Slashdot Mirror

Even more damning for Sun is this:
http://www.top500.org/stats/list/30/os

Even excluding Hybird systems, Linux = 415, Solaris = 2.

Also given that 1 of the systems is SPARC, I would say that there is only a single Solaris x86/x64 system.

Is the era of massive vertical systems over?

The list that proves you wrong is right here

Both pie charts have the same date, November 2007.

The list that proves you wrong is right here

Both pie charts have the same date, November 2007.

The list that proves you wrong is right here

Now go back to the kid's table and play with your toys. The grownups are talking important business. We know you're enthusiastic about today's fad but we don't care. We have work to do and that means using tools that don't have the lifespan of a McDonald's Happy Meal toy.

The list that proves you wrong is right here

Now go back to the kid's table and play with your toys. The grownups are talking important business. We know you're enthusiastic about today's fad but we don't care. We have work to do and that means using tools that don't have the lifespan of a McDonald's Happy Meal toy.

HPL, the code that is run for the Top500 number (or an equivalent implementation, optimized for the target architecture), is double precision. This is perhaps the most important rule of the benchmark: the calculation must be carried out in full double precision. The 1.026 Pflop/s number does represent a double-precision workload.

If one looks at http://www.top500.org/ list and compare the CPU frequencies of the top supercomputers - all BlueGene CPUs were running at less than a GHz. And it seemed those low power cores were key to HPC (high performance computing). Cell and opteron - both run at multiple GHz and (presumably consume more power). IBM still has next generation of BlueGene/Q in works and is also for +Petaflop computation.

That's what I wanted to say, was looking for a comment if someone had already mentioned it.

According to "Ny Teknik" or whatever page I found it's made up of a cluster of 2128 blade servers from HP.

Theoretical max 182 Tflops, seems like it made second place when compared to the june list / when it was done / news out.

It's number fifth on november 2007 list:
http://www.top500.org/lists/2007/11

System in question:
http://www.top500.org/system/8819

That's what I wanted to say, was looking for a comment if someone had already mentioned it.

According to "Ny Teknik" or whatever page I found it's made up of a cluster of 2128 blade servers from HP.

Theoretical max 182 Tflops, seems like it made second place when compared to the june list / when it was done / news out.

It's number fifth on november 2007 list:
http://www.top500.org/lists/2007/11

System in question:
http://www.top500.org/system/8819

• ...
• we do know that they have the world's fifth most powerful computer, in competition mostly with nuclear physics labs.
• "Customers" that will be able to place requests for searches include all authorities (all some 500 of them including Department of Transportation, Department of Agriculture, etc., but notably the police, secret service and customs)....

• At about 20 points in the national information infrastructure network, all traffic is spliced off and fed into the Försvarets Radioanstalt (FRA) agency. These points are placed as to catch all traffic entering and leaving the Swedish borders, but will catch much - if not most - domestic traffic too, for technical routing reasons. Electronic traffic, in particular, always takes the scenic route.
• This affects all Internet traffic and all telephony traffic, meaning web surfing, e-mail, phone, and fax are affected, to mention but a few.
• The FRA will scan all traffic in real time according to about 250,000 search criteria. The traffic that matches will be automatically saved for manual intelligence analysis. This obviously takes a lot of computing power. We don't know the exact extent of FRA's computing power, but we do know that they have the world's fifth most powerful computer, in competition mostly with nuclear physics labs.
• "Customers" that will be able to place requests for searches include all authorities (all some 500 of them including Department of Transportation, Department of Agriculture, etc., but notably the police, secret service and customs).
• The political administration may order (not request, but order) a political wiretapping to catch communications they are interested in.
• Major businesses will also get access to the wiretapping grid, but will have to go through an authority.
• The bill specifically allows for singling out Swedish people for specific wiretapping, although only under certain qualifiers.
• The mandate for the agency's own intelligence gathering is broadened from "external military threats" to "external threats", which are exemplified as international crime; trafficking in drugs, weapons, or people; migration movements; religious or cultural conflicts; environmental imbalances and threats; raw materials shortages; and currency speculation. More examples are listed.

