Slashdot Mirror

Open source software is a dying fad. There's really no need for Linux. Windows and OS X are just fine for most people. Even half of the new supercomputers in the top 500 in the past year are running Windows or OS X. Linux was a fad in the past 15 years, but it's going away and people are switching back to Windows and OS X.

It's a lovely day for feeding the trolls...

http://www.top500.org/statistics/sublist/

OS Family:
Linux: 476 out of 500 INCLUDING numbers 1-43 consecutively
Unix: 16
BSD: 1 (# 342)
Windows: 3 (#'s 187, 241, 289)
OSX: lol?

By the way, the 1 BSD system is SUPER-UX.

... so we've named a supercomputer after her:

http://www.top500.org/system/176952

It's a pretty giant beastie. There are some pretty awesome pictures of the front rack floating around:

http://news.techgenie.com/files/Hopper-Supercomputer.jpg

http://www.top500.org/list/2013/06/
Top 100 fastest supercomputers in the world include 6 that actually do use Ethernet. Two are using gigabit Ethernet (the other 4 use 10G). Modern super computers do have interconnect issues, but part of that is because a typical system has >10000 cores for general compute plus specialized compute cores (possibly GPU) on top of that. That is beyond what a Linksys can do.

You bring up a good point and part of the reason why Livermore, Sandia and Los Alamos have those nice big supercomputers testing decay rates and doing simulations on warheads.

There's an interesting device in the Bradbury Science Museum aka the Atomic Museum in Los Alamos, It's a phone..

Anyway, from this: http://www.nationaltlcservice.us/2013/05/report-from-the-hilltop-highlights-of-the-los-alamos-bradbury-science-museum-museum-profile-1/

A phone analogy inaugurated the display: Adjacent to a clear-plastic telephone (which reminded me of those see-through Swatch phones of the 80s), a placard explains: “Like many of the weapons currently in the nuclear arsenal, this phone was manufactured in the late 1960s and was designed to last about 15 years. You were asked to verify that this phone will work—but you weren’t allowed to make or receive a call to fully test it.” Nearby, the question “What does this phone have to do with nuclear weapons?” is answered with the motto: “safe, secure, and reliable nuclear weapons.” The exhibit further explains the connection to LANL’s mission: “We are asked to verify that the weapons in the stockpile are safe and reliable—but without performing underground nuclear tests. Instead, we use an integrated set of scientific tools to inspect and evaluate individual parts and subsystems. The military counts on us to guarantee that US nuclear weapons will perform as designed if they are ever needed. That’s our mission, and that’s a call we can make.”

The DOE still has quite a few on the Top 500 List..

And even if you stand on your head, this guy surely has seen a TFlop/s much earlier than you. And he probably gave the TFlop/s and the Petaflop/s its name - as his program is the tool to actually figure if a computer is able to put out a TFlop/s or a PFlop/s.

It's interesting to browse this website:
http://www.top500.org/
And look at the Statistics section, such as Operating System Family
http://www.top500.org/statistics/list/
Operating system Familyâf Countâf System Share (%)âf Rmax (GFlops)âf Rpeak (GFlops)âf Coresâf
Linux 476 95.2 217,913,963 318,748,391 18,700,112
Unix 16 3.2 3,949,373 4,923,380 181,120
Mixed 4 0.8 1,184,521 1,420,492 417,792
Windows 3 0.6 465,600 628,129 46,092
BSD Based 1 0.2 122,400 131,072 1,280

Linux has owned the big-machine (Server, HPC) and small-machine (embedded, phone) markets forever. It's the desktop where Linux can't get traction.

It's interesting to browse this website:
http://www.top500.org/
And look at the Statistics section, such as Operating System Family
http://www.top500.org/statistics/list/
Operating system Familyâf Countâf System Share (%)âf Rmax (GFlops)âf Rpeak (GFlops)âf Coresâf
Linux 476 95.2 217,913,963 318,748,391 18,700,112
Unix 16 3.2 3,949,373 4,923,380 181,120
Mixed 4 0.8 1,184,521 1,420,492 417,792
Windows 3 0.6 465,600 628,129 46,092
BSD Based 1 0.2 122,400 131,072 1,280

Linux has owned the big-machine (Server, HPC) and small-machine (embedded, phone) markets forever. It's the desktop where Linux can't get traction.

