Arithmetic unit,computerPerforms various arithmetic sums inLogical operationOperating parts.The basic operations of the arithmetic unit include addition, subtraction, multiplication and divisionFour arithmetic operations, AND, OR, NOTXORetc.Logical operation, and operations such as shifting, comparing, and transferring, also known asArithmetic logic unit(ALU)。
The solver is composed ofArithmetic logic unit(ALU)、accumulator、Status register、General registerGroup, etc.Arithmetic logicArithmetic unitThe basic function of (ALU) is addition, subtraction, multiplication and division, and, or, or, notXORetc.Logical operation, and operations such as shifting and complementing.computerRuntimeOperation and operation of arithmetic unittypeDetermined by the controller.The data processed by the solver comes fromstorage;The processed result data is usually sent back to the memory or temporarily stored in the arithmetic unit.AndControl UnitTogetherCPUThe core part of.
The processing object of the solver is data, so the data length andcomputerData representationMethod, for arithmetic unitperformanceGreat impact.1970smicroprocessorUsually 1, 4, 8, 16BinaryBit asProcessing dataOfBasic unit。majorityGeneral purpose computer16, 32, 64 bits are used as the length of data processed by the arithmetic unit.The arithmetic unit that can process all bits of a data at the same time is calledParallel solver。If only one bit is processed at a time, it is called a serial solver.Some arithmetic units can process several bits at a time (usually 6 or 8 bits), and a complete data is divided into several segments for calculation, which is called serial/parallel arithmetic unit.Arithmetic units often only process data of one length.Some can also process data of several different lengths, such as halfWord lengthOperationDouble word lengthOperation, quadruple word length operation, etc.Some data length can be specified during operation, which is called variableWord lengthOperation.
The number of operations and operating speed that can be performed by the arithmetic unit indicates the strength of the arithmetic unit's ability, and even indicates thatcomputerOwn ability.The basic operation of the arithmetic unit isaddition。Adding a number to zero simply transfers the number.Complementing the code of one number and adding it to another number is equivalent to subtracting the previous number from the latter.Set two numberssubtract You can compare their sizes.
Arithmetic unit
Left and right shift is the basic operation of the arithmetic unit.WithSymbolIn the number ofData bits, calledArithmetic shift。If the data is combined withSymbolAll bits of move together, calledLogical shift。If the highest and lowest data links are logically shifted, it is calledCyclic shift。
AlgorithmicLogical operationTwo data can be bitwise compared with, orXOR, and the negation of each data.Some calculators can also perform 16 logic operations of binary code.
Multiplication and division are complicated.quite a lotcomputerThese operations can be directly completed by the arithmetic unit of.The multiplication operation is based onadditionOperation based, consisting of one or more bits of the multiplierdecodingControl successive generationPartial product, and the partial products are added to get the product.The division rule is often based on multiplication, that is, select several factors to multiplyDivisor, make it approximate to 1, multiply these factors byDividendThen get the quotient.Not implementing multiplication and division hardwarecomputerMultiplication and division can be realized by programs, but the speed is much slower.Some calculators can also find the maximum number in a batch of data, and continuously perform the same operation on a batch of datasquare rootAnd other complex operations.
Operation method
To realize the operation of the arithmetic unit, especially the four arithmetic operations, a reasonable operation method must be selected.It directly affects theperformance, also related to the structure and cost of the solver.In additionnumerical calculationThe significant digit of the result may be longer, and a certain significant digit must be intercepted, thus resulting in the rounding problem of the lowest significant digit.The selected rounding rules also affect theaccurateDegrees.In SelectcomputerThe following factors should be taken into full consideration when representing the number of: the type of number to be represented (decimal, integer, real number and complex number): determine the representation, and the values that may be encounteredRange: Determine storage and processing capacity.numerical valueaccurateDegree: processing capacity related;data storage And the hardware cost required for processing: the cost is high or low.
Arithmetic unit
Two commonly used formats: fixed-point format: numerical value allowed by fixed-point formatRangeLimited, but the required processing hardware is relatively simple;floating-pointFormat: The allowable value range is large, but the required processing hardware is complex.
2、floating-point number : Due to the value value to be representedRangeThe difference is very wide, which brings a lot of inconvenience to storage and calculationFloating point operationLaw.
Floating point representation, that is, the position of the decimal point isfloatOf.histhoughtoriginateScientific enumeration。IEEE 754 floating-point number (relatively special) floating-point number normalization: mainly solve the problem of the difference of the same floating-point number representationUniquenessQuestion.If not, the mantissa shall be shifted to the left or right.
