second FA.

"With the benefit of 20/20 hindsight, there would not be an Itanium today," [an analyst at Illuminata] said.

ready, set,

[cyberbob2010]

que the linux/amd fanboy bashings of WINTEL!!! ;-)

Re:ready, set,

[ozmanjusri]

que the linux/amd fanboy bashings of WINTEL!!! ;-)

Por que no? Pasa el tiempo, si?

manipulation of the ignorant

[Anonymous Coward]

Perhaps its not ready for the mass market who dont know why linux is better, and would rather have 50 comercial programs (not all of the illegal spyware) all just go ahead and force themselves memory resident as soon as the desktop boots. oh thats right, linux isnt as good for the 'corporations' that want to load first on your computer and because the sole reason for purchasing your expensive hardware is simply so company X can fill your "startup folder" with their junk (if only they simply used the startup folder (perhaps after asking) instead of the rootkit/registry maul that passes for an install these days under the windows "operating" "system")

look if something as big and mainstream as "steam" _must_ load ~50meg as soon as I boot, reducing a sub min boot to a couple of min of grinding. I just no longer give a fuck about windows. Linux can be totally understood, from "Linux from scratch" right on up.. CLI rox, and with modern conveniences like intelligent autocompletion and emacs-navigation keys you can save time and be more expressive and effective than ever. oh and if you want a gui try enlightenment gnome or kde they're well established, or say enlightenment etc..

gentoo might take a while to install but you'll save time after not needing to install windows oh on about the 3rd windows reinstall you aviod. which should only take a few months.. ;)

Why no Itanic

[Groo Wanderer]

If you want a good explanation for the Itanic drop-off, look to the funding side of things. Follow the money, and all will be explained. Read a lot into this.

                  -Charlie

Re:Why no Itanic

[Anonymous Coward]

Ok, I followed the money...

And very little of it is going to Itanic! So that tells me that other processors are more popular, probably for a veriaty of reasons.

Very deep.

Re:linux? Not exactly.

[Wesley Felter]

May I suggest the following paper:

J. E. Moreira, G. Almási, C. Archer, R. Bellofatto, P. Bergner, J. R. Brunheroto, M. Brutman, J. G. Castaños, P. G. Crumley, M. Gupta, T. Inglett, D. Lieber, D. Limpert, P. McCarthy, M. Megerian, M. Mendell, M. Mundy, D. Reed, R. K. Sahoo, A. Sanomiya, R. Shok, B. Smith, and G. G. Stewart: Blue Gene/L programming and operating environment.

Summary: It's not all Linux.

Re:linux? Not exactly.

[daknapp]

In fact, here is the most relevent passage from the article:

The computational core of 65,536 compute nodes is partitioned into 1,024 logical processing sets, called psets. Each pset consists of one I/O node running Linux and 64 compute nodes running a custom compute node kernel(CNK).

Re:I see that...

[Pantero Blanco]

Why would Macs be in the top five Supercomputing slots? Or in the top 100 for that matter? That's not their game...Macs are geared to the Desktop, and only a certain portion of that market.

Re:linux? Not exactly.

[TheKubrix]

Where does it say that its definatly very NOT linux?

Re:linux? Not exactly.

[Kadin2048]

Yeah the Blue Gene running Linux thing is a bit overstated. According to this article from Newsfactor, "only 1,000 of the nodes will be powered by Linux, while the rest will use specialized software to power the machine." I expect that the 'specialized software' is z/OS or something similar, or custom code specific to the application or problem being solved.

This proves that Apple is the best !!!

[Anonymous Coward]

This proves that Apple is #1 in super computing.

Windows CE-ME-NT

[Anonymous Coward]

As hard as a rock, and as dumb as a brick.

Oh but they are

[Sycraft-fu]

Look on Slashdot's front page. Do enough refreshes and you'll see a massive ad for a book that basically proclaims you are a moron if you don't switch to Linux. All the time I see evangilism on how much better things would be if we all ran Linux. No viruses, no license fees, no crashes, etc, etc. There are a great many in the Linux community that seem to believe the ultimate Linux goal should be the domaination of the OS market. Of course non are willing to change Linux in the ways needed for this, but they state it as a goal all the same.

Some people don't care about Linux market share, but many do, including most of Slashdot and some very notable figures. I've often been berated because I won't switch, and told my reasons for refusing to do so are invalid.

Re:Oh but they are

[Grishnakh]

There's probably at least two camps of Linux users, based on their belief on whether marketshare is important or not. I'm somewhere in the middle.

On the one extreme is people who don't give a rat's ass if other people use it or not, or at least say this. On the other extreme is people who want global domination.