It's interesting to browse this website:
http://www.top500.org/
And look at the Statistics section, such as Operating System Family
http://www.top500.org/statistics/list/
Operating system Familyâf Countâf System Share (%)âf Rmax (GFlops)âf Rpeak (GFlops)âf Coresâf
Linux 476 95.2 217,913,963 318,748,391 18,700,112
Unix 16 3.2 3,949,373 4,923,380 181,120
Mixed 4 0.8 1,184,521 1,420,492 417,792
Windows 3 0.6 465,600 628,129 46,092
BSD Based 1 0.2 122,400 131,072 1,280

It's interesting to browse this website:
http://www.top500.org/
And look at the Statistics section, such as Operating System Family
http://www.top500.org/statistics/list/
Operating system Familyâf Countâf System Share (%)âf Rmax (GFlops)âf Rpeak (GFlops)âf Coresâf
Linux 476 95.2 217,913,963 318,748,391 18,700,112
Unix 16 3.2 3,949,373 4,923,380 181,120
Mixed 4 0.8 1,184,521 1,420,492 417,792
Windows 3 0.6 465,600 628,129 46,092
BSD Based 1 0.2 122,400 131,072 1,280

30.65 petaflops is about double the 17.6 petaflops of the current top performer on the TOP 500 list.

Of course, the devil will be in the details. It is easy to deliver high peak scores in supercomputing, and more difficult to hit high average scores. Also, the current list is from November, and it is possible that the American supercomputers are newer / faster / better too.

I don't buy your response: http://top500.org/statistics/list/ ... click accelerator and hit submit.

87.6% of the top 500 super computers have no NVIDIA etc. coprocessing

is there a top500 for diy clusters?

The top systems today are approximately at the same capacity as a human brain.

Brain neurons perform an operation that's similar to a dot product. Their operation can be simulated by a weight for each dendrite that's multiplied by that dendrite's input.

In rough order of magnitude, a human brain has a hundred billion neurons, or 1e11 in standard computer language notation. Each neuron has an average of one thousand inputs, 1e3, and performs a hundred operations per second. That is 1e11 * 1e3 * 1e2 = 1e16 flops, or 10,000 teraflops.

According to Top500, the highest powered computer system in November 2012 had a capacity of 17,590 teraflops.

This doesn't mean it has the same ability as a human brain, because there's also the software involved. There is a project, sponsored by Google, that tries to implement a computer system operating close to what the human brain does.

When they tested that system presenting to it one million random screenshots from Youtube videos, the system learned all by itself to recognize objects that appeared on those videos, like human faces and cats.

There's a good technical tutorial on this system at the Stanford university site, and a more basic explanation can be found in several popular articles if you google for "deep learning".

And yet all those government supercomputers like those run by the NSA (Never Say Anything ...oops, National Security Agency), National Nuclear Security Agency, Central Intelligence Agency, and places like Argonne National Labs, Lawrence Livermore National Labs, and others (see here), are using soo much Linux. Is it that after affording the million CPU's in the machine, they can't afford an operating system? US defence spending not enough? Ummm Nope. Linux really is the best system out there (bar none). Where it really counts, use Linux. I'm just waiting for US congress to make a kerfuffle about it all, and when the truth comes out, the wider public finally finds out.

Now this is domination. And this is starting to look like domination. Looks like embedded still has a way to go, though Linux overall looks healthier than ever.

NASA's Pleiades is 14th in top500 (at the level of Oct 2012)... I guess it still qualifies to "plenty of computation power".

Hum, why BSD? He mentioned linux not because is the best solution ever (which might or might not be), but because a lot of petaflop-capable code was written specifically to run on it.. and because the big names (IBM, Crazy) fully support it. In fact, I don't remember ever using a BSD-based supercomputer. The top500 only shows one machine, at 0.1 petaflop, running a bsd-based OS. Search for os here: http://top500.org/statistics/sublist/

Am I wrong thinking that this is not dramatically faster than Titan (27 TF peak)
http://www.top500.org/system/177975

The specifications in the doc are interesting nonetheless!