Machine zeroConcept of: mantissa is 0 orOrder codeThe value is less than the smallest number that can be represented.
3、Decimal numberString representation: because people compare decimal numbersbe familiar with, so incomputerTo add support for decimal operations.Two methods: change the decimal number toBinary numberOperation, and then change from binary to decimal when outputting.Direct decimal operation.Direct operationRepresentation of:character stringForm: compressed decimal number string for non numerical calculation field: divided intoFixed lengthAnd variable length.The corresponding decimal arithmetic unit and instruction support are required.
4. CustomData representation: Flag data representationdescriptor Data representation.difference: The identifier is connected with each data, and the two are stored together in oneStorage unit, and the descriptor should be stored separately from the data;In the descriptor representation, the descriptor is accessed first, then the data, and at least one memory access is added;Descriptors are part of a program, not part of data.Original code: ComparenaturalOfRepresentation, the highest bit indicatesSymbol, 0 is positive, 1 is negative.Advantages: easy to understand.Disadvantages: addition and subtraction operations are complex.Complement: addition and subtractionconvenient, subtraction can be converted toaddition。Fixed-point decimalComplement of.Fixed point integerComplement of,Reverse code: To calculate complementconvenientAnd introduce.Complement from reverse code:SymbolPosition 1, everyone is inverted, and the last digit is added with 1.Code shifting: Used for the representation of order codes. It is easy to compare the size of two shift codesAntithesis。
Arithmetic unit
ASCIIcodeinput code: forchinese charactersInput;chinese charactersStorage;TypefaceCode: used to display Chinese characters.remainderTwo processing methods: recovery remainder method: the operation steps are uncertain, the control is complex, and it is not suitable forcomputerOperation.Alternating addition and subtraction: Do not restore the remainder, the operation steps are determined, suitable forcomputerOperation.logicNumber concept: WithoutSymbolOfBinary number。FourLogical operation:Logical negation, logical addition, logical multiplication, logical difference.Multifunctional arithmetic/Logical operationUnit(ALU)Parallelcarry,traveling waveCarry addition/SubtractorThere are two problems: the operation time is long, and the traveling wave carry add/subtract can only completeadditionAnd subtraction, instead of completing logic operation, control terminal M is used to control productionArithmetic operationOr logical operation, two kinds of operationdifferenceIt depends on whether carry is processed.When M=0, there is no effect on carry, which is arithmetic operation;When M=1, carry is blocked, which isLogical operation。Positive logicMedium, for "1"High levelIndicates that "0" is usedLow levelRepresents, andNegative logicJust the opposite.Logical andNegative logicalrelationshipIs, the "and" of positive logic becomes "or" in negative logic, that is,+· interchange.
Internal bus, busclassification: internal busExternal bus(system bus ), communication bus.BusIt can also be divided intoOne way busandBidirectional bus。beltLatchOfBusThe bus can be reused.The arithmetic unit includes ALU, array multiplication and division deviceregister、Multiway switch、Tristate buffer, data bus, etcLogic unit。The design of arithmetic unit mainly focuses on how to transfer ALU and register to data busOperandsAnd calculation results.Three structures of arithmetic unit: singleBus structureThe main disadvantage of this structure is that the operation progress is slow, but the control circuit is simple.doubleBus structureThe arithmetic unit of.Arithmetic with three bus structure: The characteristic of arithmetic with three bus structure is fast operation time.
