What's new

China’s Tianhe-2 Retains Top Spot on List of World’s Top Supercomputers

Exactly.

That is why I say, that acquiring a powerful supercomputer is not a big deal, given of course, that you have cash.

The Saudis, literally did NOTHING concerning their supercomputer, except to roll out cash, yet were able to procure a top 10 supercomputer, that is still a thing of envy for most of the world.

The whole focus of HPC was astray sometime back, with just efforts to get brute force out of the machine. Not much work was done to build an accompanying ecosystem, of application softwares etc.

Similarly, China still doesn't have similar core competencies like the US, for most of the competencies are similar to the ones commercially used widely. That is the most fundamental aspects of supercomputers are processors, and accelerators, (their design, fabrication) all of which China is behind the US in.

China is catching up, will continue to do so, but it will still take time. Essentially the capability to design good accelerators and processors is the same, be its application in normal PCs, or in supercomputers. That is the reason why the commercial companies leading the processor and graphic card market in commercial space, Intel and Nvidia, also have the best supercomputing products.

The overall architecture is actually the most important part of the supercomputer. You can always buy commercial chips or even use less efficient chips. Otherwise your network topology will be extremely inefficient and the processors will interfere with each other. This is why a supercomputer is fundamentally different than a 1/2 core computer. Also, the front end chips of the Tianhe 2 are Chinese, I believe. Only the accelerators are Intel.
 
not for long.

the U.S has in the works 3 supercomputers with 100 to 300 petaflop performance

NVIDIA Volta GPUs and IBM Power9 CPUs To Deliver Up To 300 PetaFlops of Performance in 2017 With Summit and Sierra Supercomputers
Intel's 10nm Knights Hill Powered Aurora Supercomputer To Feature Up To 180 PetaFlops Computational Power - 2018 Launch Scheduled


so tell me how is China suppose to build it's next gen supercomputers and even upgrade the Tiahe-2 if it can't buy U.S semiconduters o_O

are you going to use longsoon processors or even Russia Elbrus :rofl:

maybe if you play nice with Japan they'll let you use SPARC 64 :lol:

At least do some research before you start trolling.

China May Develop Two 100 Petaflop Machines Within a Year - insideHPC
100 petaflop.JPG
 
I thought the "stick" measuring prevailed only between Pakistan & India :)

On a side note China will still clock 100 petaflops one year earlier than US so this indeed is an achievement worth feeling proud of. And, if China can deliver 100+ pertaflops with indigenous processors and hardware then the US supremacy (achieved in 2017) will be strongly threatened

like I said 3 times or more times before if China can built a supercomputer with only Chinese hardware I will eat crow :D


but anyhow the U.S and China will dominate the HPC sector for years if not decades to come.


the race to exaflop is on.


MAY
EXPECTS

not definitive :disagree:

who is trolling? I said I would eat my words if I am wrong :angel:
 
  • China's supercomputer growth is exploding while the US is losing steam
    It comes as no surprise that Linux is the operating system of choice for supercomputers. What is surprising is that China is now the world's fastest growing supercomputer power while the US has fallen to its lowest level ever.


    By Steven J. Vaughan-Nichols for Linux and Open Source | November 17, 2015 -- 13:22 GMT (21:22 GMT+08:00) | Topic: Hardware

    • 10
    • Linux runs the supercomputing world. In the latest, November 2015. Top500 list of the world's fastest supercomputers, Linux reached a new high of 98.8 percent, 494 of the 500.



      tianhe-2.jpg

      China's Linux-powered Tianhe 2 supercomputer remains, by a wide margin, the world's fastest supercomputer.
      All the other high-performance operating systems have been knocked out. Only AIX, IBM's Unix, remains with six supercomputers to its credit. Windows? Solaris? They're history.


      There's nothing shocking about that. What is surprising is that China's supercomputer growth has exploded.

      According to the Top500 organization, "China nearly tripled the number of systems on the latest list, while the number of systems in the United States has fallen to the lowest point since the TOP500 list was created in 1993. China is also carving out a bigger share as a manufacturer of high performance computers with multiple Chinese manufacturers becoming more active in this field."

      Specifically, China has leaped to 109 systems, from the 37 it boasted on the previous Top500 list. The United States still has the overall lead with 200 of the fastest supercomputers. However, that's down down from 231 in July. Worse still, it's the U.S.'s lowest number since the Top500 list was created in 1993.

      Elsewhere, Europe has fallen to 108 systems compared to 141 earlier this year. Japan's share has dropped slightly from 40 to 36 systems.

      At the same time, China's computer vendors are becoming leaders in high performance computers (HPC). Lenovo, following its acquisition of IBM's x86 server business in 2014, is now up to 25 Top500 systems from just three systems on the July 2015 list.

      Some systems that were previously listed as IBM are now joint IBM/Lenovo productions and Lenovo/IBM (five systems). Eventually, as the last of the contracts behind these supercomputers end, they'll go to Lenovo's count. Sugon, a Chinese HPC and server vendor that's relatively unknown in the West, has overtaken IBM with 49 systems.

      Overall, Hewlett-Packard Enterprise (HPE), with 136 supercomputers, is the leading HPC vendor. It's followed by Cray, 69, and Sugon.