The problem with the first extreme is that for the whole open-source concept to work, you have to have a critical mass of users so that you have a large enough base of developers (which are a subset of the users) to keep the project working. Since most open-source software isn't owned by corporations (though some is certainly supported to some extent by them), open-source requires a large number of users to help work on the various projects.

The problem with the other extreme is that, in order to make Linux (as a group of distributions) a viable choice for all current Windows users, certain concessions and changes might have to be made, such as providing an API for closed-source drivers, removing features from the most popular software because it's "too flexible" for many users, standardizing on one desktop (GNOME or KDE), etc. Many powerful people in the open-source community don't like these things, and it's quite debatable whether they might end up hurting or destroying the open-source movement instead of helping it. (For instance, if it became easy to distribute closed-source drivers, then while Linux might become more popular initially, it might suffer from the same problem as Windows where companies release crappy drivers for their hardware, which makes Linux systems unreliable, and the companies refuse to help any open-source driver efforts).

Personally, I don't want Linux to become a commercialized, closed-source OS with a few open-source bits, but all the important stuff closed as some companies are trying to do. I also don't care if "Aunt Tillie" uses it, as long as she doesn't ask me for free support for her Windows computer, so I don't really care about it becoming the dominant OS. What I do want is for it to gain enough marketshare so that it's taken seriously, most hardware is supported on it (by open-source drivers), most worthwhile application software is ported to it, and that there's enough business in it that the dominant distros can make very polished versions without any major shortcomings like we still see today. In a nutshell, I want to be able to use Linux at work and at home to do anything I need to do (including buying and using the latest TurboTax or AutoCAD, for instance), without ever running into any major problems because I don't use Windows. If Linux reached 50% marketshare, this dream would probably be realized. The Windows users could happily live with their BSODs, activations, client-access licenses, high license prices, etc., and me and the other Linux users could happily ignore all that crap without being hindered because some web site is "optimized" for IE, TurboTax doesn't have a Linux version, ATI cards have crap drivers for Linux, etc.

Already, we're getting fairly close: certain types of hardware still have serious driver problems (video cards and WLAN adaptors), most lower-end commercial software does not have a Linux version (although much high-end software, such as that by MentorGraphics, Cadence, etc., does), and we still have serious problems with non-HTML-compliant websites. But on the plus side, we have a very reliable kernel and OS, we have very functional desktop environments (GNOME and KDE, and apps from one will work in the other), we have tons of free software to satisfy most of your needs both on the server and the desktop, and we have tons of drivers for most popular and also much older and obscure hardware. We're at the point now where you can get a recent Linux distro and install it, easily and quickly, on the hardware of your choice, and probably not run into any problems at all. You'll get tons of included software (web browsers, CD burners, word processors, etc.), and be able to do just about anything you reasonably need to do with a computer, unless perhaps your raison d'etre is to play all the latest 3D PC games.

Re:Oh but they are

[Anonymous Coward]

Look on Slashdot's front page. Do enough refreshes and you'll see a massive ad for a book that basically proclaims you are a moron if you don't switch to Linux.

<Insert comment about stupid Microsoft, Borland, etc ad about new wiz-bang product being the second coming of Jesus>

Ie, there's books and ads for just about anything.

All the time I see evangilism on how much better things would be if we all ran Linux. No viruses, no license fees, no crashes, etc, etc.

Well, most evangilism is about Linux now. Yes, there are some idiots who believe there would be no viruses (history would disagree) or no crashes (there's crashes now, though thankfully having open source means it's more likely that once can find the cause if one's willing to do some work).

There are a great many in the Linux community that seem to believe the ultimate Linux goal should be the domaination of the OS market.

Cue the moron comment again. Fanatics tend to be like that. Just look at Apple fanatics.

Of course non[e] are willing to change Linux in the ways needed for this, but they state it as a goal all the same.

Well, some probably are actually. But you're right that the majority of fanatics aren't interested in changing Linux to meet the demands of the general public. To a certain extent I'm happy with that, since some of the demands of the general public are stupid (mostly, the removal of various security features to make it "easier"). Of course, working to solve those shortcomings that makes sense would make sense.

Some people don't care about Linux market share, but many do, including most of Slashdot and some very notable figures.

I care about Linux market share. I wouldn't try to claim to know what most (which I take to mean, a majority of) Slashdoters think. Now, yes, many (as in, a sizable number or percentage, but not necessarily a majority) Slashdoters care about Linux's market share (I'm going on the fact that I've seen several thousand comments by various people). But many care primarily because of the side effects of a Linux market share. That doesn't necessitate that they're out there evangelizing. It's hard to say, though, that they don't have something of a vested interest in insuring that there isn't a single company using their monopoly to interfere with the development, deployment, or sale of competing goods. So, again, I care about Linux market share. That doesn't mean I, or others, am an idiot who would try to push everyone towards it.

I've often been berated because I won't switch, and told my reasons for refusing to do so are invalid.