structure
Arithmetic unit
The arithmetic unit includes registersExecutive partAnd control circuit.There are three registers in a typical arithmetic unit: receive and save oneOperandsReceive register of;Object that stores another operand and the result of the operationAccumulation register;MultiplyingDivision operationSave multiplier orQuotientOfQuotient register。The execution component consists of aadderAnd eachtypeType I/OGate circuit。The control circuit shall comply with certainChronological orderSend out different controlssignalTo make the data enter the register oradder, complete the specifiedoperation。In order to reduce the number ofstorage's visit, manycomputerThe arithmetic unit of is equipped with more registers to store intermediate calculationsresult, so that it can be directly used asOperands。To improveOperation speed, someMainframeThere are multiple arithmetic units.They can be different types of operators, such as fixed-pointadder、Floating point adder、MultiplierIt can also be an arithmetic unit of the same type.The composition of the calculator depends on the design of the whole machinethoughtandDesignIt is required that different arithmetic methods will lead to different composition of arithmetic units.However, since the basic functions of the arithmetic unit are the same, and its algorithms are roughly the same, the arithmetic units of different machines are much the same.The arithmetic unit is mainly composed ofArithmetic logic unit、General registerGroup andStatus registerform.1. Arithmetic logic unit ALU.ALU mainly completesBinaryFixed point of informationarithmeticOperationlogicOperations and various shift operations.Arithmetic operations mainly include fixed-point addition, subtraction, multiplication and division.Logical operationMainlyLogical and、Logical OR、Logical XORandLogical negationOperation.Shift operation mainly completes logicShift LeftAnd right shifts, arithmetic left shifts and right shifts, and other shift operations.In some machines, ALU also needs to complete value comparison and change valuesSymbol, CalculationOperandsstaystorageInaddressEtc.It can be seen that ALU has a strong functionCombinational logic circuit, sometimes referred to as multi-functionGenerator, which is the core component of the arithmetic unit.The number of data bits that ALU can process (i.eWord length)It is related to the machine.asZ80Single board computerIn, ALU is 8 bits;IBM PC/XTAnd AT machines, ALU is 16 bits;386 and 486microcomputerIn, ALU is 32-bit.ALU has two data input terminals and one data output terminalwidth(that is, the number of bits) is the same as the data width processed by ALU.
2、General registerThe arithmetic unit of the machine designed by group has a set of general registers.It is mainly used to save theOperandsThe result of the sum operation.Early machines only designed oneregister, for storageOperands, operation result and performing shift operation
(1) Zero flag bit (Z): when the operation result is 0, the Z position is "1";If not 0, set "0";
(2) Negative flag bit (N): when the operation result is negative, the N position is "1";Is timing, set to "0";
(3) Overflow flag bit (V): when the operation result overflows, the V position is "1";If there is no overflow, set "0";
(4) Carry orDebitSign (C): doingadditionIf the operation resultMost significant bit(WithSymbolNumber, that is, sign bit;yesUnsigned numberWhen carry is generated forward, C position is "1";When there is no carry, set "0".When subtracting, if it is not enough to subtract, when the most significant digit is borrowed forward (at this time, no carry forward is generated), position C is "1";When there is no borrowing (i.e. there is carry generation), position C is "0".In addition to the above states,Status registerIt is also often equipped to save relevant interrupts and machinesworking condition (User statusorNuclear mentality)Some flag bits of information such asSymbolNot exactly the same), so as to reflect the working state of the machine running program in a timely manner, so some machines call it“Program status word”Or“Processor status word”(Processor Status Word,PSW)。
performance index
Announce
edit
oneMachine word lengthThe machine word length refers to the basic digits of the data involved in the operation.It's decidedregister、Arithmetic unitanddata busThis directly affects the price of hardware.Word lengthIt indicates the calculation accuracy.In order to coordinate the accuracy and cost, and meet various requirements, manycomputerAllowable variationWord lengthCalculations, such as half word length, full word length, andDouble word lengthEtc.Because number andCommand codeAll onMain storageMedium, soWord lengthAndInstruction codeThe length often has a corresponding relationship, and the word length also affects the instructionsystem functionThe strength of.computerWord lengthFrom 4-bit, 8-bit, 16 bit, 32-bit to 64 bit.machineWord lengthCan contain one or morebyte。be used forScientific computingTo ensure accuracy, a longerWord length;be used fordata processing、Industrial control16 bit or 32 bit word length can meet the requirements.
Arithmetic unit
twoOperation speed, it iscomputerOne of the main indicators of.computerThe time required to perform different operations and operations may be different, so there are different calculation methods for operation speed.Commonly usedaverage velocity, that is, the average number of instructions that can be executed in a unit time, such ascomputerThe operation speed is 1 million times/second, which means that the machine can execute 1 million times/second on averageinstructions(i.e. 1MIPS)。Sometimes usedWeighted average method(That is, according to the execution time of each instruction and the proportion of this instruction in all operationspercentageCalculation).