27 Petaflops at Oak Ridge
20 Petaflops at Lawrence Livermore

http://top500.org/lists/2012/11/

The probe in question is Voyager 1 and was launched in 1977. Let's not call it modern. It was designed in an era when there was no such thing as a personal computer. A high end cellphone probably has more battery-powered computing power in your pocket now than all of the compute resources of NASA back then. Imagine what those engineers could achieve with this. Materials science has progressed also. But the biggest gift of days is in our understanding the rich resources available in the space around us. Water is abundant everywhere from Mercury to the edge of the solar system. We didn't know that back then. Almost all stars have planets in the habitable zone. We didn't know that either.

It's unlikely a mission to Vega would launch any sooner than 2037, or 60 years after the launch of the first Voyager. We have learned a lot of things since Voyager 1 was launched, and will have learned more. That none have gone faster is an artifact of 30 years of neglect of space operations, but not space science. At the moment Vega is too far to a man to reach in his span of years with the science we have, though another star might be. There is no reason to expect that this will always be so.

With VASMR 200KW thrusters entering service on the ISS in a few years, and the development of suitable power plants ongoing, we still would need fuel - LOTS of fuel - on orbit or somewhere near zero-G to make a go of it. Fortunately in 26 months the NASA Dawn mission will arrive at Ceres and find there a practically unlimited supply of Xenon, Argon, Hydrogen and Oxygen ready for mining as well as a surface amenable to easily building human habitats on. You may schedule two years from now for the space Gold Rush to begin.

Ceres is not only the perfect source for interstellar fuels: it's also the perfect launchpad as it should be possible to build a railgun there 1000KM long capable of launching interstellar probes with solar system escape velocity that don't require any fuel at all. It's also the only minor planet so situated within easy reach.

Planetary Resources, SpaceX, Virgin Galactic and others are all over this. The people behind these efforts are some of the brightest, most successful minds the world has ever known. Elon Musk. Sergey Brin. Larry Page. Eric Schmidt. Richard Branson. These are but a few. They know something you don't know.

They used that line with computers too "Every computer and OS has its problems, and happy users probably aren't as vocal." And it was a bullshit lie then too. Linux *is* more fucking stable. It *does* perform much better. The shills yelp and yelp like little kids till they start believing their own crap. There's a reason Linux is used by 93.8% of the 500 fastest supercomputers, and has 96.1% of the performance share: it works. Its like: we can get 500,000 8 core processors running at super high efficiency, and so can also get a 2 core arm processor make a tablet perform really well. And unlike other systems, there is 1 linux kernel, and it runs a galaxy S3 as well as the DOE/SC/Oak Ridge National Laboratory Titan - Cray XK7 , Opteron 6274 16C 2.200GHz, Cray Gemini interconnect, NVIDIA K20x (currently #1 on the list with 560640 cores). Don't believe me? Well then look over here ya putz!

Those 8 TFLOPS would have landed it somewhere at the top of the #500 supercomputer performance list in November, 2011. ASCI White used 8192 375MHz Power3 cores to achieve this performance. It took up a fair bit of space and used 3 MW to run the machine with a further 3 MW needed for cooling. It had a theoretical processing speed of 12.3 teraflops.

Those 8 TFLOPS would have landed it somewhere at the top of the #500 supercomputer performance list in November, 2011. ASCI White used 8192 375MHz Power3 cores to achieve this performance. It took up a fair bit of space and used 3 MW to run the machine with a further 3 MW needed for cooling. It had a theoretical processing speed of 12.3 teraflops.

.

From the following, it sounds like 1 Teraflop - not theoretical, but on Linpack - Is available on a desktop, now or very soon:

Intel has been working hard on its many-integrated core (MIC), which it describes as a 50+ core capable of one teraflops real-world performance. Intel revealed its strategy and product branding at the International SuperÂcomputing Conference held last June in Hamburg. It showed the Xeon Phi as an AIB that fits into one PCIe slot. The system had two Xeon E5 processors and a Knights Corner co-processor running the Linpack benchmark and hitting the magic one teraflops number.

August 31, 2012

That would be the fastest supercomputer in the world until early 1997.

The Top500 reports actual performance as measured with LINPACK, hardware vendors report the theoretical performance of their chips, which in the case of GPUs is often quite a bit more than you'd be able to squeeze out with LINPACK.