Perhaps you should try to ignore such people, just as you have to ignore any fanatic. It's virtually impossible to argue with a fanatic. It's best to simply drop the subject. Just like it's necessary at times to drop the subject about what happens to a person after they die, since some people don't like the "I don't know" as a complete answer.

Re:Oh but they are

[Sycraft-fu]

I'm just pointing out that there is a big "Linux on everything" movement to the grandparent. He has a misperception that I've seen before that Windows people are all about marketshare and Linux people are all about the OS. In truth there's both kinds on both sides.

However you can't dismiss the evangilists for Linux if for no other reason than that they work hard to be the voice for Linux most people hear and to push their viewpoint. Notice that my orignal post got modded offtopic and flamebait. It's in no way offtopic, it's a direct response to the post above it. It's not that the first post was on topic and mine isn't, it's that the Linux evangilists, of which there are many on this site, like what he had to say but not what I have to say.

At any rate I don't have a problem either way, if Linux does make a move for market dominance I think it'll have to change in ways that fix most of the problems I have with it, and the lack of software availability problem will be self solving. If Linux remains small, I'll use it when convenient and not otherwise. I just want to try and clear up the misconception that Linux users don't care about marketshare. SOME don't, some care about it more than almost anything.

Re:Oh but they are

[Kadin2048]

Although I don't necessarily agree with them per se (the folks who think that Linux should be the One True OS for Everyone), I'm not sure that you can convince me that the reverse of the current situation regarding Windows and Linux would be worse or even as bad as it is now. The inherently fragmented, decentralized nature of Linux wouldn't allow it to be leveraged by one company in the same way that Microsoft uses Windows. If some company decided to start charging a few hundred bucks a processor for their Linux-based OS, plus a few hundred more for "Linux Terminal Services," plus a few more for the client system's OS, it would only take one reasonably deep-pocketed client one time to say 'hell with this' and fork the code. Perhaps over time, some company could add enough clean-room-developed pieces of proprietary code (or one totally proprietary 'killer app') to get people locked in to their system, but I think it would take a new generation of users to not be able to see through that play.

What I do agree with is that rule by Apple might concievably be almost as bad as rule by Microsoft (and given that I'm writing this on a Power Mac which I'm very fond of, it pains me to say it), I'm not sure that Linux has enough of a strong governing body to exert what power it might gain if it ever became dominant.

That said, it's probably not going to be a situation we'll have to deal with, at least in the forseeably near future, so the whole argument does fall a bit deeply into the 'academic wanking' category.

Re:Oh but they are

[VENONA]

Mod parent up. In the days when humor posts can be +5, why can't an essentially sane and common viewpoint?

Re:Oh but they are

[Anonymous Coward]

Academic Wanking...

OH, OH, You're a bad OS!

Re:Oh but they are

[strikethree]

removing features from the most popular software because it's "too flexible" for many users

i apologize but i just HAVE to respond to this...

why would you have to remove features? you will limit the power users and then the power users (aka programmers who may, or do, contribute to open source projects) will have no reason or desire to use the software! if the newbs (sorry for the pejorative) can't handle the wide feature set, partition off the most used features to an area that newbs are likely to find them... but do not EVER reduce the feature set. i just won't use it *cough* gnome *cough*

strike

Re:Oh but they are

[civilizedINTENSITY]

"He has a misperception that I've seen before that Windows people are all about marketshare and Linux people are all about the OS. In truth there's both kinds on both sides."

Agreed. But also ask yourself: "Of the people using Microsoft, what percentage got an 'A' in Operating Systems?" It is a two edged sword, you see. While Linux is seeing some adoption, it is still primarily used on servers, and by geeks who either run those servers or whose "other computer" is made by someone who made 64bit CPUs when Intel was limping towards '486.

Myself, I have to admit that gaining marketshare would mean more clout with peripheral drivers. However, I don't have to like seeing good systems dumbed down so that they resemble (what I consider to be) old and ugly designs. If Linux could shake a magic wand and split the desktop market with MS, but had to become more like windows to do it, I'd hate it.

Re:Oh but they are

[civilizedINTENSITY]

Corporate ideology is differenet enough it appears alien to mere citizens. However, there are certainly differences between companies, and Microsoft has been abusive for as long as they've had the power to be so. To suggest that any company that ascended would behave like Microsoft is to suggest that anyone elected president would be impeached.

Re:Oh but they are

[Grishnakh]

Yep, I feel the exact same way. I want software with the widest feature set possible (as long as the features actually make sense), and don't want to be limited because newbs are too confused.

But some other people are saying that this must be done to increase Linux's popularity among non-programmers, so I just hope that they're wrong. Unfortunately, as far as GNOME goes, it seems that it's somehow getting more uptake by the large distros than KDE, and it seems to go for the "less features" philosophy.

