DE2001342B2 - DIGITAL COMPUTER - Google Patents

Description

3 4

can also occupy memory 2 regardless of the respective sign bit. The capacity of the battery

Unchanged instructions for an operand aus- mulators 5 can be fixed or changeable; she can

is able to feed through an operand - possibly on the word greetings of the in the

Address is designated. This results in the memory 2 being related to stored values.

Advantage that the said control device can also be used by the computing power of the computer

normal command execution can be exploited. specified instruction set should have a normal »Addi-

In accordance with a further expedient embodiment, hold the "tion" or "subtraction" command so that

According to the invention, the linking device is made up of the content of the accumulator through the content of

formed that they are used for changing the values stored in the selected Jn memory

Operand address part contained operands- io can be modified to be in the accumulator

Address in addition to that in the sign bit position to form an end variable that is a positive or negative

hanclenen sign bit also has a control bit tive sign. In addition, the

uses that at least one instruction of an instruction set contains a normal "load" instruction,

The instruction set located in the accumulator contains the content of a specific address position

hold is. This has the advantage that 15 is introduced into the accumulator 5 in the memory 2

a greater possibility of variation in terms of can be. Furthermore, an output command should be provided

named change of address is achieved. be seen through the content of a particular

In the following an embodiment of the storage location of the memory 2 or the accumulator

Invention explained with reference to the drawings. tor 5 transmitted to an output device 5 out

Fig. 1 shows in a block diagram an AU-20 which can become, like a magnetic tape, a

purpose digital computer; Typewriter, to an electronic drawing

F i g. Figure 2 illustrates a typical instruction format; establishment, etc. Furthermore, the command salt should be a

F i g. 3 shows a flowchart for executing a conditional jump instruction by which the

the digital recording of a straight line; Sign of a specific memory address (or

F i g. Fig. 4 shows a flow chart for multiplying how the contents of the accumulator is checked in order to

two binary numbers; determine the address of the next command.

F i g. 5 illustrates how a double-length Ak- Finally, the instruction set should contain an accumulator-

kunullator in a multiplication program radio shift command, an input command, a load

tionicrt; Accumulator command and a store command

F i g. 6 illustrates the multiplication 30 hold process.

lions algorithm of four-bit binary although the application of internal indexing

numbers when multiplying three by six; and the address modification according to the invention

F i g. Figure 7a illustrates how the accumulator will be used below with regard to a few of the above

The sign bit and the least significant bit for the designated commands should be explained

Actuation of a four-way pump will be used, it can be seen that the command set has a large number

the can; can contain further commands and that the basic

F i g. 7b illustrates in a diagram the concepts of the invention with such commands as well

four possible combinations between the pre-executable.

character bit and the least significant bit. According to FIG. 2 includes a typical normal leg F i g. 1, there is shown in a block diagram a typical general purpose digital computer denoting certain basic operations, consisting of normal codes (bits 1 through H) such as e.g. B. the operations consists of elements and one input-output "addition", "subtraction", "store", "load", device 1, a memory 2, hereinafter also "output", "input" etc. Besides this normal Arithmetic unit 3 and command, designated as a data center, has an operand address (bits a control device 4 contains. For the purpose of 45 H + \ to λ), which a specific memory location in explanation of the invention, the memory 2 can be referred to as the memory, which is the in the case of the respective any memory can be assumed as containing operands to be used as an instruction. If core memories with arbitrary access are to be coded as thin-film iterative program loops, it is expedient to change the address part of the command to more with serial access than disk memory 5 °. In this context z. B. on or as a delay line. In the following it is also assumed that there is a desire to form the sum of a digital computer, referred to only as a computer, so long series of numbers, which are organized in, that it has a fixed or variable word - other addresses of the computer memory of a size of any length ( Number of bits). the addresses 030 to 077 are included. It is so-The number system used can be a pure binary 55 with possible the same addition command for aufsystem, whereby negative values are used by means of characters 1 consecutive addition, with the or 2 being represented in complement notation. Address of the "addition" command changed will, consider. However, it can also be done with a sign and in front of the respective addition, so that the content amplitude form using the binary address of the next following address of the respective partially coded decimal system can be used or the sum is added. In known digital with any other system in which the pre-arithmetic units are used, the address change characters of the representing a specific data word are made by an auxiliary device, so that information is given by a specific bit at a particular operand to which the command was given Position of the word is shown. The actual access has, depending on the data center, contains as a main element a content of the auxiliary equipment that can be changed. Accumulator 5, which actually consists of hardware. The auxiliary device used (such as a set of flip-flops or the like) is an independent index register, the content of which is usually modified by a special command

will. This special command is independent of the commands that change the contents of the accumulator. In the example above, an index register can be loaded with the value of the first address. Then this value can be incrementally increased by one counting unit with each addition in order to gain access to consecutive memory addresses.