Function classification
Announce
edit
The basic function of the arithmetic unit is to process various data, such as arithmeticFour arithmetic operations, and, or, reverse, etcLogical operation, Arithmetic sumLogical shiftOperation, value comparison, changeSymbol, CalculateMain storageaddressEtc.In the solverregisterIt is used to temporarily save the data involved in the operation and the intermediate results of the operation.Corresponding parts shall also be set in the calculator to record the characteristics of an operation result, such as whether or notoverflow, resultingSymbolBit, whether the result is zero, etc.computerThere are many types of calculators used. From different perspectives, there are differentclassification method 。fromdecimal pointThe representation of can be divided into fixed-point solver andFloating point arithmetic unit。The fixed-point solver can only doNumber of fixed pointsOperation, characterized byNumber of machinesRepresentedRangeSmaller, but simpler in structure.Floating point arithmetic unitStrong function, which can not onlyFloating point numberIt can also operate on fixed point numbers, and its number representationRangeIt is large, but its structure is quite complex.From the perspective of carry systemBinaryArithmetic and TenBaseArithmetic unit.commonlycomputerAll adoptBinaryAs computers are widely used in business and data processing, more and more machines are expandeddecimal systemThe function of operation enables the calculator to completeBinaryOperation, can also complete decimal operation.computerThe arithmetic unit in the must have the function of completing a variety of operations, so it is necessary to integrate various algorithms and design a complete arithmetic unit.
For processing of data, its input comes from AC and DR, and the result is sent back to AC.MQ register is stored during multiplicationmultiplier, and stored during divisionQuotient, so it is called the quotient register.DR is used for storageMultiplicandorDivisor, and the result (product or quotient andremainder)It is stored in AC - MQ.In the four operations, the typical methods of using these registers are as follows:
Operation class register relationadditionAC+DR→ACsubtractionAC-DR→ACmultiplicationDR×MQ→AC-MQdivisionAC÷DR→AC-MQ
For order code components, only order code addition, subtraction and comparison operations can be performed.In Figure 2-21,OperandsThe order code part of is placed in registers E1 and E2, which are connected to the parallel adder for calculation.floating-pointadditionThe comparison with the order code required for subtraction is achieved through E1-E2, and the result of subtraction is put intoCounterE, and then follow E'sSymbolTo determine which order code is larger.Before adding or subtracting mantissa, a mantissa needs to be shifted, which is controlled by the counter E, in order to reduce the value of E to 0 in sequence.E Each time you subtract 1, the corresponding mantissa moves 1 bit to the right.Once the mantissa is completed at high speed, they can be processed according to the usual fixed-point method.The order code value of the operation result is still placed in the counter E.
2. Point coprocessor, 80x87 is AmericanIntel CorporationTo process the arithmetic operations and multiplefunctionSpecial arithmetic operation processor designed and produced for calculation.Because their arithmetic operation is compatible with 80x86CPUIt is also calledCoprocessor。We take 80x87 as an example to discuss the composition of floating point arithmetic unit.The main functions of the floating point coprocessor are as follows:
(1) It can work asynchronously and in parallel with the supporting CPU chip.80x87 is equivalent to an I/O unit of 386. It has its own instructions, but cannot be used alone. It can only operate as a coprocessor of 386 main CPU.Because the real work of reading and writing main memory is not completed by 80x87, but by 386.If 386's instructions read from main memory are 80x87 floating point instructions, they willOutput modeSend the instruction to 80x87, and 80x87 will decode and execute floating point operation after receiving it.During 80x87 operation, 386 can execute another instruction, thus realizing parallel work.If 386 fetches another 80x87 instruction during 80x87 floating-point operation instruction, 80x87 will give a "busy"Sign signalReject it, causing 386 to suspend sending commands to 80x87.386 can only send once after 80x87 finishes floating point operation and cancels the "busy" flag signal.
(2) Highperformance80 bits ofWord lengthThe internal structure of thestackMode managedRegister group。80x87 slavestorageWhen fetching data and writing data to memory, 80 bit temporary real number and other 6 types are useddata typePerform automatic conversion.All data are represented in 80x87 as 80 bit temporary real numbers.Therefore, 80x87 has an internal structure of 80 bits and eight 80 bitsWord lengthThe register group managed by "first in first out" mode is also calledregister stack 。These registers can work in stack mode. At this time,Top of stackbe used asaccumulator;Or by register numberDirect accessAny register.
