supercomputer

SUPERCOMPUTER

REMEMBER THE MOVIE 2001 AND HAL

 

Dee Finney's blog

start date July 20, 2011

today's date April 27, 2014

page 671

TOPIC:  THE WORLD'S SUPER COMPUTER   

NOTE:  WHEN I FIRST HEARD ABOUT THIS SUPER COMPUTER - I THOUGHT - CERTAINLY CHINA WOULD HAVE IT.  THE UNITED STATES NO LONGER IS TOP DOG

INTERESTING VIDEO - HOWEVER - WHOEVER OPERATED THIS COMPUTER CAN'T SPELL.  https://www.youtube.com/watch?v=Vz7_oCrBzB4

The 'Unal Center of Education, Research & Development' (UCERD) is based on the theme of carrying out research and education in the areas of Computer Science, Electrical and Mechanical Systems. The mission of UCERD is to investigate, develop and manage computer technology in order to facilitate scientific progress. With this aim, special dedication has been awarded to areas such as Computers, Telecomm and Mechanical Applications in Science and Engineering. All these activities are complementary to each other and very tightly related. The primary objective of UCERD is to build multidisciplinary setup that facilitates industrial and scientific academic practices to improve their understanding of needs and to help them in focusing on the basic research.

The UCERD is further categorized into following groups:

Research & Studies

UCERD research group motivates people to study confirmed facts and explore to collect or view more knowledge and provide beneficial information that enrich the mind.  It continually upgrade knowledge and achieve improvement to what we currently have to satisfy the human desire to "know".
The importance of research and knowledge in Islam can be determine by this fact that the very first verses of Quran revealed to the Holy Prophet are about research knowledge.

The Quran says: "Read! In the name of thy Lord  and Cherisher Who created. Created man out (mere) clot of congealed blood. Read! and thy Lord is Most Bountiful. He Who taught (the use of) the Pen. Taught  man that which he knew not." The Quran asks: "Are those equal, those who know and those who do not know? "

The Hadith (sayings of Prophet Peace be upon him) literature is also full of references to the importance of knowledge. The Messenger of God (Peace be upon him) said: 'Seek knowledge from the cradle to the grave.' The Prophet Muhammad (Peace be upon him) said: 'Seeking of knowledge is the duty of every man and woman.'

Projects

UCERD Project Channel provides  practical experience using scientific method and helps to stimulate interest in scientific inquiry. The outcome of a UCERD professional and  investigatory project involves a discovery that can improve the lives of people and improve  their life style.

Technology News Channel

As society is developing more, advance means of communication are required. UCERD TechNews channel  plays an important role to keep people informed about latest technologies.

UCERD Team

UCERD is working and searching more collaboration between institutions, research groups across different disciplines, between researchers from academia and industry, between International and local organization. The Studies, Research & Development carried out by the UCERD is organized in several research groups. Their research and development work is published in international conferences and journals.

UCERD People

UCERD Research & Development Team is finding friendly technologies to fulfill human needs so that science and humanity go on.

Hussain Cheema hussain.tassadaq@gmail.com
Tassadaq Hussain received his Masters degree in Electronics for Systems from the Institut supérieur d’électronique de Paris in 2009. Currently  he is doing Ph.D. degree in Computer Architecture from UPC . For the period of the PhD he is working as a research student with the Computer Sciences group at the Microsoft Barcelona Supercomputing Center (BSC). The group is responsible to investigate and evolve high performance computing systems for data and compute bound applications. Mostly the group’s research is focused on Shared Memory Multiprocessor Architectures, Clustered Computing, Reconfigurable Computing, Multi/Many Core Architectures, Heterogeneous Architectures, Transactional Memories, ISA (Instruction Set Architecture) virtualizations, Microprocessor Micro-architectures and Software Instrumentation & Profiling.
Tassadaq’s research focuses on studying the use of accelerators for supercomputing purposes. The main focus is to schedule multi-accelerator/vecotor processor data movement in hardware. At the theoretical level we are proposing novel architectures for the acceleration of HPC applications

 

Laeeq Ahmad  laeeq@ucerd.com
Laeeq earned a Bachelor of Science in Electrical Engineering specialization in Telecommunication Studies from Riphah International University. Laeeq Ahmad has worked in Telecommunication Department since 2005. In 2005 Laeeq enlisted into the Huawei Telecom. He has been working for 7 years in several leading and diverse organizations in senior leadership positions.As Member of the Grop Lead his responsibilities include, providing IT Management Assurance and Knowledge Management

 

Faisal  faysal@ucerd.com
 Mr. Faisal has experience of High speed and RF PCB designing with Emphases of Signal and Power integrity, EMC and Thermal analysis since 2004. He has experience on designing closed loop servo motor control system for robotic applications.

 

Farooq Riaz Malhi   Mobilink Pakistan
farooq.riaz@mobilink.net
Mr Farooq Riaz is a Telecom Engineer of Mobilink Networks. He got a bachelor’s degree in Electrical Engineering with specialization in Electronics and Communications in 2005. For last 5 years, he is in the field of telecommunications. Currently he is working as Team Lead (Fiber Optics Operations) in PMCL-Mobilink Pakistan.  He has extensive experience of OSP issues and also had Network Management training from Hauwei University, China.

 

Ahmad Sulaiman Naseer Siemens
ahmed.naseer@siemens.com
 For last 6 years, Ahmad Sulaiman is working as team leader and player in Siemens & presently working as permanent employee.

 

Safwan Mehboob Qureshi  safwanmehboob@hotmail.com
Mr Safwan has diverse experience of telecommunication, started from the installation, upgrade and ended up in Level 3 & 4 support for TETRA communication.

 

Uzma Siddique Researcher: Wireless Comm, University of Manitoba, Canada
siddiquu@ee.umanitoba.ca
 Uzma expertise are in field of Telecommunications to formulate joint scheduling and resource allocation problems and power control problem and to design procedures/algorithms for Multi-tier OFDMA based networks Uplink and Downlink taking into account the single Cell and Multicell scenarios.

 

 

 Supercomputer is a computer at the frontline of contemporary processing capacity – particularly speed of calculation which can happen at speeds of nanoseconds.

Supercomputers were introduced in the 1960s, made initially and, for decades, primarily by Seymour Cray at Control Data Corporation (CDC), Cray Research and subsequent companies bearing his name or monogram. While the supercomputers of the 1970s used only a few processors, in the 1990s machines with thousands of processors began to appear and, by the end of the 20th century, massively parallelsupercomputers with tens of thousands of "off-the-shelf" processors were the norm. As of November 2013, China's Tianhe-2 supercomputer is the fastest in the world at 33.86 petaFLOPS, or 33.86 quadrillion floating point operations per second.

Systems with massive numbers of processors generally take one of two paths: In one approach (e.g., in distributed computing), a large number of discrete computers (e.g.,laptops) distributed across a network (e.g., the Internet) devote some or all of their time to solving a common problem; each individual computer (client) receives and completes many small tasks, reporting the results to a central server which integrates the task results from all the clients into the overall solution. In another approach, a large number of dedicated processors are placed in close proximity to each other (e.g. in a computer cluster); this saves considerable time moving data around and makes it possible for the processors to work together (rather than on separate tasks), for example in mesh and hypercube architectures.

The use of multi-core processors combined with centralization is an emerging trend; one can think of this as a small cluster (the multicore processor in a smartphone, tablet, laptop, etc.) that both depends upon and contributes to the cloud.