An important feature of the invention is that it makes it easier to modify an address on the fly without using special index commands and additional registers besides the accumulator to fall back on. The sign bit 10 of the accumulator 5 essentially acts as a single-bit index register. In addition, the computer is organized in such a way that it responds to a set of commands, which has a control bit, with the help of which the programmer is enabled, a select unmodified or modified (index) addressing. If index addressing is selected, the operand address is used as the direct value that is specified in the instruction, if the accumulator contains a positive choice (sign bit positive). In contrast to this, it is If the sign bit is negative, the operand address is used as the direct value plus a known one Step value is used (typically +1 or + 2). Which value is added depends on the particular one Command off. This function is provided by the in F i g. 1 shown linking device 8 executed.

In order to illustrate the advantages of a digital computer organized in this way, let the following instruction set be considered closer.

LDA (Y), 0 Load accumulator with the content of memory location Y

LDA (Y), 1 load accumulator with the content of the storage location Y ₁ if the accumulator (A) is currently positive, or with the content of YM. if A is even negative.

ADD (Y), 0 Add the content of memory location Y to the content of the accumulator.

ADD (Y), 1 Add the content of memory location Y if the accumulator is positive, or the content of memory location Y + 1 if the accumulator is currently negative.

OUT (Y) ₁ O Transfer the content of memory location Y to an output device.

OUT (Y), 1 Transfer the content of memory location Y to an output device if the accumulator is positive, or the contents of Y + 1 if the accumulator is currently negative.

Gradually reduce the content of the memory cell immediately following the memory cell in which this command is stored. If the memory cell contents do not underflow, then N commands jump back (ie take the next command from a memory location N of the memory cells before the DIJB command is found). With only these few selected commands, a computer program can be written that z. B. is able to implement a commonly used algorithm to approximate a straight line by performing step-by-step character movements. In this connection it is assumed that the Z and Y direction components of the line are given in accordance with a ratio of A to B and that the line should have a number of C steps or jumps. If the drawing device is coupled to the computer as an output device in order to carry out step-by-step movements parallel to the Z or to the Y axis, then the line can be drawn in accordance with the flow chart shown in FIG. The step-by-step movements take place depending on whether a code character X or a code character Y is fed to the output. The complete program can simply be written in the following way:

OUT (Z), 1
ADO (A), I.
DIJB (2)

The constant code character X and the code character Y are stored in two consecutive memory locations, so that the command OUT (Z), 1 results in the code character Z being selected when the accumulator is positive. In contrast, the code character Y is selected when the accumulator is negative. In a corresponding manner, the constants A and B are stored in two successive memory locations, so that either A is added to the accumulator or that B is subtracted from the accumulator. Which function is executed depends on the sign of the accumulator for the ADD (/ 1), 1 command. The count value CI is stored in the memory location which immediately follows the DIJB command, so that this value can be reduced step by step until an underflow occurs with the respective recursion.

The three-step program described above for carrying out the algorithm relating to the representation of the straight line is typical of the program simplification, as it results when the calculating machine is suitable for utilizing the sign of the accumulator content for a simplified address modification. For the purpose of furtheri explanation of the versatility of the application of the special index in connection with the information supplied by the accumulator sign bit purpose; Execution of other selection operations is shown in FIG. 4 shown flowchart entered in more detail. This flow chart illustrates the multiplication of two binary numbers. In this application, the sign bit is used as a value display to determine whether the multiplicand is to be added or not. In this context, it is assumed that the computer s is designed to have an accumulator twice as long, as shown in FIG. 5 illustrates. The multiplier can first be entered in the leftmost bit positions (upper half, and the lower half can be set to zero. The sign bit of the accumulator is used to determine, c

ft

hold the accumulator positive, then continue the journey

or not Each multiplactor bit is jumped _{in the p 0SltlV}

_the contents of the accumulator are positive, taking

nriift and, by means of a left shift operation, _{the next} command is obtained from the address «. Is the

deletes This operation also has the effect of a _{positive content of the accumulator} , so set the

Left shift of the subtotal for the sake of the sequence.