For comparison: Tsubame 2.0 consists of 1400 nodes with approx. 4200 NVIDIA Tesla C2075, which should yield -- according to your estimate -- 2.1 PFLOPS (4200 * 0.5 TFLOPS), yet it is listed at 1.2 PFLOPS. So just add two years to your estimate and you should be fine...

In November, 2001, the fastest supercomputer was 12 TFlops. You can achieve that today for less than $5,000 on your desktop by ganging together four GPGPU cards (such as the 3 TFlops Radeon 7970 for less than $500 each). Go back to 1999 and it's only 3 TFlops and to match today you wouldn't even need a special motherboard.

So just wait 11 years for the prices to come down.

Would I buy any current AMD processors for a server farm? Probably not.

The predecessors of this series, the Opteron 6200 is used in quite a few supercomputers. Actually I counted 21 Opteron bases systems in the last supercomputer top 100 list.

You are absolutely wrong. 75% of super computers run on Linux. Go and see.

I thought that sounded low, so I went and checked at http://i.top500.org/stats . Linux has 92.4 % of the top 500. Then you have "Unix" at 4.8 and Mixed at 2.2.

You are absolutely wrong. 75% of super computers run on Linux. Go and see.

I thought that sounded low, so I went and checked at http://i.top500.org/stats . Linux has 92.4 % of the top 500. Then you have "Unix" at 4.8 and Mixed at 2.2.

You are absolutely wrong. 75% of super computers run on Linux. Go and see.

Shocking! Say it ain't so! It must be because nasty old Linux stole all that technology from Bill Gates and Steve Jobs.

You are absolutely wrong. 75% of super computers run on Linux. Go and see.

The official list of Top 500 (last updated 2012/06) states "Sequoia - BlueGene/Q, Power BQC 16C 1.60 GHz, Custom" as the number one super computer. Sequoia is nearly as powerful as Titan.

Windows isn't used for high-performance computing, nor as a real-time OS, so it probably won't ever get this feature.

O RLY? (okok, only 2 of the top500, but it's not like it's NOT used. I'd be surprised if it's used anywhere that's not being paid by Microsoft to do it though.)

As I said :-)

Windows isn't used for high-performance computing, nor as a real-time OS, so it probably won't ever get this feature.

O RLY? (okok, only 2 of the top500, but it's not like it's NOT used. I'd be surprised if it's used anywhere that's not being paid by Microsoft to do it though.)

http://www.top500.org/lists/2007/11
http://www.top500.org/lists/2012/06
Should be interesting to see them double the rate of growth over the preceding five years

http://www.top500.org/lists/2007/11
http://www.top500.org/lists/2012/06
Should be interesting to see them double the rate of growth over the preceding five years

The Top 500 is a specific list: http://top500.org/

It's more correct to say it's the fastest on the list, than the fastest in the world. There are any number of metrics you can use to compare supercomputers. Top 500 just uses the most popular metric. Another machine could easily be the fastest on a different list, like http://www.graph500.org/.

The other specific consideration is that the list is ONLY for those that volunteer to run the Linpack benchmark and wish to publicize the results. It is presumed that governments with classified computing facilities withhold this information, for obvious reasons, so there are likely many "supercomputers" (perhaps even a "fastest") that will never be part of the Top 500. The US NSA, for example, is widely believed to operate facilities at or near the top of the list, but they are nowhere in sight for obvious reasons.

The Top 500 is a specific list: http://top500.org/

It's more correct to say it's the fastest on the list, than the fastest in the world. There are any number of metrics you can use to compare supercomputers. Top 500 just uses the most popular metric. Another machine could easily be the fastest on a different list, like http://www.graph500.org/.

AMD doesn't exist outside of the low end server space. Precisely what sector were you talking about, where you can price out an AMD machine on the web? The machines I was talking about scale out to greater than 4 sockets, and possess features aimed at the highest end of the market such as RAS and multi-year service/software support contracts. AMD's latest Interlagos models don't even support more than 4 sockets without custom interconnects and glue logic, and if you're going for more sockets than that, you're in Top500 supercomputer territory, where cost is not a consideration anyway. If it were, AMD would be doing better than it is, as that is the only reason to buy an AMD server nowadays.

Top 500 stats by processor family:

x86-64 - 435

Intel Xeon - 372

AMD Opteron - 62

If you look at the historical graphs, AMD has been falling in share ever since 2009.