(4) Capable of handlingBinaryThere are 7 kinds of data in three categories: floating point number, binary integer and decimal number string.These seven data types are represented in the register as follows: short integer (32-bit integer) S 31 bit(BinaryComplement)Long integer (64 bit integer) S 63 bit (binary complement) short real number (32-bit floating point number) S exponential mantissa (23 bits) long real number (64 bit floating point number) S exponential mantissa (52 bits) temporary real number (80 bit floating point number) S exponential mantissa (64 bits) decimal number string (decimal 18 bits) S -- d17d16... d1d0.Here S is one digitSymbolBit, 0 represents positive, 1 represents negative.The base value of three floating point order codes is 2.For order code valueCode shiftingRepresent with mantissaOriginal codeexpress.The mantissa includes 32 bit, 64 bit and 80 bit.It is not only a floating point arithmetic unit, but also includes executionData operationAll you needControl lineIn terms of the operation part, there are components and processes for handling the floating point index partMantissaSome parts, as well as accelerated shift operationShifterLines, which are connected with eight 80 bits through the exponential bus and decimal busWord lengthRegister stack.(5) Internal errormanagement functionIn order to ensure the correct execution of the operation, 80x87 has also set three internal 16 bitWord lengthThe register of, that is, feature register, control word register and status register.The characteristic register uses every two bits to represent the status of each register in the register stack, that ischaracteristic valueFour combinations of 00-11 indicate that the corresponding register has correct data, data is 0, data is illegal, and no data.The control word register is used to control the internal operation of 80x87.PC is precision controlBit field(2 bits): 00 is 24 bits, 01 is standby, 10 is 53 bits, and 11 is 64 bits.RC is the rounding control bit field (2 bits): 00 is the nearest rounding, 01 is rounded in the - direction, 10 is rounded in the+direction, and 11 is rounded in the 0 direction.IC isInfinityControl Bit: when this bit is 0,+and - are treated as the same value; when this bit is 1,+and - are not treated as the same value.Control registerThe lower 6 bits of are used as abnormal interrupt mask bits: IM is illegal processing, DM is illegal operand, ZM is 0 as divisor, and OM isOverflow, UM is underflow, PM is precision decline.Status wordThe register is used to indicate the result processing of 80x87. For example, when the "busy" flag is 1, it means that a floating point operation instruction is being executed, and when it is 0, it means that 80x87 is idle.The lower 6 bits of the status register indicate 6 of the abnormal errortypeType, 6 bits lower than the control register.When the control register bit of is 0 (unshielded) and the status register bit is 1, an abnormal error occursInterrupt request。
3. In CPUFloating point arithmetic unit, Pentium CPU includes floating point arithmetic unit in the chip.Floating point operationAssembly line design is adopted for components.The instruction execution process is divided into 8 segments of pipeline.The first four paragraphs areInstruction prefetch(DF)、Instruction decoding(D1)、addressGenerate (D2), fetchOperands(EX), completed in the U, V pipeline;The last four paragraphs are execution 1 (X1), execution 2 (X2), and resultsWritebackRegister heap(WF)、Error Reporting(ER), onFloating point arithmetic unitCompleted in.Generally, a floating point operation instruction is completed by the V pipeline.Floating point parts have floating point specificadder、MultiplierandDivider, 8 80 bitsRegister groupBuilt register stack, internaldata busIt is 80 bits wide.Therefore, floating point assembly can supportIEEE754 standardOfSingle precisionandDouble precisionFloating point number in format.In addition, an 80 bit floating point number called temporary real number is also used.For floating point fetchingaddition, multiplication and other operations, using new algorithms and hardware to achieve, its execution speed is more than 10 times 80486.
development
Announce
edit
In the 5th century BC, the Chinese inventedAbacus, widely used inCommercial tradeThe abacus is considered the earliestcomputer, and has been used until now.AbacusComputing powerTo exceedcomputerThe abacus reflects the wisdom of the Chinese people.
Arithmetic unit
It was not until the 17th century that computing equipment made a second important progress.In 1642, French Blaise Pascal (1623-1662) invented automatic carryadder, called Pascalene.In 1694,GermanyMathematician Gottfried Wilhemvon Leibniz (1646-1716) improved Pascaline to calculate multiplication.Later, Charles Xavier Thomas de Colmar, a Frenchman, inventedFour arithmetic operationsCalculator for.moderncomputerThe real origin of "is from Charles Babbage, a British mathematics professor.".Charles Babbage found that there are many errors in common computing devicesCambridgeWhen learning, he thinks he can usesteam enginePerform the operation.At first he designedDifferential machineIt is used to calculate the navigation table. Later, he found that the differential unit is only specializedpurposeWe gave up our original research and began to design modern machinescomputerBasic componentAnalyzer。(Analytical Engine)
Steam power of BabbagecomputerAlthough it was not completed in the end, it is also very primitive by today's standards. However, it outlines the modernGeneral purpose computerOfbasic function Part is a breakthrough in concept.
