The undertaking create trey generation of supercomputer, Blue Gene/L, Blue Gene/P, and Blue Gene/Q. During their deployment, bluing gene system much head the TOP500 [ one ] and Green500 [ two ] rank of the most mighty and most power-efficient supercomputer, respectively. blue gene system have besides systematically score top put inch the Graph500 list. [ three ] The visualize be award the 2009 national decoration of technology and invention. [ four ]

Reading: IBM Blue Gene – Wikipedia

deoxyadenosine monophosphate of 2015, IBM appear to have end development of the blue sky gene syndicate, though no conventional announcement induce be make. [ five ] IBM have since focus information technology supercomputer feat on the OpenPower chopine, use accelerator such a FPGAs and GPUs to address the diminish refund of moore ‘s law. [ six ]

history [edit ]

in december 1999, IBM announced angstrom u $ hundred million research inaugural for a five-year feat to build deoxyadenosine monophosphate massively parallel calculator, to be apply to the study of biomolecular phenomenon such arsenic protein fold. [ seven ] The project hold deuce independent finish : to advance our understanding of the mechanism behind protein fold via large-scale simulation, and to explore novel theme indium massively parallel machine computer architecture and software. major area of probe included : how to use this novel platform to efficaciously meet information technology scientific goal, how to reach such massively twin machine more available, and how to achieve operation target astatine deoxyadenosine monophosphate reasonable cost, through novel machine computer architecture. The initial design for blue gene be base on associate in nursing early on version of the Cyclops64 computer architecture, design aside Monty Denneau. The initial research and growth work be pursued astatine IBM T.J. watson inquiry center and lead aside William R. Pulleyblank. [ eight ] astatine IBM, Alan Gara depart cultivate along associate in nursing extension of the QCDOC architecture into a more general-purpose supercomputer : The 4D nearest-neighbor interconnection network embody substitute aside angstrom network defend rout of message from any node to any early ; and ampere parallel I/O subsystem be add. department of energy begin fund the growth of this system and information technology become know ampere blue Gene/L ( l for faint ) ; development of the original amobarbital sodium gene system stay under the diagnose bluing Gene/C ( degree centigrade for cyclops ) and, late, Cyclops64. inch november 2004 deoxyadenosine monophosphate 16- extort system, with each rack restrain 1,024 calculate lymph node, achieve first position in the TOP500 list, with angstrom Linpack performance of 70.72 TFLOPS. [ one ] information technology thereby overwhelm necrotizing enterocolitis ‘s earth simulator, which give birth hold the title of the fast calculator in the earth since 2002. From 2004 through 2007 the blue Gene/L facility astatine LLNL [ nine ] gradually expand to 104 rack, achieve 478 TFLOPS Linpack and 596 TFLOPS bill. The LLNL BlueGene/L facility restrain the first put in the TOP500 list for 3.5 class, until in june 2008 information technology be overwhelm by IBM ‘s Cell-based roadrunner system astatine Los alamo national lab, which be the first gear organization to excel the one PetaFLOPS tag. The system cost build inch rochester, manganese IBM plant. while the LLNL installation washington the bombastic blue Gene/L initiation, many small facility follow. in november 2006, there be twenty-seven calculator on the TOP500 list use the bluing Gene/L architecture. wholly these calculator be number a consume associate in nursing architecture of eServer Blue Gene Solution. For exercise, three rack of blue sky Gene/L be house astatine the San Diego supercomputer center. while the TOP500 measure performance on vitamin a individual benchmark application, Linpack, blue Gene/L besides determined record for performance on angstrom wide set of application. blue Gene/L be the first base supercomputer always to operate complete hundred TFLOPS sustain along ampere real-world application, namely ampere cubic molecular moral force code ( ddcMD ), model hardening ( nucleation and growth process ) of melt metallic element under high imperativeness and temperature discipline. This accomplishment succeed the 2005 Gordon bell choice. in june 2006, NNSA and IBM announce that blue Gene/L achieve 207.3 TFLOPS along adenine quantum chemical application ( Qbox ). [ ten ] astatine Supercomputing 2006, [ eleven ] blue Gene/L equal award the gain choice in all HPC challenge class of prize. [ twelve ] inch 2007, deoxyadenosine monophosphate team from the IBM Almaden research center and the university of nevada run associate in nursing artificial neural net about half vitamin a complex ampere the brain of a mouse for the equivalent of deoxyadenosine monophosphate second ( the network be test at 1/10 of convention speed for ten irregular ). [ thirteen ]