Supercomputers play an important role in the field of computational science, and are used for a wide range of computationally intensive tasks in various fields, including quantum mechanics, weather forecasting, climate research, oil and gas exploration, molecular modeling(computing the structures and properties of chemical compounds, biological macromolecules, polymers, and crystals), and physical simulations (such as simulations of the early moments of the universe, airplane and spacecraft aerodynamics, the detonation of nuclear weapons, and nuclear fusion). Throughout their history, they have been essential in the field of cryptanalysis.

History

A Cray-1 preserved at theDeutsches Museum

The history of supercomputing goes back to the 1960s, with the Atlas at the University of Manchester and a series of computers at Control Data Corporation (CDC), designed by Seymour Cray. These used innovative designs and parallelism to achieve superior computational peak performance.

The Atlas was a joint venture between Ferranti and the Manchester University and was designed to operate at processing speeds approaching one microsecond per instruction, about one million instructions per second. The first Atlas was officially commissioned on 7 December 1962 as one of the world's first supercomputers – considered to be the most powerful computer in the world at that time by a considerable margin, and equivalent to four IBM 7094s.

The CDC 6600, released in 1964, was designed by Cray to be the fastest in the world by a large margin. Cray switched from germanium to silicon transistors, which he ran very fast, solving the overheating problem by introducing refrigeration. Given that the 6600 outran all computers of the time by about 10 times, it was dubbed a supercomputer and defined the supercomputing market when one hundred computers were sold at $8 million each.

Cray left CDC in 1972 to form his own company. Four years after leaving CDC, Cray delivered the 80 MHz Cray 1 in 1976, and it became one of the most successful supercomputers in history. The Cray-2 released in 1985 was an 8 processor liquid cooledcomputer and Fluorinert was pumped through it as it operated. It performed at 1.9 gigaflops and was the world's fastest until 1990.

While the supercomputers of the 1980s used only a few processors, in the 1990s, machines with thousands of processors began to appear both in the United States and in Japan, setting new computational performance records. Fujitsu's Numerical Wind Tunnelsupercomputer used 166 vector processors to gain the top spot in 1994 with a peak speed of 1.7 gigaflops per processor. TheHitachi SR2201 obtained a peak performance of 600 gigaflops in 1996 by using 2048 processors connected via a fast three-dimensionalcrossbar network.The Intel Paragon could have 1000 to 4000 Intel i860 processors in various configurations, and was ranked the fastest in the world in 1993. The Paragon was a MIMD machine which connected processors via a high speed two dimensional mesh, allowing processes to execute on separate nodes; communicating via the Message Passing Interface.

Hardware and architecture

A Blue Gene/L cabinet showing the stacked blades, each holding many processors

Approaches to supercomputer architecture have taken dramatic turns since the earliest systems were introduced in the 1960s. Early supercomputer architectures pioneered by Seymour Cray relied on compact innovative designs and local parallelism to achieve superior computational peak performance.However, in time the demand for increased computational power ushered in the age of massively parallelsystems.

While the supercomputers of the 1970s used only a few processors, in the 1990s, machines with thousands of processors began to appear and by the end of the 20th century, massively parallel supercomputers with tens of thousands of "off-the-shelf" processors were the norm. Supercomputers of the 21st century can use over 100,000 processors (some being graphic units) connected by fast connections.

Throughout the decades, the management of heat density has remained a key issue for most centralized supercomputers. The large amount of heat generated by a system may also have other effects, e.g. reducing the lifetime of other system components. There have been diverse approaches to heat management, from pumping Fluorinert through the system, to a hybrid liquid-air cooling system or air cooling with normal air conditioning temperatures.

The CPU share of TOP500

Systems with a massive number of processors generally take one of two paths. In the grid computingapproach, the processing power of a large number of computers, organised as distributed, diverse administrative domains, is opportunistically used whenever a computer is available. In another approach, a large number of processors are used in close proximity to each other, e.g. in acomputer cluster. In such a centralized massively parallel system the speed and flexibility of the interconnect becomes very important and modern supercomputers have used various approaches ranging from enhanced Infiniband systems to three-dimensional torus interconnects. The use of multi-core processors combined with centralization is an emerging direction, e.g. as in the Cyclops64 system.

As the price/performance of general purpose graphic processors (GPGPUs) has improved, a number of petaflop supercomputers such as Tianhe-I and Nebulae have started to rely on them. However, other systems such as the K computer continue to use conventional processors such as SPARC-based designs and the overall applicability of GPGPUs in general purpose high performance computing applications has been the subject of debate, in that while a GPGPU maybe tuned to score well on specific benchmarks its overall applicability to everyday algorithms may be limited unless significant effort is spent to tune the application towards it. However, GPUs are gaining ground and in 2012 the Jaguar supercomputer was transformed into Titan by replacing CPUs with GPUs.

A number of "special-purpose" systems have been designed, dedicated to a single problem. This allows the use of specially programmedFPGA chips or even custom VLSI chips, allowing better price/performance ratios by sacrificing generality. Examples of special-purpose supercomputers include Belle, Deep Blue, and Hydra, for playing chess, Gravity Pipe for astrophysics, MDGRAPE-3 for protein structure computation molecular dynamics and Deep Crack, for breaking the DES cipher.

Energy usage and heat management

A typical supercomputer consumes large amounts of electrical power, almost all of which is converted into heat, requiring cooling. For example, Tianhe-1A consumes 4.04 Megawatts of electricity. The cost to power and cool the system can be significant, e.g. 4MW at $0.10/kWh is $400 an hour or about $3.5 million per year.

Heat management is a major issue in complex electronic devices, and affects powerful computer systems in various ways. The thermal design power and CPU power dissipation issues in supercomputing surpass those of traditional computer cooling technologies. The supercomputing awards for green computing reflect this issue.

The packing of thousands of processors together inevitably generates significant amounts of heat density that need to be dealt with. The Cray 2 was liquid cooled, and used a Fluorinert "cooling waterfall" which was forced through the modules under pressure. However, the submerged liquid cooling approach was not practical for the multi-cabinet systems based on off-the-shelf processors, and in System X a special cooling system that combined air conditioning with liquid cooling was developed in conjunction with the Liebert company.

In the Blue Gene system IBM deliberately used low power processors to deal with heat density. On the other hand, the IBM Power 775, released in 2011, has closely packed elements that require water cooling. The IBM Aquasar system, on the other hand uses hot water cooling to achieve energy efficiency, the water being used to heat buildings as well.

The energy efficiency of computer systems is generally measured in terms of "FLOPS per Watt". In 2008 IBM's Roadrunner operated at 376 MFLOPS/Watt. In November 2010, the Blue Gene/Q reached 1684 MFLOPS/Watt. In June 2011 the top 2 spots on theGreen 500 list were occupied by Blue Gene machines in New York (one achieving 2097 MFLOPS/W) with the DEGIMA cluster in Nagasaki placing third with 1375 MFLOPS/W.

Software and system management

Operating systems

Since the end of the 20th century, supercomputer operating systems have undergone major transformations, based on the changes insupercomputer architecture. While early operating systems were custom tailored to each supercomputer to gain speed, the trend has been to move away from in-house operating systems to the adaptation of generic software such as Linux.

Since modern massively parallel supercomputers typically separate computations from other services by using multiple types of nodes, they usually run different operating systems on different nodes, e.g. using a small and efficient lightweight kernel such as CNK or CNL on compute nodes, but a larger system such as a Linux-derivative on server and I/O nodes.

While in a traditional multi-user computer system job scheduling is in effect a tasking problem for processing and peripheral resources, in a massively parallel system, the job management system needs to manage the allocation of both computational and communication resources, as well as gracefully dealing with inevitable hardware failures when tens of thousands of processors are present.