At the very top of the list, and for the sixth straight time, Tianhe-2, a supercomputer developed by China's National University of Defense Technology, is the world's fastest computer. This time around it scored 33.86 petaflop/s (quadrillions of calculations per second or Pflop/s) on the Linpack benchmark.

Holding on to the No. 2 spot, Titan, a Cray XK7 system installed at DOE's Oak Ridge National Laboratory, is the top U.S. system. It scored 17.59 petaflop/s. In other words, Tianhe-2 remains much faster than it.

There are only two new systems in the Top 10. The first, at number six, is the Trinity supercomputer. It was built by Cray and is jointly deployed by the Department of Energy's (DOE) Los Alamos and Sandia national laboratories. In the number 8 spot, there's Cray'sHazel-Hen system. It's installed at Germany's HLRS Supercomputer center.

Cray, long synonymous with supercomputers, is on comeback trial. In terms of pure, performance, Cray systems now claim 24.9 percent share of installed total performance. That's up from 24 percent. While IBM has far fewer systems, the ones that it has left took second place with a 14.9 percent share. HPE, even though it easily has the most Top500 supercomputers, has 12.9 percent of the overall performance power. This places HPE in third place.

Overall, supercomputer performance is slowing down. The last place system's performance had grown by only 55 percent. That sounds impressive, but it pales compared to 1994 to 2008's annual 90 percent growth rate.

Technically, supercomputers are increasingly using accelerator/co-processor technology to boost performance. 104 systems on the Top500, up from 90 in July 2015. are now relying on this floating point technology. 66 of these use NVIDIA chips, three use ATI Radeon, and 27 systems work with Intel's Xeon Phi processors. Four systems use a combination of Nvidia and Intel Xeon Phi accelerators/co-processors.

What to make of all this? Well, there's no question about it. Linux rules supercomputing. But, that's old news. What's interesting, and disturbing, if you're interested in the U. S remaining a high-technology "leader" is that we're declining. And, with IBM now moving out of supercomputers, it appears that fall from supercomputing power will only continue. Within two years it seems all too likely that China will be the top supercomputing country.

China's supercomputer growth is exploding while the US is losing steam | ZDNet
 

Attachments

  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 6
  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 3
  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 4
  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 3
  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 3
  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 4
  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 4
  • upload_2015-11-18_8-43-59.gif
    upload_2015-11-18_8-43-59.gif
    43 bytes · Views: 3
Just wondering if China will be the first to introduce an exaflops supercomputer. Obama executive order is shooting to be the first.

The US Government is not satisfied with being number two, and is eager to surpass China. As per Obama's Executive Order, DoD, Department of Energy along with the National Science Foundation will lead the initiative.

Let the race continue, wish China continues to maintain the top spot.
 
The US Government is not satisfied with being number two, and is eager to surpass China. As per Obama's Executive Order, DoD, Department of Energy along with the National Science Foundation will lead the initiative.

Let the race continue, wish China continues to maintain the top spot.

Let The Good vs The Evil game begin! :bounce:
 
These high speed super computer are very critical. They help build Y-20,C919, W14 glider , CZ-5 rocket and many other system by conducting accurate simulation.

Yes, supercomputers are used for many purposes, some critical to national security (like the weaponry you just mention) and the country's economic viability. Scientists could use supercomputer to pioneer research into all fields ranging from nuclear energy to climate change.
 
The US Government is not satisfied with being number two, and is eager to surpass China. As per Obama's Executive Order, DoD, Department of Energy along with the National Science Foundation will lead the initiative.

Let the race continue, wish China continues to maintain the top spot.

The US is already number one in supercomputer technologies and core competencies.

Also, developing an exa scale is fundamentally different from current supercomputers, because the greatest thing required right now is research to reduce energy consumption per TFlop.

While, you can have an exa scale, with present technologies, the power consumption will be simply massive, and the system would be of no much use.
 
The US is already number one in supercomputer technologies and core competencies.

Also, developing an exa scale is fundamentally different from current supercomputers, because the greatest thing required right now is research to reduce energy consumption per TFlop.

While, you can have an exa scale, with present technologies, the power consumption will be simply massive, and the system would be of no much use.


Don't get me wrong, no one doubt US competencies in supercomputing. I refer to the fact that US is second only to China in computing speed, yes that's also a hard reality well acknowledged even by US government.

On future applications let's not worry, there is no limit on scientific frontier for mankind, history is a testament to that. However yes the energy consumption is an issue, under current technological limit an exascale computer will need power equivalent to that of a small city. Let the race continue.
 
Last edited:
The US is already number one in supercomputer technologies and core competencies.

Also, developing an exa scale is fundamentally different from current supercomputers, because the greatest thing required right now is research to reduce energy consumption per TFlop.

While, you can have an exa scale, with present technologies, the power consumption will be simply massive, and the system would be of no much use.


Thank you, sir! Someone really needs to pass on this important massage to those Chinese supercomputer scientists, otherwise they are going to make a grave mistake!!! They may end up building a supercomputer that they don't know where to get electricity to power it up!
 
Thank you, sir! Someone really needs to pass on this important massage to those Chinese supercomputer scientists, otherwise they are going to make a grave mistake!!! They may end up building a supercomputer that they don't know where to get electricity to power it up!

They already know.

The whole supercomputing community does. That is why the major emphasis of HPC in modern world is on energy efficiency, and energy consumption per TFLOP, rather than the peak performance alone.
 

Latest posts

Back
Top Bottom