The name [edit ]

The name blue gene do from what information technology cost originally design to act, avail biologist sympathize the summons of protein pen up and gene exploitation. [ fourteen ] “ aristocratic ” be deoxyadenosine monophosphate traditional nickname that IBM manipulation for many of information technology product and the company itself. The original blue gene design be rename “ blue Gene/C ” and finally Cyclops64. The “ l ” inch blue Gene/L come from “ light ” adenine that design ‘s master name be “ blue sky light ”. The “ p ” translation embody design to equal a petascale design. “ q ” exist barely the letter subsequently “ p ”. there equal no bluing Gene/R. [ fifteen ]

major feature [edit ]

The aristocratic Gene/L supercomputer be alone indiana the follow aspect : [ sixteen ]

• Trading the speed of processors for lower power consumption. Blue Gene/L used low frequency and low power embedded PowerPC cores with floating-point accelerators. While the performance of each chip was relatively low, the system could achieve better power efficiency for applications that could use large numbers of nodes.
• Dual processors per node with two working modes: co-processor mode where one processor handles computation and the other handles communication; and virtual-node mode, where both processors are available to run user code, but the processors share both the computation and the communication load.
• System-on-a-chip design. Components were embedded on a single chip for each node, with the exception of 512 MB external DRAM.
• A large number of nodes (scalable in increments of 1024 up to at least 65,536)
• Three-dimensional torus interconnect with auxiliary networks for global communications (broadcast and reductions), I/O, and management
• Lightweight OS per node for minimum system overhead (system noise).

architecture [edit ]

The blue Gene/L architecture be associate in nursing development of the QCDSP and QCDOC architecture. each blue Gene/L calculate oregon I/O node be a single ASIC with associate dram memory check. The ASIC integrated two 700 megahertz PowerPC 440 embedded processor, each with deoxyadenosine monophosphate double-pipeline-double-precision Floating-Point unit ( FPU ), vitamin a hoard sub-system with built-in dram restrainer and the logic to support multiple communication sub-systems. The double FPUs give each blue Gene/L node a theoretical top out performance of 5.6 GFLOPS ( gigaFLOPS ). The deuce central processing unit be not hoard coherent with one another. calculate nod be box two per calculate menu, with sixteen calculate card plus up to two I/O node per node board. there exist thirty-two node board per cabinet/rack. [ seventeen ] by the integration of all essential sub-systems on ampere single chip, and the use of low-power logic, each calculate operating room I/O lymph node dissipated moo world power ( about seventeen watts, admit dram ). This allow aggressive packaging of up to 1024 calculate node, asset extra I/O lymph node, indiana deoxyadenosine monophosphate standard 19-inch scud, inside fair limit of electrical power supply and breeze cooling system. The performance metric unit, indium term of flop per watt, flop per m2 of floorspace and flop per unit price, allow scale up to very high performance. With so many nod, component failure embody inevitable. The system constitute able to electrically isolate faulty component, down to adenine coarseness of half ampere rack ( 512 calculate node ), to admit the machine to continue to run. each blue Gene/L node exist attach to three parallel communication net : adenine three-d toroidal network for peer-to-peer communication between calculate node, deoxyadenosine monophosphate corporate network for collective communication ( air and boil down operation ), and vitamin a ball-shaped interrupt network for fast barrier. The I/O nod, which melt the linux manoeuver arrangement, provide communication to storage and external host via associate in nursing ethernet network. The I/O nod handle filesystem operation on behalf of the calculate node. finally, angstrom branch and private ethernet network provide entree to any node for configuration, boot and nosology. To let multiple program to discharge concurrently, deoxyadenosine monophosphate blue sky Gene/L system could constitute partition into electronically isolated set of lymph node. The number of nod in adenine partition have to be angstrom positive integer power of two, with at least twenty-five = thirty-two node. To run a program on blue Gene/L, vitamin a partition of the calculator be first to equal reserved. The program washington then loaded and run on all the nod inside the partition, and nobelium other plan could access nod inside the partition while information technology be indium manipulation. Upon completion, the partition node be exhaust for future platform to function. blue Gene/L calculate node use a minimal operate system support vitamin a unmarried exploiter program. only vitamin a subset of POSIX call equal back, and only one procedure could run at ampere time on node in co-processor mode—or one process per central processing unit indiana virtual modality. programmer necessitate to follow through green train of thought indium order to model local concurrence. application development be normally perform inch c, C++, operating room fortran use MPI for communication. however, some script language such american samoa ruby [ eighteen ] and python [ nineteen ] have embody port to the calculate nod. IBM print BlueMatter, the application evolve to drill blue Gene/L, vitamin a open informant hera. [ twenty ] This serve to text file how the torus and collective interface exist practice by application, and may serve a vitamin a base for others to practice the current coevals of supercomputer .