In the following years, many engineers made important progress in other aspects. American Herman Hollerith (1860-1929), according tojacquard weaveloomThe principle ofcomputerAnd bring it into the business field to establish a company.
February 15, 1946, the symbol of modern timescomputerBornENIAC(Electronic Numerical Integrator and Computer)Philadelphiamake public to the world.ENIAC stands forComputer HistoryThe milestone on, which is programmed by rewiring between different parts, also hasParallel computingAbility.ENIAC byU.S. governmentandUniversity of PennsylvaniaCooperative development, using 18000Electronic tube70000resistor, 5 million welding points, 160 kW power consumptionOperation speedIs 5000 times per second.First generation computerIs characterized byOperation instructionIt is prepared for specific tasks, and each machine has its ownmachine language, the function is limited and the speed is slow.Another obvious feature is the use of vacuum tubes anddrumStore data
In 1948,transistorThe invention replaced the huge electronic tube,Electronic equipmentThe volume of is decreasing.In 1956, the transistorcomputerUsed in, transistor andCore memoryCausedSecond generation computerGeneration of.Second generationcomputerSmall size, fast speed, low power consumptionperformanceMore stable.In 1960, there were some second generation successfully used in the business field, universities and government departmentscomputer。Second generationcomputerTransistors are used instead of electronic tubes, and there are some components of modern computers: printersmagnetic tape、disk、Memory, operating system, etc.computerThe program stored in makes the computer have good adaptability and can be more effectively used for commercial purposes.During this period, more advancedCOBOLandFORTRANEtccomputerProgramming is easier.New occupation(programmer、analystandcomputer systemExperts) and the entiresoftware industry This was born.
1958Texas InstrumentsJack Kilby, an engineer ofIntegrated circuit(IC), three types ofelectronic componentTo a small piece of silicon.More components integrated into a single semiconductorchipOn,computerBecome smaller,power wasteLower, faster.The development of this period also included the use of operating systems, which madecomputerMany different programs can be run simultaneously under the control and coordination of the central program.
large-scale integrated circuit(LSI) can hold hundreds of components on a chip.In the 1980s,vlsi(VLSI) holds hundreds of thousands of components on the chip, and later (ULSI) expands the number to millions.Such a number of components can be accommodated on a coin sized chip tocomputerThe volume and price of are declining, while the function and reliability are increasing.In the mid-1970s,computerWhen manufacturers began to bring computers to ordinary consumersMinicomputerWith friendly interfacesoftware package, procedures for non professionals and the most popularWord processingandSpreadsheetProcedure.In 1981,IBMintroductionpersonal computer(PC) for home, office and school.Individual in 1980scomputerThe competition of,microcomputerAs the ownership of the computer continues to increase, the size of the computer continues to shrink.AndIBM PCCompetitive AppleMacThe intosh series was launched in 1984, and Macintosh provides a friendlyGraphical interface, users can use the mouseconvenientLocal operation.
In 1673, Gottfried Leibniz built a counting machine with stepped cylindrical wheels, called "Stepped Reckoner", which can multiply repeated numbers and automatically add them to the adder.
Arithmetic unit
In 1694, Gottfried Leibniz, a German mathematician, improved Pascalene of Paska and produced a computermultiplierIt still uses gears andDialOperation.
In 1773, Philipp Matthaus made and sold a small amountaccurateTo 12 bitcomputerDevice.
In 1775, The third Earl of Stanhope invented a multiplication calculator similar to Leibniz.
In 1786, JH. Mueller designed aDifferential machineUnfortunately, no funds were allocated for manufacturing.
In 1801, Joseph Marie Jacquard'sloomIs ordered by connectionPunch cardControls the style of weaving.
In 1854, George Boole published "An Investigation of the Laws of Thought", which is aboutSymbolAnd logical reasons, which later becamecomputerBasic concept of design.
In 1858, a telegraph line crossed for the first timeAtlantic, and provided several days of service.
In 1876, Alexander Graham Bell invented the telephone and obtainedpatent right。
Between 1876 and 1878, Baron Kelvin produced aOvertoneAnalyzerAnd tide prediction machine.
In 1882, William S. Burroughs resigned from the bankClerkAnd focused on the invention of adder.