Linux has needed a single, unified, vision

Good idea! When will you have it ready for us? :-)

I'm joking. Linux has developed into an ecosystem, not just a single product, so why reduce it to that?

The people using Linux on their mobile phones, and those using it on their research supercomputers might disagree with your idea of optimizing it for the desktop only.

There's room for everyone.

Some operating systems provide a well defined (documented) set of interfaces/APIs/libraries to develop with. Version info included.

Some don't.

The market shows which philosophy is more successful.

Yes, the Webserver market, the embedded market and the supercomputer markets do indeed show which OS is superior.

Thanks for the insight, AC! :)

it will most likely just be used for more nuclear weapons simulations [emph mine]

The majority (but not all) supercomputers on the top 500 supercomputer list are related not to nuclear weapons research, but meteorological/oceanographic & other scientific uses.

Again, at best you've proved that Windows popularity has fallen and that Linux "could make inroads". "Could", if for example, it started selling.

Cute. You got Walmart to stop selling linux in stores. Nobody cares. Walmart.com is open 24/7 and they offer three different kinds of Linux on 24 different platforms. Wanna try again?

Mac sales are up true, in the US mainly and this has slightly dented Vista sales.

Up 50% quarter over year ago, if you don't mind, and it hasn't slightly dented Vista sales. Vista has driven customers to it. Don't you get it? This still doesn't explain your loss of 1/3 of your 90% market share. That's not bleeding. That's hemorraging.

And what of the server end? Well, Windows seems to be gaining there too; let's be honest, if there's one area Windows could do better in, it's the server.

Actually if there's one area Windows could be doing better in, it's supercomputing where linux owns 85% of the space.

On the server side if you could find a way to sell server operating systems to Google you might make a dent. Maybe you'll have a better chance with Yawho? though, because I really don't see you making any inroads with the Google. Server 2003 isn't horrible. Server 2008 isn't bad. Neither of them is Open and that's the death of them. How many parked pages you can buy to geek your numbers on Netcraft isn't fooling anybody since 2005. Oh, Hey, did DHS fix that nasty SQL injection bug that had IIS serving malware to all of your Windows clients? Us Linux users don't worry about such nonsense of course, but I'm worried about some of my customers who are still afflicted with IE.

Now, i'm not going to be disingenuous here; the OLPC and the eePC are doing well; almost in their own category, but even they run Windows now.

If those items had actually shipped in the first quarter you might have a point. Since they didn't, you've defeated yourself. Those items became popular with Linux and any way Microsoft can corrupt them is not going to change the fact that they became popular under Linux. Will they be popular with XP? We will see. They will never run Vista well and it's two years before you can offer anything better. You had better really bring your A game on that day, because Compiz is kicking your butt all over Youtube right now.

Look, if anyone just does a basic analysis, you'll see that there's this circular process where the heavier operating system requires new hardware, forcing people to buy both to keep up with the times, which both them and the manufacturer want.

According to this basic analysis(pdf), debian Etch is an order of magnitude larger and more complex than Vista. And yet it doesn't require this "new hardware" you're speaking of.

In fact in addition to the x86-32 and x86-64 targets Vista aims for it also runs on alpha, sparc, arm, powerpc, hppa, ia64, mips and s390. From the toys to spacecraft, from webservers to 85.2% of the world's top 500 supercomputers it'll run on almost anything. That's engineering.

This will not end until they have a solid competitor, period, and that means the linux geeks have got to get off their high horse and make an easy, packaged, "buy your box from dell with it pre-loaded" version of it your grandma can use.

You have been able to buy PCs preloaded with linux from Walmart, Dell, IBM, HP and many others for several years.

Because, personally, i'm getting a little sick of getting these operating systems from Microsoft which I swear to God have code running several extra loops just to bog it down so that only the most bleeding edge (aka money I don't want to spend) boxes can handle it reasonably.

So switch. It's time. Ballmer says Vista is a work in progress. Gates says its replacement is a year out. Let's take their word for it. This is a great window of opportunity to justify looking at alternatives.