Re:Oh but they are

[Sycraft-fu]

Adn that's the thing is it really does have to become more Windows like to ever have a real chance of becomming huge. Firt off binary compatibility on the driver level is a must. You are just never going to get all companies to sign on for OSS drivers, so you have to cater to them. However along those lines you have to standardise other things. There has to be a graphics server, a window manager, a graphics tool kit and so on. While there could still be others that are optional installs, there'd have to be a standard that was defined as required as part of Linux. People need consistency in UI and operation if they are going to go for it. Likewise installs would need to be simplified. No asking which one of 4 browsers you want, even in advanced mode. Tell people how it's going to be, let them change it later if they can figure out how. Most people want direction, not choice.

Now the whole of Linux wouldn't have to go this way, one particular distro could, and it could seperate itself from Linux as a whole. The problem would be, that if it really got popular, it's standards would become defacto standards all other distros woudl ahve to comply with, or shit wouldn't run on them.

That's what many Linux people don't want to come to terms with. They want Linux to be huge, but as a geek OS. Well in essence that's asking the world to become geeks and we know that's not going to happen. So you kinda have to pick which you want more: The geek Linux or the dominant Linux. Personally I think the geek Linux is the way to go. Microsoft really does make a fine user OS. So forget that market, let Apple try for it if they want, make a geek OS as good as you can and just deal with the limitations.

I've got no problem with that, and no problem with those that want to change Linux to take over MS. I have a problem with those that think that everyone should fit in to the Linux mold and if you don't fuck you and the horse you rode in on, you shouldn't have a computer.

Re:Oh but they are

[somersault]

"Microsoft really does make a fine user OS." err.. I'm not a maaaassive linux fanboy as I use Windows at home and work, and have only occasionally flirted with Linux.. but even though Windows is more dumbed down than Linux, there are still plenty of things in it that your average user wont just understand without you teaching them about it first. Most people, if Linux was the only system in the world, would be able to learn to use Linux fine. It's just that M$ is so pervasive that users are so used to it, and we all know that it's one of the worst, yet best marketed OSs ever (and I can say that truthfully without actually thinking of a system that is worse, having used AmigaOS and MacOS a lot in my younger days). I think that most normal people are becoming 'geekier' all the time anyway, eg learning to fileshare to swap mp3s etc, everyone uses email/the internet.. there must be a way to get a better option out there for people without necessarily losing functionality etc. Windows is slooooooowly becoming more reliable and usable anyway, and more OSS is being converted to it, so at least people can start to migrate to things like OpenOffice as a start to getting into Open Source. If people get used to things like FireFox, the GIMP and OO.org , then they're going to start finding Linux more familiar without even thinking about it. I know that's a small example, but the more stuff that is converted over, the easier it will be for someone to switch over to Linux in the end (and the same goes for porting good professional apps, such as CAD apps, over to Linux).

Re:Oh but they are

[civilizedINTENSITY]

Well no, you don't have to cater to them. You just don't use shitty equipment, and part of defining "shitty" is whether it will run for me. No big deal.

We are talking at cross purposes in terms of GUIs, because it is the inflexability in MS Windows that I hate. I can't stand using it for more than a couple days before I feel like I never want to see their layout again. You can't make more than insignificant changes without paying for third party extentions, and I fear Windows is too britle for layering very much on top of it, anyways.

A happy medium would be the development of a "Linux for Dummies" GUI that Windows Power Users could learn to use. Just don't expect to see anyone switching over to it except for said Windows Power Users.

And yet none

[Trogre]

could withstand the mighty power of a Slashdotting!

ERROR
The requested URL could not be retrieved

While trying to retrieve the URL: http://www.top500.org/lists/2005/11/basic

The following error was encountered:
        * Connection Failed

The system returned:
        (60) Connection timed out

The remote host or network may be down. Please try the request again.

Re:I see that...

[Anonymous Coward]

Two of the top 20 are Mac clusters running Mac OS X.

But...

[dcapel]

Does it run....

D'oh.

Image a Beowolf cluster...

D'oh.

Re:But...

[sydneyfong]

No, really.

Imagine a Beowolf cluster of these...

Re:MS anything? any where?

[Anonymous Coward]

How did this post get modded up when it doesn't contain any hint of a coherent thought?

Seriously, I'm at a loss as to what the poster was even trying to say. Perhaps one of the people who modded it up could help me overcome the problem I have with not being able to read gibberish?

In case you were wondering...

[vectorian798]

...the one in the top 5 that is not running Linux is ASCI Purple, and it is running AIX. In case you haven't heard of it, AIX is a version of Unix developed by IBM:

IBM AIX 5L
Wikipedia: AIX Operating System

Lennon

[Anonymous Coward]

Imagine... a Beowulf cluster of these... ;).

