1 / 42

C OMPUTER ARCHITE C T U R E (for Erasmus students)

C OMPUTER ARCHITE C T U R E (for Erasmus students) Asso c. Prof. Stasys Maciulevičius C omputer Dept. sta sys. ma ciulevicius @ktu.lt stasys@ecdl.lt Our course Presentations - ifko.ktu.l t /~stama/CompArch/CompArchE.html Final grade : Simulation of o pera t io n using Small (20%)

Melvin
Télécharger la présentation

C OMPUTER ARCHITE C T U R E (for Erasmus students)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. COMPUTER ARCHITECTURE(for Erasmus students) Assoc.Prof.Stasys Maciulevičius Computer Dept. stasys.maciulevicius@ktu.lt stasys@ecdl.lt

  2. Our course • Presentations - ifko.ktu.lt/~stama/CompArch/CompArchE.html • Final grade: • Simulation of operationusing Small (20%) • Abstract on computer architecture topics (20%) • Colloquium from 1stpart (processors)(30%) • Final exam from 2nd part (30%) ©S.Maciulevičius

  3. Coursware • J.L.Hennessy, D.A.Patterson. Computer Architecture: A Quantitative Approach. - Morgan Kaufman, San Mateo, CA, 1990, 1996, 2003 • W.Stallings. Computer Organization and Architecture: Designing for Performance. - Prentice Hall Int., 2000 • My presentations www.ifko.ktu.lt/~stama • Internet sites ©S.Maciulevičius

  4. Evolution ofcomputers • Increasing of performance • Decreasing of component size • Increasing of memory size • Increasing of input/output streams and bandwidth Important: balancing of performance between different components ©S.Maciulevičius

  5. We will study … • processors – main engine of computer • memory hierarchy • buses, connectingcomputer devices together • principles of input and output Looking through the prism of performance ©S.Maciulevičius

  6. Past and Present Time before emergence of computers: • abacus • 1642, B.Pascal -the first mechanical computing device • 1673, G.Leibnitz -mechanical computing machine • 1822,Ch.Babbage -difference engine • 1854,G.Boole-book "The Laws of Thought " ©S.Maciulevičius

  7. Past and Present Computer era begins • J.V.Atanasoff(USA, ABC computer ), K.Zuse(Germany, 1942), H.Aiken (USA, 1944) • John Eckert and John Mauchly invented the first general-purpose electronic digital computer (ENIAC - Electronic Numerical Integrator and Calculator • 1944 - John von Neumann described a computer architecture in which data and program memory are mapped into the same address space -“von Neumann architecture ” ©S.Maciulevičius

  8. Past and Present Computer-era mid • IBM invested 5 billion USD and in1964 anounced system IBM/360: • (allmost) same instruction set • (allmost) same operating system • increasing speed • increasing memory ©S.Maciulevičius

  9. Past and Present IBM System/360 ©S.Maciulevičius

  10. Past and Present Further.... • 1965 - DEC PDP-8 - first commercial minicomputer (round$20 000) • 1971–first microprocessor (Intel 4004) • 1973 -Gary Kildall presents operating system CP/M (Control Program/Monitor or Control Program for Microcomputer ). • 1981 IBM presentsIBM 5150 PC (4.77-MHz Intel 8088 CPU, 64KB RAM, 40KB ROM, one 5.25” floppy discand PC-DOS 1.0), beyond$3000 • 1985 m. Microsoft presents Microsoft Windows 1.0, beyond$100. etc. ©S.Maciulevičius

  11. Parametrs of somecomputers ©S.Maciulevičius

  12. Parametrs of somecomputers ©S.Maciulevičius

  13. Tendences Technology Transistor countin chip grows quickly – duplicates every 3 years Frequence grows relatively slowly ©S.Maciulevičius

  14. Von Neumann architecture • While consulting for the Moore School of Electrical Engineering on the EDVAC project, von Neumann wrote an incomplete set of notes titled the First Draft of a Report on the EDVAC • The paper described a computer architecture in which data and program memory are mapped into the same address space • This architecture became the de facto standard and can be contrasted with a so-called Harvard architecture, which has separate program and data memories on a separate bus ©S.Maciulevičius

  15. Von Neumann architecture ©S.Maciulevičius

  16. Instruction processor Data processor Input/ output processor Communication device (bus) Memory Von Neumann architecture Some another drawing: ©S.Maciulevičius

  17. Von Neumann architecture • Todayinstructionand data processors are combined into single unit – central processing unit (CPU). ©S.Maciulevičius

  18. data bus 1 addres bus RN AR AC T IR PC M R1 DR ALU data bus2 Instructionprocessor Data processor Structure of computer More mindful look intothe computer inside alowws us to present such a possible view: ©S.Maciulevičius

  19. Structure of computer Here are: • AC - accumulator – special register for storing (keeping) of data to be processed (operand) and operation result, • T - register for temporare storing of data (second operand), • IR- instruction register, • PC - program counter;it indicates address of instructionto be executed, • M - memory, • AR - memoryaddress register, • DR - memorydata register ©S.Maciulevičius

  20. Executing of instruction Process of instruction executingcan be described more detailed as follows: 1. Fetch instruction from memory and write into instruction register IR 2. Change content of PC 3. Determinetype of fetched instruction 4. If data to be processed are in memory, determinetheir location 5. If needed, select data from memoryand transfer them to CPU registers 6. Execute operation specified by instruction 7. Write results into appropriate place 8. Return back to 1ststep for execution of next instruction ©S.Maciulevičius

  21. Input device Main memory Output device Programand Initial data Results External memory Processor Structureof first generation computers Processor acts as centralcontrol device: CPU interprets program and executes instructions; manages the initial data input and output of results also ©S.Maciulevičius