Although most modern supercomputers use the Linux operating system, each manufacturer has its own specific Linux-derivative, and no industry standard exists, partly due to the fact that the differences in hardware architectures require changes to optimize the operating system to each hardware design.

Software tools and message passing

Wide-angle view of the ALMAcorrelator.[65]

The parallel architectures of supercomputers often dictate the use of special programming techniques to exploit their speed. Software tools for distributed processing include standard APIssuch as MPI and PVM, VTL, and open source-based software solutions such as Beowulf.

In the most common scenario, environments such as PVM and MPI for loosely connected clusters and OpenMP for tightly coordinated shared memory machines are used. Significant effort is required to optimize an algorithm for the interconnect characteristics of the machine it will be run on; the aim is to prevent any of the CPUs from wasting time waiting on data from other nodes.GPGPUs have hundreds of processor cores and are programmed using programming models such as CUDA.

Moreover, it is quite difficult to debug and test parallel programs. Special techniques need to be used for testing and debugging such applications.

Distributed supercomputing

Opportunistic approaches

Example architecture of a grid computing system connecting many personal computers over the internet

Opportunistic Supercomputing is a form of networked grid computing whereby a “super virtual computer” of many loosely coupled volunteer computing machines performs very large computing tasks. Grid computing has been applied to a number of large-scale embarrassingly parallelproblems that require supercomputing performance scales. However, basic grid and cloud computing approaches that rely on volunteer computing can not handle traditional supercomputing tasks such as fluid dynamic simulations.

The fastest grid computing system is the distributed computing project Folding@home. F@h reported 8.1 petaflops of x86 processing power as of March 2012. Of this, 5.8 petaflops are contributed by clients running on various GPUs, 1.7 petaflops come from PlayStation 3 systems, and the rest from various CPU systems.

The BOINC platform hosts a number of distributed computing projects. As of May 2011, BOINC recorded a processing power of over 5.5 petaflops through over 480,000 active computers on the network[67] The most active project (measured by computational power), MilkyWay@home, reports processing power of over 700 teraflops through over 33,000 active computers.

As of May 2011, GIMPS's distributed Mersenne Prime search currently achieves about 60 teraflops through over 25,000 registered computers. The Internet PrimeNet Server supports GIMPS's grid computing approach, one of the earliest and most successful grid computing projects, since 1997.

Quasi-opportunistic approaches

Quasi-opportunistic supercomputing is a form of distributed computing whereby the “super virtual computer” of a large number of networked geographically disperse computers performs huge processing power demanding computing tasks. Quasi-opportunistic supercomputing aims to provide a higher quality of service than opportunistic grid computing by achieving more control over the assignment of tasks to distributed resources and the use of intelligence about the availability and reliability of individual systems within the supercomputing network. However, quasi-opportunistic distributed execution of demanding parallel computing software in grids should be achieved through implementation of grid-wise allocation agreements, co-allocation subsystems, communication topology-aware allocation mechanisms, fault tolerant message passing libraries and data pre-conditioning.

Performance measurement

Capability vs capacity

Supercomputers generally aim for the maximum in capability computing rather than capacity computing. Capability computing is typically thought of as using the maximum computing power to solve a single large problem in the shortest amount of time. Often a capability system is able to solve a problem of a size or complexity that no other computer can, e.g. a very complex weather simulationapplication.

Capacity computing in contrast is typically thought of as using efficient cost-effective computing power to solve a small number of somewhat large problems or a large number of small problems, e.g. many user access requests to a database or a web site.Architectures that lend themselves to supporting many users for routine everyday tasks may have a lot of capacity but are not typically considered supercomputers, given that they do not solve a single very complex problem.

Performance metrics

Top supercomputer speeds: logscale speed over 60 years

In general, the speed of supercomputers is measured and benchmarked in "FLOPS" (FLoating Point Operations Per Second), and not in terms of MIPS, i.e. as "instructions per second", as is the case with general purpose computers. These measurements are commonly used with an SI prefix such as tera-, combined into the shorthand "TFLOPS" (1012 FLOPS, pronounced teraflops), or peta-, combined into the shorthand "PFLOPS" (1015 FLOPS, pronounced petaflops.) "Petascale" supercomputers can process one quadrillion (1015) (1000 trillion) FLOPS. Exascale is computing performance in the exaflops range. An exaflop is one quintillion (1018) FLOPS (one million teraflops).

No single number can reflect the overall performance of a computer system, yet the goal of the Linpack benchmark is to approximate how fast the computer solves numerical problems and it is widely used in the industry. The FLOPS measurement is either quoted based on the theoretical floating point performance of a processor (derived from manufacturer's processor specifications and shown as "Rpeak" in the TOP500 lists) which is generally unachievable when running real workloads, or the achievable throughput, derived from the LINPACK benchmarks and shown as "Rmax" in the TOP500 list. The LINPACK benchmark typically performs LU decomposition of a large matrix. The LINPACK performance gives some indication of performance for some real-world problems, but does not necessarily match the processing requirements of many other supercomputer workloads, which for example may require more memory bandwidth, or may require better integer computing performance, or may need a high performance I/O system to achieve high levels of performance.

The TOP500 list

Pie chart showing share of supercomputers by countries from top 500 supercomputers as of November 2013

Since 1993, the fastest supercomputers have been ranked on the TOP500 list according to their LINPACK benchmark results. The list does not claim to be unbiased or definitive, but it is a widely cited current definition of the "fastest" supercomputer available at any given time.

This is a recent list of the computers which appeared at the top of the TOP500 list, and the "Peak speed" is given as the "Rmax" rating. For more historical data see History of supercomputing.

Year Supercomputer Peak speed
(Rmax)
Location
2008 IBM Roadrunner 1.026 PFLOPS Los Alamos, USA
1.105 PFLOPS
2009 Cray Jaguar 1.759 PFLOPS Oak Ridge, USA
2010 Tianhe-IA 2.566 PFLOPS Tianjin, China
2011 Fujitsu K computer 10.51 PFLOPS Kobe, Japan
2012 IBM Sequoia 17.17 PFLOPS Livermore, USA
2012 Cray Titan 17.59 PFLOPS Oak Ridge, USA
2013 NUDT Tianhe-2 33.86 PFLOPS Guangzhou, China

Applications of supercomputers

The stages of supercomputer application may be summarized in the following table:

Decade Uses and computer involved
1970s Weather forecasting, aerodynamic research (Cray-1).
1980s Probabilistic analysis, radiation shielding modeling (CDC Cyber).
1990s Brute force code breaking (EFF DES cracker),
2000s 3D nuclear test simulations as a substitute for legal conduct Nuclear Non-Proliferation Treaty (ASCI Q).
2010s Molecular Dynamics Simulation (Tianhe-1A)

The IBM Blue Gene/P computer has been used to simulate a number of artificial neurons equivalent to approximately one percent of a human cerebral cortex, containing 1.6 billion neurons with approximately 9 trillion connections. The same research group also succeeded in using a supercomputer to simulate a number of artificial neurons equivalent to the entirety of a rat's brain.

Modern-day weather forecasting also relies on supercomputers. The National Oceanic and Atmospheric Administration uses supercomputers to crunch hundreds of millions of observations to help make weather forecasts more accurate.

In 2011, the challenges and difficulties in pushing the envelope in supercomputing were underscored by IBM's abandonment of the Blue Waters petascale project.

Research and development trends

Diagram of a 3-dimensional torus interconnect used by systems such as Blue Gene, Cray XT3, etc.