Re:I see that...

[Anonymous Coward]

Holy Christ, I hate Mac users more than ever now. And I don't even believe in Jesus!!!

The fifth supercomputer was an Amiga

[wildzeke]

http://en.wikipedia.org/wiki/Amiga

Go us

[ajkst1]

Am I the only one looking at the top 5 going "USA! USA! USA! USA!" with the Hulk Hogan theme song "I am a real American!" playing in my head? Probably. Think the Japanese aren't planning a new super-duper computer that will accurately predict the precise location Godzilla will destroy?

Re:Go us

[Anonymous Coward]

You need all that power for the ever-increasing number of security checks being done on foreign travellers.

Re:Go us

[Polly_Morf]

Am I the only one looking at the top 5 going "USA! USA! USA! USA!" with the Hulk Hogan theme song "I am a real American!" playing in my head?

Not at this precise moment, but when i see a fat person eating at McDonalds. Oh, and a bulletwound. The american way is the model for all of us

Japanese SUPER HYPER MEGA EFFICIENT ENGINEERING

[garrett714]

I find it funny that the US's challenger to the Earth Simulator came out 3 years later, used almost twice as many processors, and only has a slight performance advantage.

6) Sandia National Laboratories
United States Red Storm Cray XT3, 2.0 GHz
Cray Inc. #Processors: 10880 Year: 2005 Rmax: 36190 Rpeak: 43520

7) The Earth Simulator Center
Japan Earth-Simulator
NEC #Processors: 5120 Year: 2002 Rmax: 35860 Rpeak: 40960

Re:Japanese SUPER HYPER MEGA EFFICIENT ENGINEERING

[Wesley Felter]

Be sure to include price, power consumption, and floor space in your engineering comparison. Vector processors are fast, but if you can only amortize the design cost over 5,120 units instead of 50M they get pretty expensive.

Re:Japanese SUPER HYPER MEGA EFFICIENT ENGINEERING

[Anonymous Coward]

Isn't it funny also when you compare the prices of the clusters?

The Earth Simulator was a very impressive feat for the time. It impressed the hell out of everyone and made the US computer manufacturers shit their pants.

Anyways, the Japanese supercomputer has been beat for a year or so now. Blue Gene did it and did it with gusto. The reason you see '2005' in the date is because it's been continually updated since it was introduced.

The odd thing about BlueGene is that it's a completely revolutionary design. It's basicly nothing but CPUS on their onn little boards with interconnects connecting to other cpus in logical three dimensional torus layout. Thats pretty much what it is.

It has a simple batch control system running the majority of the cluster backed up by Linux running on part of it which, I suppose, takes care of I/O of the total cluster, administration interfaces, and job control systems.

The other freaky thing is the vast numbers of CPUs, which is over one hundred thrity one THOUSAND proccessors... plus their speed which is 700mhz.

IBM basicly kicked the japanese supercomputer technology to floor and stomped all over it multiple times. It would take almost 9 'Earth Simulators' to equal about the same gigaflop performance as the BlueGene\L supercomputer.

Re:Japanese SUPER HYPER MEGA EFFICIENT ENGINEERING

[Shinobi]

And then you throw a somewhat complex and not-so-parallell problem on the Blue Gene, and practical efficiency drops down to somewhere around 15%.

Linux Claims 4 of the Top 5 Supercomputer Spots

[Anonymous Coward]

Well done linux, i yearn for the day when the headlines read

Female Linux Users Claim 4 of the Top 5 Supermodel Finalists

There must be some mistake...

[EnderWiggin99]

That link for AIX points to IBM's website.

Try here instead.

I hate it when you long-haired smellies pass off false information for truth.

Sincerely,
Anonymous

Yawn, part 2:

[imsabbel]

Following your argument:
Oh, vector cpus are SO boring. You just put some more execution units on the cpu, add some more memory channels. Just takes die-space and board layers, thus money, not invention or innovation.

Windows Compute Cluster Server 2003

[PCM2]

You jest, but Microsoft actually does have a horse in this race.

...and the fifth one?

[HermanAB]

Looking further, it seems that the top ten all run some flavour of UNIX and seven of them run Linux.

Less duh, more history

[fm6]

The alternative is Unix, which is what most supercomputers used to run. Or more precisely, they ran proprietary OSs that had started out as ports of Unix to their particular hardware. Then in the late 90s everybody realized that they couldn't afford to keep developing their own processors, and started shifting to commodity processors, such as Itanium. Rather than go to the expense of porting their own OSs to the new processors, they just adopted Linux. A commodity OS for a commodity processor, if you will.