Many cheap processors are great, until your definition of "Many" expands to over 200000 and you still need them to be very reliable, accurate, and play well with others (Power6 interconnects run at half clock speed.)

Well, all the computers in my institute run on Linux.

Almost all of the supercomputers run Linux or BSD.[1]

As John Ousterhout said: "The second problem I have seen (really more of a limitation) is that open-source software hasn't broken out of the "tools and systems" arena."
So while I can give you that, it makes me think about Apache, Firefox, MySQL etc.

So, yeah, "open source has been a ridiculously huge success".

Build the 4th fastest supercomputer:

1. Build the worlds fastest supercomputer
2. Load Vista

Want to drop off the end of the Top 500list? Load SP1.

Dude the site you provided ... was very useful ... I just had to look up the list for November 2007 & look up the details of the fourth supercomputer to check if the original post got it right.

And guess what? Its right there ... don't believe me? Well look up the top 10 list for Nov '07.

And to quote from that site you pointed to:

Try as I might, I could find no information that this list is only for supercomputers that are publicly funded. May be I didn't try hard enough ... could you pls give reference to the claim you make?

My understanding is that the list of Fastest Supercomputers is for supercomputers that are publicly funded - which is why we don't see any mention of google or microsoft clusters. But yet the summary claims that this one has not taken any government funding.

So then how can this cluster claim to be fourth fastest, when they really are not competing with the publicly funded clusters of the main list?

Obviously, we cannot expect the big corporate supercomputers to spill their beans to contest this, but it doesn't really look like it would be fair to compare this non-publicly-funded supercomputer to those that are.

The system was initially slated to use Intel's maligned V8 platform, but was later changed to the current Skulltrail - incorporating two quad-core CPUs natively running at 3.2GHz on a motherboard that supported four graphics cards - when the design became available.

A little-known fact is that Armari's technical director, Dan Goldsmith, being the eccentric chap he is, bought a decommissioned Cray supercomputer - used in the Cold War - a while back. Cray's extra-large computers (by today's standards) required some serious cooling, as you would expect, and Cray engineered some class-leading liquid cooling to keep the voluminous beast operating within tolerances.

Dan has used the inspiration from Cray's research, and indeed the coolant itself, which works in a temperature-range of -110C through to 90C, as a base for the XCP (eXtreme Concept Prototype) - the total immersion model.

Umm, this runs Linux... SLES to be specific:

http://www.top500.org/system/ranking/8966

According to http://www.top500.org/stats/list/30/os when the linux is identified, most are suse.

I'm glad to see AMD based projects like this, as they have certainly took a hit in the HPC space as of late.

Reading the article, the goal is nowhere near building a real exaflop computer, but more about thinking about issues (like processor data feeding).

In a year and a half, we shouln't have more than 100 GFlops per socket, which means that you will still need 10 millions of processors (not cores!) to achieve the exaflop computer. No chance to build a cluster that big (at least these years).

The all-times progression of the top500 shows that exaflop computers should arrive around year 2020, definetly not tomorrow. (x10 every ~4 years, 2008:1 PF, 2012:10 PF, 2016:100 PF, 2020:1 EF)

And what would be the killer app that needed all that extra power?

I can think of at least two applications that are often in the news: protein folding and physical simulations of continuous media, like weather and climate, aerodynamics, water, oil, and gas flow in porous rocks, etc.

But I think the future applications for personal supercomputers haven't been invented yet. We don't have the brains to predict what super-human artificial intelligence will be like.

then you haven't been paying attention to http://www.top500.org/list/2007/11/100.
nnote the 3rd position (and there are several others down the list).

and what would have happened if Dell went all out putting Linux on the front page, only selling Linux machines no MS Windows and it was a failure? There's another ten years of "Linux Sucks" right there.

No. Dell did the right thing by slowly growing their Linux desktop market and now everyone is copying them.

Branding matters a lot.