Sign of the night Mu tipiftator zMg ^ ₁ ^ ^ _{Μτ &} ^ _{a; ans0Q} .

SSlXchUrH ^ however not. _st en set the order,

If one of the number of bits of the ^M _K ^ultl P ^hkat ° ' ^S _c w " _As ^{the number of ^ 1} ? ^« ^ _ £ ^! 5 should be evident from each of the above normal commands, the programmer has made a choice , then the multiplication program 15 must be ensured. Ended with the sign, and in ^ / ^ X ^ L ^^ Z Mt-Index computation, however, a four-way control is the product. By adding ^ ^^ _context, assume three more Commands, namely the Beteme. ^ B ^ _control command causes a jump to

LSA left shift of the accumulator, _{causes e} in stored in memory location a - 1

ADR (M), 1 Add the content of the S _P , - «* ^^^ ^ feTscS

(M) to the right half of the AkkurnulahenKHg ; ι> β of _{the § icherzelle ß ver} .

tors, as this is taken into account by the sign of the * 5 in addition to the ^ ^^ ^^,

Accumulator is indexed, _h ^ _{state of the bit never the} lowest valence of the

j α · AM / iimiilators from as illustrated in F i 2. 7 a, a program loop for using the de ^ ™ ^ J ^. _Vo ^ _mWt . _Inducing

Sign-indexing-use algorithm ^ _w ⁿ _{andt and the bit never} Driger value hinzu-

can be written as follows: to add, the address B is replaced by a 1

ADR (M) 1 indexed if the content of the accumulator is negative

_T ". i _st This means that the next instruction from the

^ Cr _m memory location β or β + 1 is taken into account. from

DUB (Z) ₃₅ which storage location the next command takes into account

RSA is seen depends on the algebraic sign

„W, 1 λ ·· η H _P n Multi of the contents of the accumulator. However, is the domestic

The value is zero, M + 1 contains the multi of the mn _{d so the next plicand} becomes, and DlJB (Z) is followed by a number of halt ^ des, Accum ^^ ^ ^ ^ ^ ₊ ^ ^^ bits of the multiplier. In I ig. 6 and at Bejehl V ^ memory location the next BeHand of an example the different states d ^ s ^ ng ^ _{is <held by the} sign bit of the accumulator during execution; the MuJipWatoo tem B ^ ^ ^ _{poss [chcn combina} . two four-bit numbers according to derι hear ^^ _{e chend Fig} . _{7b on em}

Changes to the above P ™ ^^ ^chlei *% ^vera SSator minimum should be reduced. The multiplicand is 3 (001, and the ^ P * ^ _{45 D} ™ indexing can also be used to be 6 (0110). First, the upper ^ aWe of the 45 J ^ ¹ f _{of alternate} i _nd s selection opera accumulators contain the Multiplier (0110), during * e almost J ^ _{Above that, the} lower half of the accumulator with characters ^ NuIl J ^^ f ^ ₅ Shen of the accumulator can be assumed that a Re * ^ ™ = ™ ^!, Character a number (as in the first example) or the accumulator contents (step 10) ^ «^ J * ^ ¹ ^ ₅₀ £ value of a selected bit (as with the product 18 (00010010) in the. ^ ™" ^ _t ^ second and third examples) are exploited. Fer following the last left shift (step yj ™ * ^{η ζ} ™ _{above froM} g _e leads. indexing au into position to take. another commands are used wi "for further example, strong the front a ^ _{s ichern.} Add, Subtract, from sign bit-lndizierung is the application of Kon ^"La ° ^ _{In Re} chnern with indirect zeptes on e A conditional jump command should be given. 55 Enter, _calculate ^ _{fa those of the} user. With the normal jump command wire1 te ^ calculator Ad essier ^ ^ ^ ^^ ^

Let the next command in ^ ^R «henfoj ^ fd bihen but rather allow it

With the normal jump b ^ ^ ^ ^ ^^ ^

caused to designate the next command in s ^ ^R "henfoj ^ f" operand, but rather to

of commands to be taken into account or ranging from »^ P _{d Memory location} ii _e at which the address

in the normal sequence the next command from B «nchm g ρ _{will be} ^^ ^ ^

a part of the memory to be taken into account ^ η Dm S _u ° _n P ^e _s ^ra _e ⁿ _dimensions ⁸ _{Tdizierung be} applied, ui