"OS X is Unix which is all Linux is pretending to be"

Huh? That may have been true a decade ago.

Have a look at http://i.top500.org/overtime and you'll see that Linux overtook and topped Unix between 2002 and 2005.

OSX today? Not of any significant relevance for the last few years.

Aww, how cute! England wants to come out and play. While I am happy to see that our mates from across the pond are getting themselves a nice little number cruncher - it is still little. NASA has a setup called Pleiades:

Total cores: 112,896
Total memory: 191 TB

But here's the real hard-to-fathom point. The sport two 11-dimension hypercube interconnect configurations using Infiniband QDR and DDR networking (mostly DDR). Now DDR is ~4GB/s and QDR is ~8GB/s, but Inifiniband is rarely singily-connected. Usually you multipath using four connections to the switch, which in this case bumps the transfer rate up to 16GB/s for DDR and 32GB/s for QDR. Peak theoretical speed for this system is ~240TB/s.

Oh, and it's running SUSE Linux. That off-the-shelf enough?

Last I checked, the system is running at 90% utilization. This is one heck of a cluster and it isn't just for show. Yet Pleiades only rates as #7 on the TOP500 list of supercomputers worldwide. The new list comes out in a few days. We'll see if they can keep that illustrious position.

Side note: The #1 supercomputer is running an interconnect called "Tofu". Since I'm an Infiniband guy, I'd like to know more. Right now, though, it rates very high on my silliness scale.

The single data rate switch chips have a latency of 200 nanoseconds, DDR switch chips have a latency of 140 nanoseconds and QDR switch chips have a latency of 100 nanoseconds.

Can these numbers be directly compared to the "fiber-to-fiber latency" reported in the article summary?

No joke, the last Windows cluster in the Top500, ShanghaiSupercomputing Center's Dawning 5000A went SLES10 and now there is not even one.

I wonder if they were able to get their Windows refund.

You think that's bad? Try being the boss of Windows HPC. That guy fell of the Top500 entirely. No joke, the last Windows cluster in the Top500, ShanghaiSupercomputing Center's Dawning 5000A went SLES10 and now there is not even one. He must be so lonely.

Let's look this up. 7 years ago #1 on the Top500 was an IBM BlueGene/L at 70 TFLOPS. I can't see that performance anywhere close on the desktop or even on the notebook market.

Assuming you're running a good SLI systems and that your GPUs actually deliver the performance the manufacturer is claiming them to have, you'd get in the best case something around 1.5 TFLOPS which corresponds roughly to a 1998 ASCI Red.

Wow, we're on Slashdot......almost like being On The Cover of the Rolling Stone.

Answers to various questions and comments:
- We support the Linux toolchain; compilers, debuggers, etc., fortunate to have some of the original gcc team. Ported pieces of various kernels to TOMI Aurora to make certain we had not left anything out and to test the memory manager. Aurora was for use in a tablet type device.
- TOMI Borealis was optimized for Big Data and unstructured data apps like MapReduce that choke at the Memory Wall. Linux could probably be ported without too much difficulty. Most massively parallel installations will use something really light weight instead.
- Potential users said give them more integer cores instead of adding FPU. We gladly cede the FP world to Itanium.
- For raw FP horsepower within a reasonable power budget, its tough to beat Nvidia's GPU approach. That is probably why 3 of the top 10 supercomputers are GPU accelerated. http://www.top500.org/ GPU-type architectures will likely be the future of scientific computing. Venray is focused on Memory Wall limited areas such as Big Data.
- From the computer architecture perspective, the distinction between Big Data and Small Data is whether the datasets will primarily fit within the onboard caches. Video compression, graphics acceleration, encryption, and much of LINPAC (http://en.wikipedia.org/wiki/LINPACK) would be classed as Small Data since most of the computing can be done without leaving the caches (high locality). Legacy architectures choke on Big Data since the datasets overflow the caches and there is much much less data reuse.
- MapReduce is important because it is currently the most visible Big Data application thanks to Google. http://research.google.com/archive/mapreduce.html
- Venray believes Big Data applications are the future of computing. So does McKinsey Consulting. http://www.mckinsey.com/Insights/MGI/Research/Technology_and_Innovation/Big_data_The_next_frontier_for_innovation We leave it to others to accelerate MS Office and Call of Duty.
- The future of Big Data appears to be RAM resident, not disk, not even flash. (See Fred Ho's work at IBM.) https://www.ibm.com/developerworks/mydeveloperworks/blogs/fredho66/?lang=en_us
- re: Mitsubish 3DRAM and other similar ventures, iRAM, Exacute, Gilgamesh, etc....they embedded DRAM into logic. Contrast with TOMI that embeds CPU cores into DRAMs.....our benefits are performance and particularly cost: http://www.edn.com/photo/294/294788-microprocessor_vs_memory_transistors_graph.jpg
- We chose a modified RISC architecture rather than a special purpose one such as Gilgamesh in order to make programming simpler with well understood Linux tools such as gcc. Submit your gcc C, C++, or Fortran to http://www.venraytechnology.com./ Statistics are returned in standard dGen format.
- TSV (through silicon vias) and HMC (hybrid memory cube) are valid attempts to push back the Memory Wall. Discussed in Part 1 for EDN. http://www.edn.com/article/520059-The_future_of_computers_Part_1_Multicore_and_the_Memory_Wall.php Decision may be determined by cost.
- Would love to dispense with caches because they add transistors. 4K data and 4K instruction caches sped us up about 10x. Unlike legacy architectures, TOMI cache lines load in a single DRAM cycle.
- Yes love Raspberry Pi. http://www.raspberrypi.org/
- Quad-