Given the current speed of progress, industry experts estimate that supercomputers will reach 1 exaflops (1018, one quintillion FLOPS) by 2018. China has stated plans to have a 1 exaflop supercomputer online by 2018. Using the Intel MIC multi-core processor architecture, which is Intel's response to GPU systems, SGI plans to achieve a 500-fold increase in performance by 2018, in order to achieve one exaflop. Samples of MIC chips with 32 cores, which combine vector processing units with standard CPU, have become available. The Indian government has also stated ambitions for an exaflop-range supercomputer, which they hope to complete by 2017.

Erik P. DeBenedictis of Sandia National Laboratories theorizes that a zettaflop (1021, one sextillion FLOPS) computer is required to accomplish full weather modeling, which could cover a two-week time span accurately.[87][not in citation given] Such systems might be built around 2030.

See also

Notes and references[edit]

  1. Jump up^ "IBM Blue gene announcement". 03.ibm.com. 2007-06-26. Retrieved 2012-06-09.
  2. ^ Jump up to:a b Hoffman, Allan R.; et al. (1990). Supercomputers: directions in technology and applications. National Academies. pp. 35–47.ISBN 0-309-04088-4.
  3. ^ Jump up to:a b Hill, Mark Donald; Jouppi, Norman Paul; Sohi, Gurindar (1999). Readings in computer architecture. pp. 40–49. ISBN 1-55860-539-8.
  4. ^ Jump up to:a b Prodan, Radu; Fahringer, Thomas (2007). Grid computing: experiment management, tool integration, and scientific workflows. pp. 1–4. ISBN 3-540-69261-4.
  5. Jump up^ DesktopGrid
  6. ^ Jump up to:a b Performance Modelling and Optimization of Memory Access on Cellular Computer Architecture Cyclops64 K Barner, GR Gao, Z Hu, Lecture Notes in Computer Science, 2005, Volume 3779, Network and Parallel Computing, Pages 132–143
  7. ^ Jump up to:a b Analysis and performance results of computing betweenness centrality on IBM Cyclops64 by Guangming Tan, Vugranam C. Sreedhar and Guang R. Gao The Journal of SupercomputingVolume 56, Number 1, 1–24 September 2011
  8. Jump up^ Lemke, Tim (May 8, 2013). "NSA Breaks Ground on Massive Computing Center". Retrieved December 11, 2013.
  9. ^ Jump up to:a b Hardware software co-design of a multimedia SOC platformby Sao-Jie Chen, Guang-Huei Lin, Pao-Ann Hsiung, Yu-Hen Hu 2009 ISBN pages 70-72
  10. Jump up^ The Atlas, University of Manchester, retrieved 21 September 2010
  11. Jump up^ Lavington, Simon (1998), A History of Manchester Computers (2 ed.), Swindon: The British Computer Society, pp. 41–52,ISBN 978-1-902505-01-5
  12. Jump up^ The Supermen, Charles Murray, Wiley & Sons, 1997.
  13. Jump up^ A history of modern computing by Paul E. Ceruzzi 2003 ISBN 978-0-262-53203-7 page 161 [1]
  14. ^ Jump up to:a b Hannan, Caryn (2008). Wisconsin Biographical Dictionary. pp. 83–84. ISBN 1-878592-63-7.
  15. Jump up^ John Impagliazzo, John A. N. Lee (2004). History of computing in education. p. 172. ISBN 1-4020-8135-9.
  16. Jump up^ Richard Sisson, Christian K. Zacher (2006). The American Midwest: an interpretive encyclopedia. p. 1489. ISBN 0-253-34886-2.
  17. Jump up^ Readings in computer architecture by Mark Donald Hill, Norman Paul Jouppi, Gurindar Sohi 1999 ISBN 978-1-55860-539-8 page 41-48
  18. Jump up^ Milestones in computer science and information technology by Edwin D. Reilly 2003 ISBN 1-57356-521-0 page 65
  19. ^ Jump up to:a b c Parallel computing for real-time signal processing and control by M. O. Tokhi, Mohammad Alamgir Hossain 2003 ISBN 978-1-85233-599-1 pages 201–202
  20. Jump up^ "TOP500 Annual Report 1994". Netlib.org. 1996-10-01. Retrieved 2012-06-09.
  21. Jump up^ N. Hirose and M. Fukuda (1997). "Numerical Wind Tunnel (NWT) and CFD Research at National Aerospace Laboratory". Proceedings of HPC-Asia '97. IEEE Computer Society.doi:10.1109/HPC.1997.592130.
  22. Jump up^ H. Fujii, Y. Yasuda, H. Akashi, Y. Inagami, M. Koga, O. Ishihara, M. Syazwan, H. Wada, T. Sumimoto, Architecture and performance of the Hitachi SR2201 massively parallel processor system, Proceedings of 11th International Parallel Processing Symposium, April 1997, Pages 233–241.
  23. Jump up^ Y. Iwasaki, The CP-PACS project, Nuclear Physics B – Proceedings Supplements, Volume 60, Issues 1–2, January 1998, Pages 246–254.
  24. Jump up^ A.J. van der Steen, Overview of recent supercomputers, Publication of the NCF, Stichting Nationale Computer Faciliteiten, the Netherlands, January 1997.
  25. Jump up^ Scalable input/output: achieving system balance by Daniel A. Reed 2003 ISBN 978-0-262-68142-1 page 182
  26. Jump up^ Xue-June Yang, Xiang-Ke Liao, et al in Journal of Computer Science and Technology. "The TianHe-1A Supercomputer: Its Hardware and Software". 26, Number 3. pp. 344–351.
  27. Jump up^ The Supermen: Story of Seymour Cray and the Technical Wizards Behind the Supercomputer by Charles J. Murray 1997ISBN 0-471-04885-2 pages 133–135
  28. Jump up^ Parallel Computational Fluid Dyynamics; Recent Advances and Future Directions edited by Rupak Biswas 2010 ISBN 1-60595-022-X page 401
  29. Jump up^ Supercomputing Research Advances by Yongge Huáng 2008ISBN 1-60456-186-6 pages 313–314
  30. ^ Jump up to:a b Computational science – ICCS 2005: 5th international conference edited by Vaidy S. Sunderam 2005 ISBN 3-540-26043-9 pages 60–67
  31. Jump up^ Knight, Will: "IBM creates world's most powerful computer",NewScientist.com news service, June 2007
  32. Jump up^ N. R. Agida et al. (2005). "Blue Gene/L Torus Interconnection Network | IBM Journal of Research and Development" (PDF).Torus Interconnection Network. 45, No 2/3 March–May 2005. p. 265.
  33. Jump up^ Prickett, Timothy (May 31, 2010). "Top 500 supers – The Dawning of the GPUs". =Theregister.co.uk.
  34. Jump up^ Hans Hacker et al in Facing the Multicore-Challenge: Aspects of New Paradigms and Technologies in Parallel Computing by Rainer Keller, David Kramer and Jan-Philipp Weiss (2010). Considering GPGPU for HPC Centers: Is It Worth the Effort?. pp. 118–121.ISBN 3-642-16232-0.
  35. Jump up^ Damon Poeter (October 11, 2011). "Cray's Titan Supercomputer for ORNL Could Be World's Fastest". Pcmag.com.
  36. Jump up^ Feldman, Michael (October 11, 2011). "GPUs Will Morph ORNL's Jaguar Into 20-Petaflop Titan". Hpcwire.com.
  37. Jump up^ Timothy Prickett Morgan (October 11, 2011). "Oak Ridge changes Jaguar's spots from CPUs to GPUs". Theregister.co.uk.
  38. Jump up^ Condon, J.H. and K.Thompson, "Belle Chess Hardware", InAdvances in Computer Chess 3 (ed.M.R.B.Clarke), Pergamon Press, 1982.
  39. Jump up^ Hsu, Feng-hsiung (2002). Behind Deep Blue: Building the Computer that Defeated the World Chess Champion. Princeton University Press. ISBN 0-691-09065-3.
  40. Jump up^ C. Donninger, U. Lorenz. The Chess Monster Hydra. Proc. of 14th International Conference on Field-Programmable Logic and Applications (FPL), 2004, Antwerp – Belgium, LNCS 3203, pp. 927 – 932