I was working at SGI in 1999 when they made their Itanium/Linux move. A lot of customers (and employees for that matter) would have liked SGI to port its version of Unix, Irix, to the Itanium. But that was just too expensive. Instead, SGI promised to continue selling the MIPS/Irix Origin line, in addition to the Linux/Itanium Altix line. So Irix is still alive — as a legacy system. If you check the Top 500 list you'll find several Altix systems but not a single Origin system.

Re:Less duh, more history

[turne10]

I was working at SGI in 1999 when they made their Itanium/Linux move. A lot of customers (and employees for that matter) would have liked SGI to port its version of Unix, Irix, to the Itanium. But that was just too expensive.

uh, maybe they didn't port Irix to the Itanium, but if they didn't it wasn't because it was too expensive, because Irix was definitely ported to their PIII-based Visual Workstation line (http://en.wikipedia.org/wiki/SGI_Visual_Workstati on)

SGI never marketed/sold Visual Workstations running Irix due to fears that they'd cannibalize the sale of Cocktanes

Re:Less duh, more history

[fm6]

...Irix was definitely ported to their PIII-based Visual Workstation line

And you know this is true because....?

Re:*yawn* (Ahem...)

[Frumious Wombat]

If you look up how the top500 benchmark, and most of the others, slapping together a heap of boxes doesn't get you anything. To actually get a decent score on parallel DP linpack, or simulation codes used as benchmarks, you need a fast, very low latency interconnect between the nodes, excellent synchronization, and fast disk access.

Even the allegedly "off the shelf" systems contain an awful lot of not off the shelf hardware. Case in point would be PNNL's Itanium cluster http://www.emsl.pnl.gov/capabs/mscf.shtml/ (at 1000 or so nodes). At SC2003 I chatted with people I know from there, and they mentioned that they had four (4) Quadrics http://www.quadrics.com/ interconnect cards Per Node, plus extra switches, in order to get the bandwidth up high enough. Even a cheap cluster will add Myrinet (at about $1500/node when the switch is factored in), and start worrying about topology after the first few dozen nodes are installed.

There are clusters (basically networks of workstations), and then there are supercomputers.

Not nearly that simple ... at all

[ToasterofDOOM]

Even if it was that simple, it is incredibly tough to create applications that can harness that power, especially since they work mostly from the ground up with no high level stuff to help them along. Can you imagine allocating memory and threads across 65,536 processors and who know how many gigabytes (terabytes?) of RAM? If so, can I have your autograph?

How is this Score +5?

[Anonymous Coward]

"I'm not sure"
"I think"
"I haven't read the article yet"

This guy has no information to contribute. It's just uninformed ass licking.

Reason why Linux is used

[Ma3oxuct]

Linux is on 4 of the top 5 supercomputers is due to the fact that it is opensource, and can be modified to fit unusuall hardware. I cannot imagine the process one would have to go through to get changes to be made to the way a closed source OS will function on more than 1000 CPUs; all I know is that it is an inconvientient one. Opensource wins out because of the tremendous flexabilty it has to change rapidly.

Processors sorted by efficiency

[Anonymous Coward]

Here are the top 500 systems sorted by efficiency:

NEC Vector: 89.9%
Cray Vector: 84.0%
Hitachi Vector: 82.2%
Itanium: 80.8%
AMD x86-64: 74.0%
POWER: 72.8%
Alpha: 71.2%
Intel x86-64: 60.2%
PA-RISC: 54.0%
Intel x86: 53.2%
and lucky last,
SPARC: 49.6%

Interesting things to note:

1) After all these years, vector machines are still on top
2) "The best of the rest" are actually Itanium (IA-64) machines
3) Efficiency is important, because for real-world applications (weather forecasting, earthquake simulation or whatever) you are likely to see even lower efficiencies. If a machine can't even run linpack (dense linear algebra) well, it might really hit a brick wall when it comes time to solving more difficult problems.
4) These stats were made by visiting the top500's "stats" page, dividing the aggregate Rmax/Rpeak for each processor family, and then sorting them.

Re:Processors sorted by efficiency

[ejona]

The most surprising is the Hitachi at 37. That only has 80 processors! That is a performance of 111 for Rmax/Processors. The top ones are only around 2 to 6 for such a stat. The next fewest number of processors used is 576 at 36 which has 15 for Rmax/proc.

Re:I see that...

[Anonymous Coward]

It would be interesting to see the cost/performance ratio of the Mac clusters compared to the others in the top 20.

Re:I see that...

[crashelite]

the only one who used macs had 1100 dual processor macs (desktop models not rack mount) and is still in top 20... and they did it by their self... unlike all the others who had the vedors do it for them hints the SELF MADE part... and who would run windows on all those computers... can u imagine the # of licenses you would need DAMN...

Re:linux? Not exactly.