The first top 500 list was published in June 1993. The fastest computer on that list was a CM-5/1024 made by Thinking Machines Corporation. It was rated at: 59.70 Rmax(GFs) and 131.00 Rpeak(GFs).

Last place on that first top 500 list (scroll down) was held by a VP-200 made by Fujitsu/SNI which had 1 core and was rated at 0.422 Rmax(GFs) and 0.533 Rpeak(GFs).

I've heard the expression about carrying a supercomputer in your pocket - how close are we? I'd expect most of the latest Android/iPhone/smartphones can beat that last-place finisher from 1993. I'm doubtful that any of these devices could beat that first place finisher, but I suspect desktops (especially with GPUs) should be there by now. If you're are interested, you can get the software from here.

Any takers? How does YOUR system compare?

The first top 500 list was published in June 1993. The fastest computer on that list was a CM-5/1024 made by Thinking Machines Corporation. It was rated at: 59.70 Rmax(GFs) and 131.00 Rpeak(GFs).

Last place on that first top 500 list (scroll down) was held by a VP-200 made by Fujitsu/SNI which had 1 core and was rated at 0.422 Rmax(GFs) and 0.533 Rpeak(GFs).

I've heard the expression about carrying a supercomputer in your pocket - how close are we? I'd expect most of the latest Android/iPhone/smartphones can beat that last-place finisher from 1993. I'm doubtful that any of these devices could beat that first place finisher, but I suspect desktops (especially with GPUs) should be there by now. If you're are interested, you can get the software from here.

Any takers? How does YOUR system compare?

The first top 500 list was published in June 1993. The fastest computer on that list was a CM-5/1024 made by Thinking Machines Corporation. It was rated at: 59.70 Rmax(GFs) and 131.00 Rpeak(GFs).

Last place on that first top 500 list (scroll down) was held by a VP-200 made by Fujitsu/SNI which had 1 core and was rated at 0.422 Rmax(GFs) and 0.533 Rpeak(GFs).

I've heard the expression about carrying a supercomputer in your pocket - how close are we? I'd expect most of the latest Android/iPhone/smartphones can beat that last-place finisher from 1993. I'm doubtful that any of these devices could beat that first place finisher, but I suspect desktops (especially with GPUs) should be there by now. If you're are interested, you can get the software from here.

Any takers? How does YOUR system compare?

Way back in '03, Virginia Tech built a cluster of 1,100 Mac G5's. It came in at #3 on the Top 500 list that year, and at $5.2M, it was a fraction of the cost of the next cheapest supercomputer in the top ten. And it was assembled by students in 3 weeks, using stock G5 towers fitted with InfiniBand cards.

It was later upgraded to G5 xServe boxes, and as of 2008, was still ranked 281 on the Top 500 list.

Here's a short promo film that VT produced: http://www.youtube.com/watch?v=vLujLtgBJC0