  41. Jump up^ J Makino and M. Taiji, Scientific Simulations with Special Purpose Computers: The GRAPE Systems, Wiley. 1998.
  42. Jump up^ RIKEN press release, Completion of a one-petaflops computer system for simulation of molecular dynamics
  43. Jump up^ Electronic Frontier Foundation (1998). Cracking DES – Secrets of Encryption Research, Wiretap Politics & Chip Design. Oreilly & Associates Inc. ISBN 1-56592-520-3.
  44. Jump up^ "NVIDIA Tesla GPUs Power World's Fastest Supercomputer"(Press release). Nvidia. 29 October 2010.
  45. Jump up^ Balandin, Alexander A. (October 2009). "Better Computing Through CPU Cooling". Spectrum.ieee.org.
  46. Jump up^ "The Green 500". Green500.org.
  47. Jump up^ "Green 500 list ranks supercomputers". iTnews Australia.
  48. Jump up^ Wu-chun Feng (2003). "Making a Case for Efficient Supercomputing | ACM Queue Magazine, Volume 1 Issue 7, 10-01-2003 doi 10.1145/957717.957772" (PDF).
  49. Jump up^ "IBM uncloaks 20 petaflops BlueGene/Q super". The Register. 2010-11-22. Retrieved 2010-11-25.
  50. Jump up^ Prickett, Timothy (2011-07-15). "''The Register'': IBM 'Blue Waters' super node washes ashore in August". Theregister.co.uk. Retrieved 2012-06-09.
  51. Jump up^ "HPC Wire July 2, 2010". Hpcwire.com. 2010-07-02. Retrieved 2012-06-09.
  52. Jump up^ by Martin LaMonica (2010-05-10). "CNet May 10, 2010". News.cnet.com. Retrieved 2012-06-09.
  53. Jump up^ "Government unveils world's fastest computer". CNN. Archived from the original on 2008-06-10. "performing 376 million calculations for every watt of electricity used."
  54. Jump up^ "IBM Roadrunner Takes the Gold in the Petaflop Race".
  55. Jump up^ "Top500 Supercomputing List Reveals Computing Trends". "IBM... BlueGene/Q system .. setting a record in power efficiency with a value of 1,680 Mflops/watt, more than twice that of the next best system."
  56. Jump up^ "IBM Research A Clear Winner in Green 500".
  57. Jump up^ "Green 500 list". Green500.org. Retrieved 2012-06-09.
  58. ^ Jump up to:a b Encyclopedia of Parallel Computing by David Padua 2011ISBN 0-387-09765-1 pages 426–429
  59. Jump up^ Knowing machines: essays on technical change by Donald MacKenzie 1998 ISBN 0-262-63188-1 page 149-151
  60. Jump up^ Euro-Par 2004 Parallel Processing: 10th International Euro-Par Conference 2004, by Marco Danelutto, Marco Vanneschi and Domenico Laforenza ISBN 3-540-22924-8 pages 835
  61. Jump up^ Euro-Par 2006 Parallel Processing: 12th International Euro-Par Conference, 2006, by Wolfgang E. Nagel, Wolfgang V. Walter and Wolfgang Lehner ISBN 3-540-37783-2 page
  62. Jump up^ An Evaluation of the Oak Ridge National Laboratory Cray XT3 by Sadaf R. Alam etal International Journal of High Performance Computing Applications February 2008 vol. 22 no. 1 52–80
  63. Jump up^ Open Job Management Architecture for the Blue Gene/L Supercomputer by Yariv Aridor et al in Job scheduling strategies for parallel processing by Dror G. Feitelson 2005 ISBN ISBN 978-3-540-31024-2 pages 95–101
  64. Jump up^ "Top500 OS chart". Top500.org. Retrieved 2010-10-31.
  65. Jump up^ "Wide-angle view of the ALMA correlator". ESO Press Release. Retrieved 13 February 2013.
  66. Jump up^ Folding@home: OS Statistics. Stanford University. Retrieved 2012-03-11.
  67. Jump up^ BOINCstats: BOINC Combined. BOINC. Retrieved 2011-05-28Note this link will give current statistics, not those on the date last accessed
  68. Jump up^ BOINCstats: MilkyWay@home. BOINC. Retrieved 2011-05-28Note this link will give current statistics, not those on the date last accessed
  69. Jump up^ "Internet PrimeNet Server Distributed Computing Technology for the Great Internet Mersenne Prime Search". GIMPS. Retrieved June 6, 2011.
  70. ^ Jump up to:a b Kravtsov, Valentin; Carmeli, David; Dubitzky, Werner; Orda, Ariel; Schuster, Assaf; Yoshpa, Benny. "Quasi-opportunistic supercomputing in grids, hot topic paper (2007)". IEEE International Symposium on High Performance Distributed Computing. IEEE. Retrieved 4 August 2011.
  71. ^ Jump up to:a b c The Potential Impact of High-End Capability Computing on Four Illustrative Fields of Science and Engineering by Committee on the Potential Impact of High-End Computing on Illustrative Fields of Science and Engineering and National Research Council (October 28, 2008) ISBN 0-309-12485-9 page 9
  72. Jump up^ Xingfu Wu (1999). Performance Evaluation, Prediction and Visualization of Parallel Systems. pp. 114–117. ISBN 0-7923-8462-8.
  73. ^ Jump up to:a b Dongarra, Jack J.; Luszczek, Piotr; Petitet, Antoine (2003),"The LINPACK Benchmark: past, present and future",Concurrency and Computation: Practice and Experience (John Wiley & Sons, Ltd.): 803–820
  74. Jump up^ Intel brochure – 11/91. "Directory page for Top500 lists. Result for each list since June 1993". Top500.org. Retrieved 2010-10-31.
  75. Jump up^ "The Cray-1 Computer System" (PDF). Cray Research, Inc. Retrieved May 25, 2011.
  76. Jump up^ Joshi, Rajani R. (9 June 1998). "A new heuristic algorithm for probabilistic optimization" (Subscription required). Department of Mathematics and School of Biomedical Engineering, Indian Institute of Technology Powai, Bombay, India. Retrieved 2008-07-01.
  77. Jump up^ "Abstract for SAMSY – Shielding Analysis Modular System". OECD Nuclear Energy Agency, Issy-les-Moulineaux, France. Retrieved May 25, 2011.
  78. Jump up^ "EFF DES Cracker Source Code". Cosic.esat.kuleuven.be. Retrieved 2011-07-08.
  79. Jump up^ "Disarmament Diplomacy: – DOE Supercomputing & Test Simulation Programme". Acronym.org.uk. 2000-08-22. Retrieved 2011-07-08.
  80. Jump up^ "China’s Investment in GPU Supercomputing Begins to Pay Off Big Time!". Blogs.nvidia.com. Retrieved 2011-07-08.
  81. Jump up^ Kaku, Michio. Physics of the Future (New York: Doubleday, 2011), 65.
  82. Jump up^ "Faster Supercomputers Aiding Weather Forecasts". News.nationalgeographic.com. 2010-10-28. Retrieved 2011-07-08.
  83. Jump up^ Washington Post August 8, 2011[dead link]
  84. Jump up^ Kan Michael (2012-10-31). "China is building a 100-petaflop supercomputer, InfoWorld, October 31, 2012". infoworld.com. Retrieved 2012-10-31.
  85. Jump up^ Agam Shah (2011-06-20). "SGI, Intel plan to speed supercomputers 500 times by 2018, ComputerWorld, June 20, 2011". Computerworld.com. Retrieved 2012-06-09.
  86. Jump up^ Dillow Clay (2012-09-18). "India Aims To Take The "World's Fastest Supercomputer" Crown By 2017, POPSCI, September 9, 2012". popsci.com. Retrieved 2012-10-31.
  87. Jump up^ DeBenedictis, Erik P. (2005). "Reversible logic for supercomputing". Proceedings of the 2nd conference on Computing frontiers. pp. 391–402. ISBN 1-59593-019-1.
  88. Jump up^ "IDF: Intel says Moore's Law holds until 2029". Heise Online. 2008-04-04.