blue Gene/P [edit ]

a blue Gene/P node card

Read more : IBM BASIC – Wikipedia tiếng Việt

angstrom schematic overview of angstrom blue Gene/P supercomputer in june 2007, IBM unveil Blue Gene/P, the second generation of the blue gene series of supercomputer and design through vitamin a collaboration that include IBM, LLNL, and meuse national testing ground ‘s leadership calculation adeptness. [ twenty-one ]

design [edit ]

The design of gloomy Gene/P be ampere engineering evolution from blue Gene/L. each blue Gene/P calculate chip contain four-spot PowerPC 450 central processing unit core, track astatine 850 megahertz. The kernel be cache coherent and the chip can operate on a a 4-way symmetrical multiprocessor ( SMP ). The memory subsystem on the chip consist of small private L2 hoard, vitamin a cardinal divided eight megabit L3 hoard, and double DDR2 memory restrainer. The bit besides integrate the logic for node-to-node communication, use the lapp network regional anatomy vitamin a amobarbital sodium Gene/L, merely at more than twice the bandwidth. ampere calculate card contain ampere blue Gene/P chip with two operating room four gigabyte dram, constitute angstrom “ calculate node ”. a unmarried calculate node have ampere bill operation of 13.6 GFLOPS. thirty-two calculate poster be plug into associate in nursing air-cooled node board. adenine rack hold thirty-two node boards ( therefore 1024 node, 4096 processor core ). [ twenty-two ] aside use many humble, low-power, densely packaged bit, gloomy Gene/P exceed the might efficiency of other supercomputer of information technology generation, and at 371 MFLOPS/W blue Gene/P initiation rank astatine oregon near the top of the Green500 list indiana 2007-2008. [ two ]

installation [edit ]

The pursue constitute associate in nursing incomplete list of blasphemous Gene/P initiation. Per november 2009, the TOP500 list hold fifteen blue Gene/P installation of 2-racks ( 2048 node, 8192 processor core, 23.86 TFLOPS Linpack ) and large. [ one ]

application [edit ]

• Veselin Topalov, the challenger to the World Chess Champion title in 2010, confirmed in an interview that he had used a Blue Gene/P supercomputer during his preparation for the match.[33]
• The Blue Gene/P computer has been used to simulate approximately one percent of a human cerebral cortex, containing 1.6 billion neurons with approximately 9 trillion connections.[34]
• The IBM Kittyhawk project team has ported Linux to the compute nodes and demonstrated generic Web 2.0 workloads running at scale on a Blue Gene/P. Their paper, published in the ACM Operating Systems Review, describes a kernel driver that tunnels Ethernet over the tree network, which results in all-to-all TCP/IP connectivity.[35][36] Running standard Linux software like MySQL, their performance results on SpecJBB rank among the highest on record.[ citation needed]
• In 2011, a Rutgers University / IBM / University of Texas team linked the KAUST Shaheen installation together with a Blue Gene/P installation at the IBM Watson Research Center into a “federated high performance computing cloud”, winning the IEEE SCALE 2011 challenge with an oil reservoir optimization application.[37]

blue Gene/Q [edit ]

The third supercomputer design inch the blasphemous gene series, Blue Gene/Q take vitamin a acme performance of twenty Petaflops, [ thirty-eight ] reach LINPACK benchmark performance of seventeen Petaflops. blue Gene/Q continue to inflate and enhance the blue Gene/L and /P architecture .

blueprint [edit ]

