Sure, Fujitsu has a right to be unapproachable of a K supercomputer — behaving over 8 petaflops with usually underneath 70,000 Venus CPUs is zero to sneeze at. Intel isn’t giving up a standing as a supercomputing CPU king, however, as it skeleton to move exascale computing to a universe by a finish of this decade. Such a appurtenance could do a single million trillion calculations per second, as well as Intel skeleton to have it occur with a Many Integrated Core Architecture (MIC). The initial CPUs written with MIC, codenamed Knights Corner, have been built upon a 22nm routine that utilizes a company’s 3D Tri-Gate transistors as well as packs over 50 cores per chip. These CPUs have been written for together estimate applications, identical to a NVIDIA GPUs that will be used in a DARPA-funded supercomputer you schooled about final year. Here you suspicion a fight in between these dual was over — looks similar to a ultimate one’s usually removing started. PR’s after a break. Show full PR textIntel Equipped to Lead Industry to Era of Exascale Computing
Intel® Many Integrated Core (Intel® MIC) Architecture Shows Strength as Critical Component of Intel’s Exascale Computing Solution
NEWS HIGHLIGHTS
With partnership partners, Intel aims to broach finish record resolution for exascale opening by a finish of a decade.
Demonstrations from pass supercomputing centers such as Forschungszentrum Juelich, Leibniz Supercomputing Centre, CERN as well as KISTI underscore movement of Intel® Many Integrated Core Architecture.
Intel processor-powered supercomputers have up 77 percent of a ultimate TOP500 list of supercomputers as well as scarcely 90 percent of all ultimate systems in 2011.
International Supercomputing Conference 2011
SANTA CLARA, Calif. & HAMBURG, Germany–(BUSINESS WIRE)–At a International Supercomputing Conference (ISC), Kirk Skaugen, Intel Corporation clamp boss as well as ubiquitous physical education instructor of a Data Center Group, summarized a company’s prophesy to grasp ExaFLOP/s opening by a finish of this decade. An ExaFLOP/s is quintillion mechanism operations per second, hundreds times some-more than today’s fastest supercomputers.
“Intel is singly versed with forlorn prolongation technologies, ultimate design innovations as well as a informed program programming sourroundings that will move us closer to this sparkling exascale goal.”
Reaching exascale levels of opening in a destiny will not usually need a total efforts of attention as well as governments, yet additionally approaches being pioneered by a Intel® Many Integrated Core (Intel® MIC) Architecture, according to Skaugen. Managing a bomb expansion in a volume of interpretation common opposite a Internet, anticipating solutions to meridian change, handling a flourishing costs of accessing resources such as oil as well as gas, as well as a crowd of alternative hurdles need increasing amounts of computing resources that usually increasingly high-performing supercomputers can address.
“While Intel® Xeon® processors have been a transparent design of preference for a stream TOP500 list of supercomputers, Intel is serve expanding a concentration upon high-performance computing by enabling a attention for a subsequent limit with a Many Integrated Core design for petascale as well as destiny exascale workloads,” pronounced Skaugen. “Intel is singly versed with forlorn prolongation technologies, ultimate design innovations as well as a informed program programming sourroundings that will move us closer to this sparkling exascale goal.”
Paving a Way to Exaflop Performance
Intel’s relentless office of Moore’s Law — doubling a transistor firmness upon microprocessors rounded off each 2 years to enlarge functionality as well as opening whilst dwindling costs — total with an innovative, rarely fit program programming indication as well as impassioned complement scalability were remarkable by Skaugen as pass mixture for channel a starting indicate of petascale computing in to a ultimate epoch of exascale computing. With this enlarge in performance, though, comes a poignant enlarge in appetite consumption.
As an example, for today’s fastest supercomputer in China, a Tianhe-1A, to grasp exascale performance, it would need some-more than 1.6 GW of appetite – an volume vast sufficient to supply physical phenomenon to 2 million homes – to illustrate presenting an appetite potency challenge.