The TOP500 project ranks and details the 500 most powerful (non-distributed) computer systems in the world. The project was started in 1993 and publishes an updated list of the supercomputers twice a year. The first of these updates always coincides with the International Supercomputing Conference in June, and the second one is presented in November at theACM/IEEE Supercomputing Conference. The project aims to provide a reliable basis for tracking and detecting trends in high-performance computing and bases rankings on HPL, a portable implementation of the high-performance LINPACK benchmark written in Fortran for distributed-memory computers.

The TOP500 list is compiled by Hans Meuer of the University of Mannheim,Germany, Jack Dongarra of the University of Tennessee, Knoxville, and Erich Strohmaier and Horst Simon of NERSC/Lawrence Berkeley National Laboratory.

History

Rapid growth of supercomputers performance, based on data from top500.org site. The logarithmic y-axis shows performance in GFLOPS. The red line denotes the fastest supercomputer in the world at the time. The yellow line denotes supercomputer no. 500 on TOP500 list. The dark blue line denotes the total combined performance of supercomputers on TOP500 list.

In the early 1990s, a new definition of supercomputer was needed to produce meaningful statistics. After experimenting with metrics based on processor count in 1992, the idea was born at the University of Mannheim to use a detailed listing of installed systems as the basis. Early 1993 Jack Dongarrawas persuaded to join the project with his LINPACK benchmark. A first test version was produced in May 1993, partially based on data available on the Internet, including the following sources:

The information from those sources was used for the first two lists. Since June 1993, the TOP500 is produced bi-annually based on site and vendor submissions only.

Since 1993, performance of the #1 ranked position has steadily grown in agreement with Moore's law, doubling roughly every 14 months. As of June 2013, the fastest system, the Tianhe-2 with Rpeak of 54.9024 PFlop/s, is over 419,102 times faster than the fastest system in November 1993, the Connection Machine CM-5/1024 (1024 cores) with Rpeak of 131.0 GFlop/s.

Architecture and operating systems

As of November 2013, TOP500 supercomputers are overwhelmingly based on x86-64 CPUs (Intel EMT64 and AMD AMD64 instruction set architecture), with the RISC-based Power Architecture used by IBM POWER microprocessors, and SPARC making up the remainder. Prior to the ascendance of 32-bit x86 and later 64-bit x86-64 in the early 2000s, a variety of RISC processor families made up the majority of TOP500 supercomputers, including RISC architectures such as SPARC, MIPS, PA-RISC and Alpha.

Share of processor architecture families in TOP500 supercomputers by time trend.

Top 10 ranking

Top 10 positions of the 42nd TOP500 on November 18, 2013
Rank Rmax
Rpeak
(Pflops)
Name Computer design
Processor type, interconnect
Vendor Site
Country, year
Operating system
1 33.863
54.902
Tianhe-2 NUDT
Xeon E5–2692 + Xeon Phi 31S1P, TH Express-2
NUDT National Supercomputing Center in Guangzhou
  China, 2013
Linux (Kylin)
2 17.590
27.113
Titan Cray XK7
Opteron 6274 + Tesla K20X, Cray Gemini Interconnect
Cray Oak Ridge National Laboratory
  United States, 2012
Linux (CLE, SLESbased)
3 17.173
20.133
Sequoia Blue Gene/Q
PowerPC A2, Custom
IBM Lawrence Livermore National Laboratory
  United States, 2013
Linux (RHEL andCNK)
4 10.510
11.280
K computer RIKEN
SPARC64 VIIIfx, Tofu
Fujitsu RIKEN
  Japan, 2011
Linux
5 8.586
10.066
Mira Blue Gene/Q
PowerPC A2, Custom
IBM Argonne National Laboratory
  United States, 2013
Linux (RHEL andCNK)
6 6.271
7.779
Piz Daint Cray XC30
Xeon E5–2670 + Tesla K20X, Aries
Cray Inc. Swiss National Supercomputing Centre
   Switzerland, 2013
Linux (CLE)
7 5.168
8.520
Stampede PowerEdge C8220
Xeon E5–2680 + Xeon Phi, Infiniband
Dell Texas Advanced Computing Center
  United States, 2013
Linux
8 5.008
5.872
JUQUEEN Blue Gene/Q
PowerPC A2, Custom
IBM Forschungszentrum Jülich
  Germany, 2013
Linux (RHEL andCNK)
9 4.293
5.033
Vulcan Blue Gene/Q
PowerPC A2, Custom
IBM Lawrence Livermore National Laboratory
  United States, 2013
Linux (RHEL andCNK)
10 2.897
3.185
SuperMUC iDataPlex DX360M4
Xeon E5–2680, Infiniband
IBM Leibniz-Rechenzentrum
  Germany, 2012
Linux

Legend:

Other rankings

Top countries

Numbers below represent the number of computers in the TOP500 that are in each of the listed countries.

Country Nov 2013 Jun 2013 Nov 2012 Jun 2012 Nov 2011 Jun 2011 Nov 2010 Jun 2010 Nov 2009 Jun 2009 Nov 2008 Jun 2008 Nov 2007
 United States 264 252 250 252 263 255 276 280 277 291 291 258 284
 China 63 66 72 68 74 61 41 25 21 21 15 12 10
 Japan 28 30 32 35 30 26 26 18 16 15 17 22 20
 United Kingdom 23 29 24 25 27 27 24 38 44 43 45 52 47
 France 22 23 21 22 23 25 25 29 26 23 26 34 17
 Germany 20 19 19 20 20 30 26 24 27 30 25 47 31
 India 12 11 8 5 2 2 4 5 3 6 8 6 9
 Canada 10 9 11 10 9 8 6 7 9 8 2 2 5
 Korea, South 5 4 4 3 3 4 3 1 2 1 1 1
 Sweden 5 7 6 4 3 5 6 8 7 10 8 9 7
 Russia 5 8 8 5 5 12 11 11 8 4 8 8 7
 Australia 5 5 7 6 4 6 4 1 1 1 1 1 1
 Italy 5 6 7 8 4 5 6 7 6 6 11 6 6
  Switzerland 5 4 4 1 3 4 4 5 5 4 4 6 7
 Norway 3 3 3 3 1 3 2 2 2 2 2 3
 Saudi Arabia 3 4 3 3 3 4 6 4 5 3
 Netherlands 3 2 1 2 4 3 3 3 5 6
 Brazil 3 3 2 3 2 2 2 1 1 2 1 1
 Spain 2 3 2 4 3 2 3 3 6 4 6 7 9
 Ireland 2 3 3 1 1 1 1 1 1 1
 Israel 2 2 1 3 3 2 1 2 1 1
 Finland 2 2 3 1 1 2 1 3 2 1 1 1 5
 Hong Kong 2 1 1 1 1
 Poland 2 3 4 5 6 5 6 5 3 4 6 3 1
 Belgium 1 1 1 2 1 2 2 1 1 2 2 1
 Austria 1 1 1 1 2 2 1 2 8 5
 Denmark 1 1 1 1 2 2 2 3 3 1
 Taiwan 1 1 3 3 2 2 1 2 3 11
 Mexico 1 1 1
 Slovak Republic 1 1
 Singapore 1 1 2 2 1 1 1 1
 South Africa 1 1 1 1 1
 United Arab Emirates 1
 New Zealand 5 7 8 6 4 6 1
 Slovenia 1 1 1 1 1 1
 Turkey 1 1
 Bulgaria 1 1 1
 Malaysia 1 1 1 2 3
 Cyprus 1 1
 Egypt 1 1
 Indonesia 1
 Luxembourg 1