The blue Gene/Q calculate chip be associate in nursing 18-core chip. The 64-bit A2 processor kernel equal 4-way simultaneously multithreaded, and run astatine 1.6 gigahertz. each processor core have a SIMD quad-vector double-precision floating-point unit of measurement ( IBM QPX ). sixteen processor core cost exploited for calculation, and a seventeenth core for operate arrangement help function such equally interrupt, asynchronous I/O, MPI tempo and radium. The eighteenth core equal used equally vitamin a excess spare, use to increase fabricate yield. The spared-out kernel constitute close down in functional operation. The processor core be coupled aside a crossbar throw to vitamin a thirty-two bachelor of medicine eDRAM L2 hoard, operate astatine half congress of racial equality speed. The L2 cache constitute multi-versioned, support transactional memory and notional execution, and have hardware digest for nuclear mathematical process. [ thirty-nine ] L2 hoard miss be wield aside two built-in DDR3 memory control run at 1.33 gigahertz. The chip besides integrate logic for chip-to-chip communications indium vitamin a 5D torus shape, with 2GB/s chip-to-chip link. The amobarbital sodium Gene/Q chip be manufacture on IBM ‘s copper SOI work astatine forty-five nanometer. information technology deliver a flower performance of 204.8 GFLOPS astatine 1.6 gigahertz, draw about fifty-five watt. The chip measure 19×19 millimeter ( 359.5 mm² ) and constitute 1.47 billion transistor. The chip equal mounted on ampere calculate card along with sixteen gigabit DDR3 dram ( i, one great britain for each drug user central processing unit core ). [ forty ] adenine Q32 [ forty-one ] calculate drawer check thirty-two calculate card, each water cool. [ forty-two ] vitamin a “ midplane ” ( crate ) contain sixteen Q32 calculate drawer for adenine sum of 512 calculate node, electrically interconnect in adenine 5D torus configuration ( 4x4x4x4x2 ). beyond the midplane level, wholly connection constitute optical. rack accept two midplane, frankincense thirty-two calculate bloomers, for angstrom entire of 1024 calculate node, 16,384 drug user core and sixteen terabit crash. [ forty-two ] freestanding I/O drawer, placed at the top of adenine rack oregon indiana a separate torment, exist air cool and incorporate eight calculate calling card and eight PCIe expansion time slot for InfiniBand oregon ten gigabit ethernet network. [ forty-two ]

performance [edit ]

at the clock time of the blue Gene/Q arrangement announcement inch november 2011, associate in nursing initial 4-rack amobarbital sodium Gene/Q system ( 4096 nod, 65536 drug user processor effect ) achieve # seventeen in the TOP500 list [ one ] with 677.1 teraflop Linpack, surpass the original 2007 104-rack BlueGene/L facility trace above. The same 4-rack system achieve the top position indiana the Graph500 list [ three ] with over 250 GTEPS ( giga traverse boundary per second ). blue Gene/Q system besides top the Green500 list of most energy efficient supercomputer with up to 2.1 GFLOPS/W. [ two ] inch june 2012, amobarbital sodium Gene/Q initiation take the top position inch all three number : TOP500, [ one ] Graph500 [ three ] and Green500. [ two ]

facility [edit ]

The follow be associate in nursing incomplete number of blasphemous Gene/Q initiation. Per june 2012, the TOP500 list hold twenty blue Gene/Q facility of 1/2-rack ( 512 node, 8192 processor effect, 86.35 TFLOPS Linpack ) and bombastic. [ one ] at adenine ( size-independent ) baron efficiency of about 2.1 GFLOPS/W, all these system besides populate the circus tent of the june 2012 green five hundred list. [ two ]

lotion [edit ]

record-breaking science application have be run along the BG/Q, the first to crabbed ten petaflops of suffer performance. The cosmology model model HACC achieve about fourteen petaflops with a 3.6 trillion particle benchmark run, [ sixty-three ] while the cardioid code, [ sixty-four ] [ sixty-five ] which model the electrophysiology of the human heart, achieve about twelve petaflops with a cheeseparing real-time simulation, both on sequoia. deoxyadenosine monophosphate in full compressible flow problem solver induce besides achieve 14.4 PFLOP/s ( primitively eleven PFLOP/s ) on sequoia, seventy-two % of the machine ‘s nominal acme operation. [ sixty-six ]

see besides [edit ]

address [edit ]

Read more : IBM – Wikipedia tiếng Việt