[tantrum]

much in the same way as my computer that has a dedicated videocard that does not run linux.. I still say that my computer is running linux though. (I might even say the same about my window box, but then I'll get modded down)

Re:I see that...

[Anonymous Coward]

Weird... I think I only partly "got it" Must be something wrong with me as I actually find some of these photos (or people in them or the products) strangely attractive.

I will admit that the girl who sliced her arms up (presumably) with razor blades might have gone too far. And the guy with the Newton logo tattooed on his arm, well, he probably went too far in a different sort of way.

And maybe I need to take my head out of the gutter or something but damn! The asian girl with the Powerbook is hot:

http://dogcow.atspace.com/473a51.html

I find something attractive about the girl with the "X" dog tag too:

http://dogcow.atspace.com/IMG_0730.html

But what do I know? I must admit that I have never been particularly attracted to those things that popular culture would have us believe to be attractive (or maybe "pulp culture" as Thomas Dolby put it.)

I actually do own a Powerbook. I don't consider myself socially elite. (Or any kind of elite really, maybe 1337 68k asm c0der back in the day elite. That part of me certainly hasn't gone away. I just design embedded computer hardware these days. And I have my own colorful fractal artwork for the desktop background on said Powerbook and switched the shell background to transparent blue and occasionally prototype DSP code in PERL and express my questionable sense of humor with silly messages hidden on the silkscreen under chip packages on prototype hardware. Is that cool? ;)

Re:I see that...

[Anonymous Coward]

There was quite a bit said about this when VT built system X. They choose the G5 based on price. (And price goes beyond just the cost of a box. If it takes twice as many CPUs to get to a certain level of performance the hidden cost is the extra rack space, extra electrical power, and extra cooling required to attain that level.) If you do a bit of math you find that the PPC CPUs do pretty well. Near the top (no 3) you have a Power5 based system delivering 6.2GF/CPU. Just below that you have what is presumably an Itanium system delivering 5.1GF/CPU. And below that you have less impressive performance from the Xeon at 4.8GF/CPU. (And for the "clock speed isn't everything" folks, note that those Xeons clock at 3.6Ghz where the Itanium is 1.5Ghz and Power5 is 1.9Ghz.) The Apple Xserve clusters (PPC970/G5) do pretty well at 5.3GF/CPU (at 2Ghz) and 5.7GF/CPU (at 2.3Ghz) On the surface it would appear that the BlueGene systems suck by this metric. (About 2.1GF/CPU but these are custom CPUs run at 500Mhz. Again "big picture" these CPUs consume a lot less electrical power and require much less physical space so you can pack a ton of these into the space and energy budget of a machine built with "commodity" CPUs. And the result? Number 1 and 2 spots.) Earth Simulator is still king in some sense at 7GF/CPU!

they did

[r00t]

Google runs on whatever is cheapest. Today Google is probably big enough to tell a motherboard designer what to design.

About 5 years ago, Google ran on about 8000 beige boxes. Each one had a Celeron CPU and two IDE drives.

OS breakdown

[Anonymous Coward]

• _______________Count____Share_%
• Linux__________361______72.20_%
• SuSE_Linux_ES8___1_______0.20_%
• Redhat_ES3_______4_______0.80_%
• HP_Unix_(HP-UX)_31_______6.20_%
• MacOS_X__________5_______1.00_%
• Solaris__________4_______0.80_%
• UNICOS__________14_______2.80_%
• Super-UX_________4_______0.80_%
• HI-UX/MPP________2_______0.40_%
• AIX_____________44_______8.80_%
• Tru64_UNIX_______4_______0.80_%
• Paragon_OS_______1_______0.20_%
• SuSE_Linux_ES9___5_______1.00_%
• UNICOS/Linux_____1_______0.20_%
• CNK/Linux_______18_______3.60_%
• Redhat_Linux_____1_______0.20_%
• Totals______500________100_%

honk if you hate the lameness filter!

watch out for bill

[torrents]

linux better watch out
http://news.com.com/Gates+to+tout+Microsofts+super computing+move/2100-1016_3-5951855.html?part=rss&t ag=5951855&subj=news

Re:watch out for bill

[Anonymous Coward]

it will propogate to a super worm and it will crash the internet

Why was this moderated off-topic?

[Ellen Spertus]

The article was about blog software, of which Drupal is an example. Why was my post moderated as being off-topic? FYI, I don't get a kickback from my hosting provider for mentioning their name.

Re:Why was this moderated off-topic?

[shish]

The article was about blog software

Pass me some of what you're smoking, kplzthx.

"Linux Claims 4 of the Top 5 Supercomputer Spots... Anonymous Coward writes to tell us that the November 2005 list of supercomputers has been published. ..."

Re:I see that...

[civilizedINTENSITY]

Yeah, I noticed the "X" dog tag too :-)

Only thing is, most of the Mac users I know are more like, named Raven, or something, ya know? and are just so totally the "elitist asshole."