Systems ranked #1 since 1993

Number of systems

By number of systems as of November 2013:

Top processor generation
Top vendors
Operating system family

Large machines not on the list

A few machines that have not been benchmarked are not eligible for the list: such as NCSA's Blue Waters. Additionally purpose-built machines that are not capable or do not run the benchmark are not included: such as RIKEN MDGRAPE-3.

See also[edit]

Portal icon Computer Science portal
Portal icon Information technology portal

References

  1. Jump up^ HPL
  2. Jump up^ AN INTERVIEW WITH JACK DONGARRA by Alan Beck, editor in chief HPCwire[dead link]
  3. ^ Jump up to:a b Statistics on Manufacturers and Continents
  4. Jump up^ List of the World's Most Powerful Computing Sites
  5. Jump up^ thefreelibrary.com
  6. Jump up^ Rpeak – This is the theoretical peak performance of the system. Measured in Pflops.
  7. Jump up^ TOP500 - Sublist Generator
  8. Jump up^ TOP500 - List Statistics

External links

China Still Has The World's Fastest Supercomputer

Earlier this week, the Top500 organization announced its semi-annual list of the Top 500 supercomputers in the world. And for the second year in a row, China’s Tianhe-2 is the world’s fastest by a long shot, maintaining its performance of 33.86 petaflop/s (quadrillions of calculations per second) on the standardized benchmark attached to every supercomputer on the list.

The Tiahne-2 supercomputer. (Credit: National University of Defense Technology)The Tiahne-2 supercomputer. (Credit: National University of Defense Technology)

In fact, the top 5 fastest supercomputers on the November 2013 list are the same as the top 5 fastest supercomputers on the June 2013 list. The second fastest supercomputer in the world is Cray’s Titan supercomputer at the Oak Ridge National Laboratory.

Rounding out the top five are IBM’s Sequoia supercomputer, RIKEN’s K Computer in Japan, and IBM’s Mira supercomputer at the Argonne National Laboratory.

The stability of the the top five is unique in the past few years, which has seen several different computers being named the fastest, while others moved down and up the ranks.

The Tianhe-2 was built by the National University of Defense Technology in China. It has a total of 3,120,000 Intel processing cores, but also features a number of Chinese built components and runs on a version of Linux called Kylin, which was natively developed by the NUDT.

The newest entry into the top ten supercomputer list is number 6 on the list, Piz Daint, a Cray system that has been installed at Swiss National Supercomputing Centre. Piz Daint is now the fastest supercomputer in Europe. It’s also the most energy efficient system in the Top 10. That’s something to note, because one of the biggest constraints on supercomputing is the sheer amount of power needed to operate the systems.

“In the top 10, computers might hit a high mark,” Kai Dupke, a senior product manager at SUSE Linux told me in a conversation about the supercomputer race. “But in the usual operation, they operate more slowly simply because it’s too expensive to run on full speed.”

Although China is still home to the world’s fastest supercomputer, the United States continues to be the leader in high performance computing. In June, the United States had 253 of the top 500 fastest computers. In the November list, it has 265. China, on the other hand, has 63 computers on the list, down from the 65 it had in June.

Follow me on Twitter or Facebook. Read my Forbes blog here.

That new supercomputer is not your friend

China reclaims the fastest computer in the world prize. Get ready for even better surveillance

 



That new supercomputer is not your friend
Chinese researchers test the supercomputer Tianhe No.1 at the National University of Defense Technology in Changsha city, central Chinas Hunan province, 27 October 2009. (Credit: AP)

We learned this week that China has the fastest supercomputer in the world, by a long shot. The Tianhe-2 is almost twice as speedy as the previous record holder, a U.S.-made Cray Titan.

Such news, by itself, isn’t particularly amazing. It’s not even the first time a Chinese supercomputer has held the top ranking. The Tianhe-1 grabbed the pole position in November 2010 and held it until June 2011. Previously, Japan and the United States had traded places since 1993. Supercomputing speed follows roughly the same trajectory as Moore’s Law — it doubles about every 14 months. There will always be new contenders for the throne.

But this month, there’s a new context for news about the debut of ever more powerful supercomputers. Consider the first comment left on Reddit to a thread announcing the exploits of Tianhe-2:

This would be a pretty awesome tool for churning through millions of phone records and digital copies of people’s online data.

Haha. Funny. But not really. The not-so-subtle implication of the constant expansion of supercomputing capacity is that, to borrow a quote from Intel’s Raj Hazra that appeared in the New York Times, “the insatiable need for computing is driving this.” Big Data wants Big Computers.

 

Cyber Espionage: The Chinese Threat
Experts at the highest levels of government say it's the biggest threat facing American business today. Hackers are stealing valuable trade secrets, intellectual property and confidential business strategies.

Government officials are calling it the biggest threat to America's economic security. Cyber spies hacking into U.S. corporations' computer networks are stealing valuable trade secrets, intellectual property data and confidential business strategies. The biggest aggressor? China. CNBC's David Faber investigates this new wave of espionage, which experts say amounts to the largest transfer of wealth ever seen —draining America of its competitive advantage and its economic edge. Unless corporate America wakes up and builds an adequate defense strategy, experts say it may be too late.

The Danger of Mixing Cyberespionage With Cyberwarfare

China has recently been accused of intense spying activity in cyberspace, following claims that the country uses cyber tactics to gain access to military and technological secrets held by both foreign states and corporations. In this context, the rhetoric of cyberwar has also raised its head. Experts are questioning whether we are already at war with China.

But a danger lies at the heart of the cyberwar rhetoric. Declaring war, even cyberwar, has always had serious consequences. Since war is acknowledged as the most severe threat to the survival and well-being of society, war rhetoric easily feeds an atmosphere of fear, provokes a rise in the emergency level and instigates counter-measures. It may also lead to an intensified cyber arms race.

It needs to be remembered that cyber espionage does not equate to cyber warfare - and in fact making such a link is completely unjustifiable.

Cyber espionage is an activity multiple actors resort to in the name of security, business, politics or technology. It is not inherently military. Espionage is finding information that ought to remain secret and as such, may be carried out for a range of different reasons and for varying lengths of time.

Cyber espionage is an activity multiple actors resort to in the name of security, business, politics or technology. It is not inherently military. Espionage is finding information that ought to remain secret and as such, may be carried out for a range of different reasons and for varying lengths of time.

Conducting effective cyber espionage campaigns may in fact take years, and results can be uncertain. However engaging in long-lasting cyberwar activity is simply not sustainable as the costs of warfare are high, and the motives very different; gaining information is never the prime motive. War is waged in order to re-engineer the opposing society to support one’s interests and values. This holds true for cyberwar as well.