It's the reason Microsoft runs it's Get the facts campaign against Linux. Having Linux associated with big brands that people have heard of increases your chance of people picking your product. It doesn't matter that Linux runs on the top 8 super computers of the world because people will make judgements based of how familiar they are with a product.

This is why Ubuntu is more popular then other distributions, because Mark S. has associated Ubuntu with larger brands. More people know about Ubuntu and are more likely to pick it compared to another distributions. A lot of people here on /. grumble about "Why Noobuntu, why not try X". Well now you know, if distribution X had better branding it would probably be more popular then Ubuntu.

Another branding example..

Have you noticed recently how "Windows Server" adverts keep popping up on websites such as top500.org, sourceforge, etc? Places that decision makers might see them, but also developers. Sourceforge in particular seems to have tons of Microsoft adverts that it is starting to put me off visiting that website at all.

With specs like those, Linux may become known as a "low quality" operating system. To the masses, at least.

I'll explain: Joe Consumer buys a system for $200. He realizes that he can't run his Windows apps easily/at all, that it's "different" and "difficult" from what he knows (Microsoft, again), and it's kind of slow. He'll associate Linux with incompatibility, difficulties, and piss poor performance. And he may tell his friends.

Looking at some US statistics of R&D expenditures on a global scale it seems UK is on 18th place, calculated _per capita_.
(http://nsf.gov/statistics/seind06/c4/tt04-13.htm) That would make the 17th rank well deserved. ;)

Israel is number one, followed by Sweden, Finland, Japan, Iceland, and the US.

Of course the absolute magnitude is of importance here. That allows the UK to engage in research never attainable by e.g. the Icelandic however high their per capita is. With the US in at a sixth place (per capita) and with the largest economy in place, we can safely assume they can do any research they wish. And buy any computer there is. The www.top500.org is a testimony of that. http://www.top500.org/overtime/list/30/countries thus also recognizes the UK economy, and gives it a rank 4! So, the original grand grand parent should not complain. The UK is fine.

I guess the situation has improved in the last year, but my point is that linux is a newcomer in the big iron world.

But I think it's not still mature enough to compete in high-performance, high scalable, mission critical environments...

> Linux has it's role in the server market, and it's a very important one. But I think it's not still mature enough to compete in high-performance,

Linux is not competing on high-performance computers. It OWNS the high-performance computers. Currently about 85% of the top500 super computers are using Linux:
http://www.top500.org/stats/list/30/osfam

Here is the OS family list.

Disclaimer: I work with SGI, so I can shed some light on their customer's perspective (NASA, gov't, research labs, etc.) and solution to this problem.

The increasing density of servers is exacerbating the problem of power and cooling in every data center. This week is the SuperComputing trade show where the the new top 500 supercomputers edition was released with "Big Turnover Among the Top 10 Systems," where you can see the first examples to address these issues.

SGI's new ICE blade system was launched a few months ago, it was designed to address the power consumption, real estate density, and cooling issues everyone will probably experience on their next server cycle. ICE has shipped and one installation is now #3 on the Top 500. It's a welcome sign that SGI is back from bankruptcy. I'm sorry if this seems like an advert, so I'm not going to link to SGI -- you can go find out more easily if you want.

Well, 6 of the top 500 are, anyway. The rest are plain ol' scalar machines, albeit with gobs of processors ("gobs" being the technical term for "OMG how many?!?").

Well the top500 list was around since 1993 listing supercomputers. So the best you can say is there's two definitions - the one everyone else in the industry (and every dictionary I can find) follows where supercomputer refers to a specific category of computers that are simply the most powerful during their time, and the one you're pushing for 1 TFlop but for which I have yet to find any objective support.

Google definition
Top500 1993

There's an interesting paper that analyzes the data accumulated in the top500 list site, which ranks the 500 most powerful supercomputers twice a year: it shows that, over time, the share of vector machines within the list is sharply declining, both in aggregated power and in number: from around 60% in 1993 to around 10% in 2003 (see Figure 3, page 6, in said paper). Still, vector machines refuse to die and always seem to maintain a presence in the top500, as is evident from the above slashdot post. Will vector machines live forever?