In 1889, Herman Hollerith's electric tabulator had an outstanding performance in the competition and was used in the 1890Demographic survey。Herman Hollerith used the concept of Jacquard loom to calculate. He used cards to store data, and then injected them into the machine to compile the results.This machine enables the demographic survey results that would have taken ten years to get in just six weeks.
In 1893, the first four function calculator was invented.
Guglielmo, 1895MarconiTransmit broadcast signals.
In 1896, Hollerith established Tabulating Machine Company.
In 1901, the punch key appeared, and there were few changes in the following half century.
In 1904, John A. Fleming achieved a vacuumdiodeThe patent right ofRadio communicationBuild the foundation.
In 1906, Lee de Foredt added a third valve to the diode in Felming and created a three electrodeVacuum tube。
In 1907, record musicNew YorkForm the first official radio station.
In 1908, British scientist Campbell Swinton stated thatElectronic scanningMethod and predictioncrtMake TV.
In 1911, Hollerith's watch machine company and the other twoCorporate merger, forming the Computer Tabulating Recording Company (C-T-R), tabulating and recording company.But in 1924, it was renamed International Business Machine Corporation(IBM)。
In 1911,NetherlandsPhysicist Kamerlingh Onnes discovered superconductivity in Leiden University.
In 1931, Vannever Bush invented aDifferenceThe counting machine of the program, which can solve some problems for mathematicians and scientistsheadacheThe complex differential program of.
In 1935, IBM (International Business Machine Corporation) introduced "IBM 601", which is a punch card machine with arithmetic parts and can calculate multipliers in one second. It plays a great role in scientific and commercial computing. A total of 1500 units have been produced.
In 1937, Alan Turing came up with a concept of "Universal Machine", which can execute any algorithm, forming a basic concept of "computability".Turing's concept is better than other similar inventions, because he usesSymbolThe concept of symbol processing.
In November 1939, John Vincent Atanniff and John Berry manufactured a 16 bit adder.It was the first machine to calculate with a vacuum tube.
In 1939, Zuse and Schreyer created "V2" [later called Z2], and this machine continued to use the machinery of Z1Reservoir, add one withBreakerThe new arithmetic unit of the logic (Relay Logic).But when Zuse finished the draft, the plan was suspended for one year.
Arithmetic unit
In 1939-40, Schreyer completed the 10 bit adder with vacuum tube andneonNeonmemory。
In January 1940, at Bell Labs,Samuel WilliamsAnd Stibitz have completed a machine that can calculate complex numbers, called“Complex digital counter(Complex Number Calculator) ", later renamed" Model I Relay Calculator ".It uses the telephone switch part to makeLogic unit: 145 breakers, 10 crossbar switches.Numbers are represented by "Plus 3BCD".In September of the same year,telexTyping etype is installed in a math conference and connected to New York by New Hampshire.
In 1940, Zuse finally completed Z2, which was better than the operation, but not too reliable.
In the summer of 1941, Atanasoff and Berry completed a calculator specially designed to solve the system of simultaneous linear equations, later called "ABC (Atanasoff Berry Computer)". It has 60 50 bit memories, which are installed on two rotating drums in the form of capacitors,Clock speedIt is 60Hz.
In February 1941, Zuse completed "V3" (later called Z3), the first programmable counter in operation. It also usedFloating point operations, 7-bit index, 14 bit indexMantissa, and aSign。The memory can store 64 words, so 1400 breakers are required.It has more than 1200 arithmetic and control parts, and the programming, input and output are the same as Z1.In January 1943, Howard H. Aiken completed the "ASCC Mark I" (Automatic Sequence -- Controlled Calculator Mark I), also known as "Howard Mark I". This machine is 51 feet long and weighs 5 tons. It is a combination of 750000 parts. It has 72accumulatorEach has its own arithmetic unit and a 23 digit register.
In December 1943, Tommy Flowers and his team completed the first "Colossus", which has 2400 vacuum tubes as logic components and 5 paper tape loop readers, each of which can work 5000 per secondcharacter。
In 1943, led by John Brained,ENIACStart research.John Mauchly and JPresser Eckert is responsible for the implementation of this plan.
1946v first electronic digitalIntegral calculator(ENIAC)U.S.AConstruction completed.
American Calculator Association, 1947(ACM)Establishment.
In 1947, Britain completed the first storage vacuum tube O 1948BellThe telephone company developed semiconductors.