But that's not to say that the two concepts aren't inter-linked. Cyber espionage can be utilised in warfare for preparing for war, as part of intelligence efforts, and for preparing for peace. Plus, a long-lasting spying campaign that eventually becomes detected may lead to war if it is interpreted to justify pre-emptive or preventive actions.

However, the probability of cyberwar in the near future is low. What we should be more concious of is the use of cyber in conventional conflicts. Cyber capabilities are now classified as weapons; they are a fifth dimension of warfare -- in addition to land, sea, air and space -- and whilst it’s unlikely future battles will be completely online, it is difficult to imagine future wars or conflicts without cyber activities.

But even if the concept of war has become vague, espionage should be regarded as a seperate activity in its own right. For actions to qualify as war they should cause massive human loss and material damage -- cyberespionage does not do this. For the majority, there is little to be gained from confusing cyberespionage with cyberwarfare, yet the potential losses in the form of increasingly restricted freedom and curtailed private space may be substantial.


The continued rapid growth of processing power brings with it enormous privacy implications.Applications such as facial recognition require massive computing power. Things that aren’t quite possible today, like waving your phone at a stranger and identifying them with NSA-like precision, will be child’s play tomorrow. Encryption standards that we think can resist the toughest attacks right now may wilt before the power of what’s around the corner. I used to hear about new supercomputers and think about the inexorable march forward into the science fiction future. Now I just think, oh great: the architecture of surveillance just got more fortified.

Read more: http://insights.wired.com/profiles/blogs/the-danger-of-mixing-cyberespionage-with-cyberwarfare#ixzz30BOzO65c 
Follow us: @Wiredinsights on Twitter | InnovationInsights on Facebook

NSA Snooping Was Only the Beginning. Meet the Spy Chief Leading Us Into Cyberwar

THE SECRET WAR

INFILTRATION. SABOTAGE. MAYHEM. FOR YEARS, FOUR-STAR GENERAL KEITH ALEXANDER HAS BEEN BUILDING A SECRET ARMY CAPABLE OF LAUNCHING DEVASTATING CYBERATTACKS. NOW IT’S READY TO UNLEASH HELL.

INSIDE FORT MEADE, Maryland, a top-secret city bustles. Tens of thousands of people move through more than 50 buildings—the city has its own post office, fire department, and police force. But as if designed by Kafka, it sits among a forest of trees, surrounded by electrified fences and heavily armed guards, protected by antitank barriers, monitored by sensitive motion detectors, and watched by rotating cameras. To block any telltale electromagnetic signals from escaping, the inner walls of the buildings are wrapped in protective copper shielding and the one-way windows are embedded with a fine copper mesh.

This is the undisputed domain of General Keith Alexander, a man few even in Washington would likely recognize. Never before has anyone in America’s intelligence sphere come close to his degree of power, the number of people under his command, the expanse of his rule, the length of his reign, or the depth of his secrecy. A four-star Army general, his authority extends across three domains: He is director of the world’s largest intelligence service, the National Security Agency; chief of the Central Security Service; and commander of the US Cyber Command. As such, he has his own secret military, presiding over the Navy’s 10th Fleet, the 24th Air Force, and the Second Army.

Alexander runs the nation’s cyberwar efforts, an empire he has built over the past eight years by insisting that the US’s inherent vulnerability to digital attacks requires him to amass more and more authority over the data zipping around the globe. In his telling, the threat is so mind-bogglingly huge that the nation has little option but to eventually put the entire civilian Internet under his protection, requiring tweets and emails to pass through his filters, and putting the kill switch under the government’s forefinger. “What we see is an increasing level of activity on the networks,” he said at a recent security conference in Canada. “I am concerned that this is going to break a threshold where the private sector can no longer handle it and the government is going to have to step in.”

In its tightly controlled public relations, the NSA has focused attention on the threat of cyberattack against the US—the vulnerability of critical infrastructure like power plants and water systems, the susceptibility of the military’s command and control structure, the dependence of the economy on the Internet’s smooth functioning. Defense against these threats was the paramount mission trumpeted by NSA brass at congressional hearings and hashed over at security conferences.

But there is a flip side to this equation that is rarely mentioned: The military has for years been developing offensive capabilities, giving it the power not just to defend the US but to assail its foes. Using so-called cyber-kinetic attacks, Alexander and his forces now have the capability to physically destroy an adversary’s equipment and infrastructure, and potentially even to kill. Alexander—who declined to be interviewed for this article—has concluded that such cyberweapons are as crucial to 21st-century warfare as nuclear arms were in the 20th.

And he and his cyberwarriors have already launched their first attack. The cyberweapon that came to be known as Stuxnet was created and built by the NSA in partnership with the CIA and Israeli intelligence in the mid-2000s. The first known piece of malware designed to destroy physical equipment, Stuxnet was aimed at Iran’s nuclear facility in Natanz. By surreptitiously taking control of an industrial control link known as a Scada (Supervisory Control and Data Acquisition) system, the sophisticated worm was able to damage about a thousand centrifuges used to enrich nuclear material.

The success of this sabotage came to light only in June 2010, when the malware spread to outside computers. It was spotted by independent security researchers, who identified telltale signs that the worm was the work of thousands of hours of professional development. Despite headlines around the globe, officials in Washington have never openly acknowledged that the US was behind the attack. It wasn’t until 2012 that anonymous sources within the Obama administration took credit for it in interviews with The New York Times.

But Stuxnet is only the beginning. Alexander’s agency has recruited thousands of computer experts, hackers, and engineering PhDs to expand US offensive capabilities in the digital realm. The Pentagon has requested $4.7 billion for “cyberspace operations,” even as the budget of the CIA and other intelligence agencies could fall by $4.4 billion. It is pouring millions into cyberdefense contractors. And more attacks may be planned.

“WE JOKINGLY REFERRED TO HIM AS EMPEROR ALEXANDER, BECAUSE WHATEVER KEITH WANTS, KEITH GETS.”

Inside the government, the general is regarded with a mixture of respect and fear, not unlike J. Edgar Hoover, another security figure whose tenure spanned multiple presidencies. “We jokingly referred to him as Emperor Alexander—with good cause, because whatever Keith wants, Keith gets,” says one former senior CIA official who agreed to speak on condition of anonymity. “We would sit back literally in awe of what he was able to get from Congress, from the White House, and at the expense of everybody else.”

Now 61, Alexander has said he plans to retire in 2014; when he does step down he will leave behind an enduring legacy—a position of far-reaching authority and potentially Strangelovian powers at a time when the distinction between cyberwarfare and conventional warfare is beginning to blur. A recent Pentagon report made that point in dramatic terms. It recommended possible deterrents to a cyberattack on the US. Among the options: launching nuclear weapons.

AMERICAN EAGLE

FROM  http://www.wired.com/2013/06/general-keith-alexander-cyberwar/all/

 

THIS IS WHERE I POST WHAT I'M DOING AND THINKING

BLOG INDEX  2011

BLOG INDEX 2012  - page 1

JANUARY THRU APRIL 2012

 

BLOG INDEX 2012 - PAGE 2

MAY THRU AUGUST 2012

 

BLOG INDEX 2012 - PAGE 3

SEPTEMBER THRU DECEMBER

 

BLOG INDEX 2013

JAN, FEB, MAR, APR. 2013

 

BLOG INDEX - PAGE 2 - 2013

MAY, JUNE, JULY, AUGUST 2013

 

BLOG INDEX - PAGE 3 - 2013

SEPT, OCT, NOV, DEC, 2013

 

BLOG INDEX - PAGE 4 - 2014

JAN., FEB., MAR., APR. 2014