8 PS3s offer at least 1.6TFLOPS. The 2007.6 TOP500 supercomputers bottom out at just over 4TFLOPS. Two years ago, this PS3 cluster might even have made it to the TOP500 list proper; nine years ago it might have topped the list.

It might not be in the top 500, but it does seem to be a supercomputer.

8 PS3s offer at least 1.6TFLOPS. The 2007.6 TOP500 supercomputers bottom out at just over 4TFLOPS. Two years ago, this PS3 cluster might even have made it to the TOP500 list proper; nine years ago it might have topped the list.

It might not be in the top 500, but it does seem to be a supercomputer.

8 PS3s offer at least 1.6TFLOPS. The 2007.6 TOP500 supercomputers bottom out at just over 4TFLOPS. Two years ago, this PS3 cluster might even have made it to the TOP500 list proper; nine years ago it might have topped the list.

It might not be in the top 500, but it does seem to be a supercomputer.

I think that he was referring to this cluster market. Windows has 0.4% and Linux has 77.8% of that market.

I'm a physics to-be grad student, so I share your gripe on the SSC. However, I should note that US has the lead in supercomputing by far. In fact, we hold the first 8 positions on the Top 500 list.

Despite the ITER fiasco, we still have a lot of promising research going on in the fusion department.
We have one of the best astrophysics programs in the world.

Despite the SSC never getting finished, it is our collider, the Tevatron, that has potentially found the Higgs boson

It is our lasers that have reached kW and MW range. (Now that's a tough google search ... it returns mW instead of MW. Damn case insensitivity).

It is our robots that have made giant strides in their walking procedures. Japan is trying to introduce their classically-modelled robots into the mainstream. We're working on getting them faster and more efficient.

We're still the primary innovators, by far.

Can we PLEASE stop trying to sell Linux as the cheap knock-off? that's the only way it's ever going to get anywhere

Linux runs nicely on machines with thousands of cores: http://www.top500.org/

but yes, that would be totally wasted on Vista.

If you want to talk about whether Real Science is being done, too many of the top machines are working on various aspects of Weapons of Mass Destruction and therefore aren't publishing a lot of results for the scientific community... For a number of years, the fastest machine was Seti@Home, but their statistics reporting has gotten a lot less useful so it's no longer easy to tell how much CPU is being used to search for signals from little green men, and I think by about 2005 it was no longer on top even ignoring the linpak-vs-bogomips disconnect.

What do you mean by 'stable' exactly? Linux has crash-free, problem-free uptimes in the 6 9's range 99.9999% or 31.5 seconds of trouble per year (a reboot may take 2 minutes), so this basically means you may need to actually reboot the computer once every four years or so. Is that what you meant? Did you know that the largest supercomputers in the US run Linux? You did know that didn't you? You know, the ones they use to do nuclear stockpile testing? Please go to http://www.top500.org/ for a list of the worlds current 500 fastest supercomputers. Please look to see how many are running Linux ... (77.8% including all in the top 20). Is that what you meant by 'stable'? If you are talking about 'crashing' or 'freezing', the version of Linux available 15 years ago was more stable than windows vista. Perhaps thats the answer you are looking for?

I write multi-threaded floating point intensive systems for a living. I simply want to use the chip that gives me the greatest floating point throughput I can get. Right now that chip appears to be Barcelona. I'm not interested with hypothetical arguments - only speed. I am looking forward to using Barcelona processors because they will get my mathematical computations done faster.

By the way, check out number 2 and 3 on your top 500 supercomputer list - they're Opterons.

When only measuring single core performance, clock for clock, Barcelona is on par with Cloverton.

Unfortunately processors are not generally sold "clock for clock." If you're on par clock for clock, but the other guy is clocked more than 50% faster than you... that could be trouble.

What good is an Intel chip that has fast floating point but the bus cannot feed it data fast enough?

Plenty good if the data can fit in cache, in which case the unit can be fed fast enough. For instance, say you're running LinPack. But then, who uses LinPack as a benchmark?