;-)

As a guy who currently works for SGI....

[Anonymous Coward]

I can assure you, we didn't do it because it was indeed too expensive.

I'm also pretty sure there's no Irix on a PIII, but hey, there's a lot of odd looking stuff in the basement, so who knows.

Re:I see that...

[FastDownload]

ah, did you see #15? 15 COLSA United States MACH5 - Apple XServe, 2.0 GHz, Myrinet

Most bang per buck

[Danta]

No, not in the top 5 any longer. But those Mac clusters that are in the top 500 give by far the most bang for the the buck. See for yourself on this list. You want to look at the 'R-max MFlop / dollar' column. Because of this fact I personally don't understand why there aren't much more Mac clusters being built.

Re:I see that...

[Danta]

Has been done, though not updated with the new list. Take a look here.

Re:*yawn* (Ahem...)

[Anonymous Coward]

That info was very true a few years ago; but is rapidly becoming out of date.

These days about half the supercomputers simply use Gigabit Ethernet top500 site. And yes, that's commodity. Dual Gig Ethernet ports can be found on even $700 off the shelf servers these days.

Worrying abou topology is extremely importing. Spending your money on $1500/node interconnects probably isn't. The highest-performance guys certainly don't use such mid-range interconnects as you described (that perform about like gigE), but reather build their own proprietary interconnects (the dark blue on the chart I linked).

Linux is a computer now???

[master_p]

what else? Linux the toaster??? Linux the car???

Re:Linux is a computer now???

[a24061]

what else? Linux the toaster??? Linux the car???

I think you mean "NetBSD the toaster".

Re:I see that...

[Anonymous Coward]

I am somehow reminded of the Quentin Tarantino line in Pulp Fiction (best movie ever) "Don't Jimmy me Jules. I don't need you to tell me how fucking good my coffee is, okay? I'm the one who buys it! I know how good it is."

Re:I see that...

[Anonymous Coward]

What about

#15 MACH5 at COLSA http://www.top500.org/system/7741
#85 Turing at University of Illinois http://www.top500.org/system/7366
#308 Xseed at Bowie State University http://www.top500.org/system/7500

Re:linux? Not exactly.

[__aanekd3853]

The compute nodes of Blue Gene/L do not run anything that may be called an OS. They basically run a single application thread per processor, and they do not do any sophisticated system work at all (no context switches, the single user process has access to all the memory, etc). The system tasks are concentrated in the so-called I/O-nodes, and those run Linux. So all the system-related things there are Linux indeed. See this paper, for instance.

Note that I/O nodes and not "front-end" nodes. All the front-end machines (there are many) run Linux as well.

All the user-level stuff (the programming model, tools, compilers, etc) is standard Linux, too.

So, is it Linux?

[Disclaimer: I have worked on some system aspects of the beast, but this post is not sanctioned by BG/L team or IBM or LLNL. I am not disclosing anything proprietary here - all this is open info that can be found in many papers on the subject. Check out IBM Journal of R&D for a wealth of information.

Re:linux? Not exactly.

[WillerZ]

It's not z/OS, it's a custom kernel which provides the absolute minimum facilities required.

z/OS doesn't even run on PPC-compatible hardware...

IBM != Linux

[Anonymous Coward]

4 of the top 5 use Linux? I think you mean "The Top 3 are from the same vendor, one that is known for supporting Linux". This is more a victory for IBM. Linux is just along for the ride.

Apple-built supercomputer at Virginia Tech

[GPS Pilot]

Virginia Tech's cluster of G5 Macs used to be in the Top 5. If lowest cost per teraflop is your figure of merit, Virgina Tech's rig ranked #1 at the time... that may still be true, not sure.

Currently, G5 Mac clusters hold the #15 and #20 spots on the Top 500 list.

Funny how you say MY HEART WILL GO ONNNNNNNN!

[Anonymous Coward]

The moment when Saline Dione sung that phrase in her musick veedio, Leonardo DiCrapio was:

Getting the Captain's Log shucked by the beaver,
Governing the transaction at the cargo bay,
Stuffing the Turkey with anglo-food cake,
Exercising drawing rights on the well,
Feeding the pussycat some cock-jerky,
Spongebobbing her squidward,
NUMA NUMA in her puma,

It is safe to ascertain that EVERYONE that boards the iTanic, will be fucked in the ears by Saline Dione and in the rears by Intel.

Re:MS anything? any where?

[hhawk]

was trying to say, do any super computers run ANY bit of software from MS?

Re:MOD THE TROLL DOWN!!!

[Anonymous Coward]

I've said this before: what troll???

--
Trolling all trolls since 2001.
Posting one week late so I don't get AC-banned from negative moderation.