To residence this challenge, Intel as well as European researchers have determined 3 European labs with 3 categorical goals: to emanate a postulated partner participation in Europe; take worth of a flourishing aptitude of European high-performance computing (HPC) research; as well as exponentially grow capabilities in computational science, engineering as well as vital computing. One of a technical goals of these labs is to emanate make-believe applications that proceed to residence a appetite potency hurdles of relocating to exascale performance.
Skaugen pronounced there is a intensity for extensive expansion of a HPC market. While supercomputers from a 1980s delivered GigaFLOP/s (billions of floating indicate operations per second) performance, today’s many absolute machines have increasing this worth by multiform million times. This, in turn, has increasing a direct for processors used in supercomputing. By 2013 Intel expects a tip 100 supercomputers in a universe to operate a single million processors. By 2015 this array is approaching to double, as well as is forecasted to strech 8 million units by a finish of a decade. The opening of a TOP500 #1 complement is estimated to strech 100 PetaFLOP/s in 2015 as well as mangle a separator of 1 ExaFLOP/s in 2018. By a finish of a decade a fastest complement upon Earth is forecasted to be means to yield opening of some-more than 4 ExaFLOP/s.
Intel MIC Architecture Software Development Momentum
The Intel MIC design is a pass further to a company’s existent products, together with Intel Xeon processors, as well as approaching to assistance lead a attention in to a epoch of exascale computing. The initial Intel MIC product, codenamed “Knights Corner,” is programmed for prolongation upon Intel’s 22-nanometer record that featuring innovative 3-D Tri-Gate transistors1. Intel is right away shipping Intel MIC program growth platforms, codenamed “Knights Ferry,” to name growth partners.
At ISC, Intel as well as a little of a partners together with Forschungszentrum Juelich, Leibniz Supercomputing Centre (LRZ), CERN as well as Korea Institute of Science as well as Technology Information (KISTI) showed early formula of their work with a “Knights Ferry” platform. The demonstrations showed how Intel MIC design delivers both opening as well as program programmability advantages.
“The programming indication worth of Intel MIC design enabled us to fast scale a applications using upon Intel Xeon processors to a Knights Ferry Software Development Platform,” pronounced Prof. Arndt Bode of a Leibniz Supercomputing Centre. “This effort was creatively grown as well as optimized for Intel Xeon processors yet due to a laxity of a programming indication you could optimize a formula for a Intel MIC design inside of hours as well as additionally completed over 650 GFLOPS of performance.”
Intel additionally showed server as well as workstation platforms from SGI, Dell, HP, IBM, Colfax as well as Supermicro, all of that have been operative with Intel to devise products formed upon “Knights Corner.”
“SGI recognizes a stress of inter-processor communications, power, firmness as well as usability when architecting for exascale,” pronounced SGI CTO Dr. Eng Lim Goh. “The Intel MIC products will prove all 4 of these priorities, generally with their expected enlarge in discriminate firmness joined with informed X86 programming environment.”
TOP500 Supercomputers
The 37th book of a Top500 list, that was voiced during ISC, shows that Intel continues to be a force in high-performance computing, with 387 systems or some-more than 77 percent, powered by Intel processors. Out of all ultimate systems to a list in 2011, Intel powered systems accounted for tighten to 90 percent. More than half of these ultimate additions have been formed upon ultimate 32nm Intel Xeon 5600 array processors that right away alone appetite some-more than 35% of all systems in TOP500 list, 3 times a volume comparing to final year. Intel additionally powers 5 systems in tip 10.
The semi-annual TOP500 list of supercomputers is a work of Hans Meuer of a University of Mannheim, Erich Strohmaier as well as Horst Simon of a U.S. Department of Energy’s National Energy Research Scientific Computing Center, as well as Jack Dongarra of a University of Tennessee. The finish inform is accessible during www.top500.org.
More report upon ISC’11 together with minute outline of showcased Intel MIC architecture-based demos as well as performance, Skaugen’s display as well as cinema have been accessible during http://newsroom.intel.com/docs/DOC-2152.
