《计算机组织与体系结构课件01》由会员分享，可在线阅读，更多相关《计算机组织与体系结构课件01（20页珍藏版）》请在金锄头文库上搜索。
1、IntroductionTongwei RenSoftware Institute, Nanjing UniversityFeb. 18, 2014Lecture 1C0864 Computer Organization and Architecture2NANJING UNIVERSITYConcept Computer General-purpose electronic digital computer Architecture (visible to programmer) The attributes have a direct impact on the logical execu
2、tion of a program Instruction set, the number of bits to represent data type, E.g.: Is there a multiply instruction? Organization (transparent to programmer) The operational units and their interconnections Control signals, memory technology, E.g.: Implement multiply by a hardware unit or repeated a
3、ddition?3NANJING UNIVERSITYA Brief History of Computers First generation: Vacuum tubes (1946-1957) ENIAC (1946-1955): decimal, manually programming4NANJING UNIVERSITYA Brief History of Computers First generation: Vacuum tubes (1946-1957) IAS (1946-1952*): binary, stored program5NANJING UNIVERSITYA B
4、rief History of Computers First generation: Vacuum tubes (1946-1957) IAS (1946-1952*): the von Neumann machine Idea: main memory storing programs and data Prototype of all subsequent computers Central Arithmetical (CA) Central Control (CC) Memory (M) Input (I) / Output (O)6NANJING UNIVERSITYA Brief
5、History of Computers (cont.) Second generation: Transistors (1958-1964) NCR and RCA, IBM 7000: transistor is smaller, cheaper, and dissipates less heat but can be used in the same way as a vacuum tube Introduction of more complex arithmetic and logic units and control units, the use of high-level pr
6、ogramming languages and the provision of system software. 7NANJING UNIVERSITYA Brief History of Computers (cont.) Third to N generation: Integrated circuits (1965-now) Idea: fabricate an entire circuit in a piece of silicon rather than assemble discrete components made from separate pieces of silico
7、n These transistors can be connected with a process of metallization to for circuits Scale: small large very large ultra large .8NANJING UNIVERSITYA Brief History of Computers (cont.) Moores law (Gordon Moore, 1965) The number of transistors that could be put on a single chip is doubling every year
8、(1965-1969) / 18 months (1970-now) Consequence The cost of computer logic and memory circuitry is dramatically fallen for the cost of a chip was unchanged Smaller size leads to flexibility and probability Operating speed is increased for the electrical path length is shorten Reduction in power and c
9、ooling requirements The interconnections on the integrated circuit are more reliable than solder connections and few interchip connections9NANJING UNIVERSITYA Brief History of Computers (cont.) Constant function Data processing Data storage Data movement Control10NANJING UNIVERSITYPossible Computer
10、Operations (cont.) Data processing E.g.: check data by antivirus software E.g.: receive data from network and store it after CRC verification11NANJING UNIVERSITYPossible Computer Operations Data movement E.g.: move the photos from camera to flash disk12 Data storage E.g.: store a music file in memor
11、yNANJING UNIVERSITYA Brief History of Computers (cont.) Improving performance13NANJING UNIVERSITYComputer Performance Key parameters of computer Performance, cost, size, security, reliability, power consumption, Performance evaluation CPU: speed Memory: capacity, speed I/O: speed, capacity14The main
12、 goal / driver is the increase of CPU speedNANJING UNIVERSITYCPU Performance System clock Clock rate / clock speed (HZ): fundamental rate in cycles per second at which a computer performs its most basic operations Clock cycle / clock tick (s): a single electronic pulse of a CPU Cycle time: the time
13、between pulses15NANJING UNIVERSITYCPU Performance (cont.) Instruction execution A processor is driven by a clock with a constant frequency or, equivalently, a constant cycle time . Let be the number of cycles required for instruction type , and be the number of executed instructions of type for a gi
14、ven program. The overall can be calculated as follows: = =1 () , = =1 Process time to execute a given program: = = +( ) 16Transfer data between processor and memoryNANJING UNIVERSITYCPU Performance (cont.) Million Instructions Per Second (MIPS) = 106 = 106 Million Floating Point Operations Per Secon
15、d (MFLOPS) = 10617NANJING UNIVERSITYCPU Performance (cont.) Benchmarks Measure the performance of systems using a set of benchmark programs Averaging results: Arithmetic mean: = 1 =1 Harmonic mean: = =1118NANJING UNIVERSITYSummary Concept organization, architecture Computer history The von Neumann machine, Moores law Possible computer operations Computer performance CPU performance evaluation19