Decision about which path will be used nnau % & ^^^ _various hints

typically depends on a breach of the one another _{o utes to scl} , _monkeys . _{ln this one}

saved values. Let the following example JJ ^ J * be _used ; _{en those in the} ressierenden Operai

given. 6 _{5 which} may be stored somewhere in memory, uti

_Neg a, .ver jump ^^ I'entfd ™ Ad _re S " SZLZ d ™

For this purpose 2 sheets of drawings 209 529/4

Claims (3)

P. 273) to provide a bit position for a possibly successful change of address. However, this bit position is one next to the an 1. Digital computer with a memory, with a bit positions that are to be provided anyway Arithmetic unit, bit position to be provided with input and output device. With the help of this separately operations and with an accumulator, the bit position to be provided for, it is possible to identify an Indest to replace an operational part for the inclusion of dexregister, with the relevant In- additional bit in the respective command indicates whether the operations to be carried out formations, at least one operand address - the address in the command to be translated absolutely or relatively stands for the inclusion of a one from an io, d. i.e., whether the address is immediately the cell Memory to be read out indicates operands in which the to be read or to be stored Operand address and at least one cherishing operand is located, or whether the respective sign bit position includes, thus valid word initially given provisionally address indicates that the sign bit position is to be added to an address in which the (10) is used as an index register and that the 15 respectively executed instruction is located. However, this can Sign bit position (10) via a link only through the additional provision of a bit position device (8) with a control device (4) take place, which, however, is not ready-connected in all cases which can be made with the memory (2) and with. the arithmetic unit (3) interacts in such a way that it is also related to the structure it is known when a certain prefix 20 of a command word is present (»data processing character bits in the sign bit position (10) a system 3003 «, command list NV Bs 107/1, May 1963), Change in the operand address part to carry out a jump from the sign bit operand address contained in it for storing a word; n dependent on an accumulator Saving a further operand from the memory. However, the jumping only takes place rather (2) in the arithmetic unit (3) causes, in addition, that each running program with a chem according to the operation part of the other command. So it becomes a certain accumulator (5) is a measured operation determined by the respective jump of the respective link of the relevant further operan-next instruction. Often, however, there is that is done by the request in front of the address, using the in an accumulation change information contained in the operand address part contains operations tene operand address is designated. To execute operands from a memory 2. Digital computer according to claim 1, are to be stored thereby. But it is not suitable for this characterized in that the control device (4) is designed as the known structure of a Beaus under consideration is that it is independent of the every word. The respective sign bit also remains unchanged. The object of the invention is to provide a fail to execute on an operand in the way of showing how an accumulator without an initial is, which by means of an operand address extension of its capacity for the designation of is designated. Operands used in a particularly effective way 3. Digital computer according to claim 1 or 2, there can be. characterized in that the linkage 40 is achieved the above-indicated problem device (8) is designed so that it can be used in a digital computer of the type mentioned Change in the operand address part according to the invention in that the sign bits Operand address next to the one in place is used as an index register and that the The sign bit position present in the sign bit position (10) via a linkage setup bit additionally uses a control bit, 45 device is connected to a control device which at least in one command, one in the one with the memory and with the arithmetic logic unit in such a way The command set located in the accumulator (5) has the effect that they are retained in the presence of one is. correct sign bits in the sign bit position a change in the operand address contained in the operand address part 50 for the output Save another operand from the memory in the arithmetic unit, in which accordingly the one indicated in the operation part of the accumulator Operation a link of the relevant 55 lower operands with the operand that The invention relates to a digital computer by prior to the change of address in the operanner with a memory, with an arithmetic unit, with the address part contained operand address be input and output devices and is marked with. The invention has the advantage Accumulator, the at least one operational part for itself that without additional bit positions with the existing Operations to be carried out include the exceptional storage capacity of the accumulator Information, at least one opera, can be sent to a change of address the address part for the inclusion of an off or a change in the contents of the accumulator The operands to be stored in the memory are used for the purpose of controlling those memory characters Operand address and at least one place from which the operand is to be taken from the sign bit position includes. 65 is saved, which is necessary for the execution of the respective In connection with the establishment of a Requested Operation is required,
wrong word it is already known (»digital computing- According to an expedient embodiment of the systems «von Speiser, 1961, Section V.2.4.3, the control device is designed in such a way that it