  22. CPU Main memory Console Output module Input module External memory Common Bus Structureof second generation computers All components of systemos are connected together using common bus This allows simply add new modules. However, the common bus - a narrow point in the system which limits the system's performance ©S.Maciulevičius

  23. Structureof third generation computers IO IO CPU Main memory Chan- nel Chan- nel MD MT Information processing is separated from information input and output. Data input and output (exchanging with external memory as well) is controlled by special devices cold channels orinput/output processors Processor interprets program and executesinstructions,and controls channel functioning ©S.Maciulevičius

  24. Input device Output device CPU Main memory Control- ler Control- ler Common Bus DMA channel Timer External memory Structureof first personal computers ©S.Maciulevičius

  25. Structureof personal computers Cache CPU Main memory Graphics Bridge 1 Hard disk Bridge 2 Output device Input device Input device LAN Output device ©S.Maciulevičius

  26. New structureof personal computers CPU Main memory Graphics Bridge 1 Hard disk Bridge 2 Output device Input device Input device LAN Output device ©S.Maciulevičius

  27. CPU CPU CPU CPU LAN, IDE, USB, ... Memory and IO controllers LAN, SCSI PCI-X 64/133 PCI-X 64/66 … IO adapters Main memory modules Modern servers ©S.Maciulevičius

  28. Some principles used in development of modern computers Developers of modern processorsare keeping the following principles: • Execution of instructions is controlled by hardware. Previously, execution of complex and long instructions were realized using principle of microprogramming: every instruction was realized by sequence of microcommands (microcommand is analog of instruction at lower level; as the program consists of a sequence of instructions, soinstruction consists of a sequence of microcommands). Microprogramms are usefull in realizationof complex and rarely used instructions. • Processor executes severalinstructions in parallel. Parallel execution of instructions increases processor productivity greatly ©S.Maciulevičius

  29. Some principles used in development of modern computers • Instructions should be easilydecoded. During decoding of instructions there are determined what resources they need, what operations should be executed, where are data to be processed. Therefore, the number of formats used for instructions encoding is reduced, the same instruction word length will be used, reducing the number of fields is reduced as well • Processor should have a large number ofregisters. A large number of registers (at least 32) helps to reduce the number of memory read and write operations (storing intermediate results in registers). As the access to memory requires much more time than the access to register, registers use significantly increases processor productivity ©S.Maciulevičius

  30. Classification of computers According to use for: • scientific calculations, • commerce, • laboratory, • processcontrol, • office, • home, • travel (compact PC). ©S.Maciulevičius

  31. Classification of computers According to size: • homecomputers, • personal computers (desktop), • workstations, • minicomputers, • mainframes, • supercomputers ©S.Maciulevičius

  32. Supercomputers Supercomputers are usedespecially for complex calculations(so calledGrand Callenge problems), knowledge banksand networked data bases Supercomputers are build: • as particularly productive specially designed computers, • using thousands of processors (multiprocessors), • using specialconnected systems of computers (cluster and grid) ©S.Maciulevičius

  33. TOP 500: June 2008 ©S.Maciulevičius

  34. IBMRoadrunner IBM Roadrunner is build using processors of different type: AMD 1,8 GHz dual-core Opteron (6562 CPUs) and IBM 3,2 GHz PowerXCell 8i (12240 CPUs; they were designed for game computers Sony PlayStation 3) Blade modules (TriBlades) are connected using Infiniband (length of opticalwires – 55 miles) TriBlade modulehas two dual-core Opteronsand four Cell 8i CPUs Main memory – 98 TB 296 racks, 560 m2 (6,000 sq ft); weigth - 226 t. Cost - 133 mln USD Power 2.35 MW, effectiveness - 437 Mflops/W ©S.Maciulevičius

  35. IBMRoadrunner ©S.Maciulevičius

  36. Blue Gene/L ©S.Maciulevičius

  37. June 2009: TOP 500 ©S.Maciulevičius

  38. Mainframes Mainframes(large computers)are build forcomplex calculations, maintaining large groups of users,networked data bases Classic large computers - systems IBM/360, IBM/370, IBM S/390 It seemed that the time of large computers has already passed, but IBM's company specialists argue that half of all data of world information systems must be kept in large computers ©S.Maciulevičius

  39. Minicomputers Minicomputers as new classwerebuild immediately after the introducing IBM/360 It was a DEC PDP-8 machine, designed for small companies. DEC later produced the even more popular PDP-11 family of minicomputers Minicomputer popularity quickly grew, so DEC has become the second company in the manufacture of computers (after IBM) The minicomputer CM-1600 (it corresponds to PDP-11/34, but special processor wasadded)was designed and produced in Vilnius SKB SM and computer factory ©S.Maciulevičius

  40. Workstations Workstationis a computer designed for professional work using computer network resources Workstation can be summarized as follows: the computer, which satisfies some special needs upraisingfrom specific professional activities It is more powerful as PC. In most cases it is intended for automated design ©S.Maciulevičius

  41. Personal computers Personal computers(desktops) are build mainly for non-professional users, mostly working in an autonomous mode But recently, when some of the work is done at home, they are used by professionals as well Network computers are less powerfull personal computers with small memory (sometimes diskless) for usingin computer network ©S.Maciulevičius

  42. Reached margins 1 Gigaflop mark was exceeded in 1987 - CRAY-2/4-256 system demonstrated performance 1.4 Gflops rate (system has 4 Cray processors running at 243 MHz) 1 Teraflopmark was exceeded in 1998 - ASCI Blue Mountain system demonstrated performance 1,6 Tflops (system has 6144 MIPS R10000 CPUs running at 250 MHz) 1 Petaflopmark was exceeded in 2008 - IBM Roadrunner ©S.Maciulevičius

More Related