DE1111430B - Data processing calculating machine for processing information words of different lengths - Google Patents

Description

The invention relates to a data processing computer, which consists of a main storage unit, an accumulator memory, a program unit and an arithmetic unit, the latter being the processed information words consisting of a plurality of character positions, the information words have different lengths (number of characters).

In the case of data-processing machines, it is generally customary to use predefined information words To process number of digits, which means a simplification of the structure of the system in terms of apparatus, but is objectionable with regard to the time economy of the arithmetic processes.

The question of time economy is discussed in computing machines that process data, for example of the following kind:

One does not write a given time span or given number for each arithmetic operation of individual arithmetic steps, but begins with the next arithmetic operation when the previous arithmetic operation is completed, so that there is no rest, the corresponding the maximum time required to perform the operation is measured. At a However, in such a known system, the information words in the main storage unit are not different The number of memory cells is stored, and the machine instructions are also stored with a predefined Address digit number processed.

Another known data processing machine provides that each information word, which comprises a predetermined number of bit positions, is assigned to an address. A multiplication time gives is then derived from the product of an addition time and the number of those used Digits of the multiplier.

It is also known to split information words into two half-words of the same length and if necessary to represent information with two words, i.e. with double the length.

It is also known to have groups of multi-digit decimal numbers, each of which is a decimal number, for example is characterized by 48 binary digits, to be combined into blocks and these blocks by to separate a special character, which also contains 48 binary digits.

A data processing calculating machine according to the invention with a main storage unit, accumulator memory, The program unit and arithmetic unit are characterized by the status of the discussed above Technology in that the main memory

Data processing calculating machine

for processing information words

different length

Applicant:

IBM Germany
International office machines

Gesellschaft mbH,
Sindelfingen (Württ), Tübinger Allee 49

Claimed priority:
V. St. v. America March 22, 1954

Nathaniel Rochester, Wappingers Falls, N.Y.,

Charles Julian Bashe, Poughkeepsie, N.Y.,

Werner Buchholz, Wappingers Falls, N.Y.,

Robert Paul Crago, Philip Everett Fox,

Poughkeepsie, N.Y.,

Jerrier Abdo Haddad, Binghamton, NY,
and Byron Eugene Phelps, Poughkeepsie, NY

(V. St. A.),
have been named as inventors

werk and in the secondary storage unit the end of an information word with a behind the lowest Character position of the same occurring final character is characterized, which is significant for numerical values Information words forms the sign and just like the other character positions of the information word consists of a binary multiplet, and control means responding to this terminator are provided, which, based on this final character, serially remove the bits of the lowest Control character position and then the bits of the other character positions from the storage units and when the final character of the preceding information word in the memory is reached, the removal end of bits and depending on the length of the extracted words the duration of the with respect to Bits and character positions carried out as standard

109 648/196

Limit the computing cycle, while machine commands are used as information words with a specified number of characters are processed.

The final character is preferably used to characterize the memory address of an information word, and the words in the main storage unit are decreasing in the sense and in the accumulator storage unit stored in the sense of ascending memory locations, and it is used in an arithmetic operation one operand is taken from the main storage unit and the other operand is taken from the accumulator storage unit and the result at the locations of the operand to be removed in series in Accumulator storage unit stored.

The accumulator storage unit expediently has twice the number of characters of the average word.

The basic principle of a machine according to the invention can be summarized as follows:

Information words, the multi-digit decimal numbers or from several are used for processing Words consisting of characters or consisting of an operation part and an address part Commands can be. Every digit of a decimal place and every letter of a word and the Digits and the operational parts of an instruction consist of a binary multiplet, where each digit and each letter occupies a memory address.

The words made up of letters and the numerical values consisting of several digits In contrast to the machine commands, they have different lengths and take accordingly different numbers of memory addresses. The information words forming the machine commands, on the other hand, consist of an operational part and a four-digit address and have the same length as each other and thus take five each Memory addresses to complete.

Each information word is followed by a final character, which for numerical values also has the The sign forms, and this final sign, hereinafter referred to as the "field mark", consists of as many binary characters as there are letters and decimal digits. Accordingly included the final character also has a complete address position.

The overall system of the data processing machine is illustrated in FIGS. 1 and 2.

The information, each consisting of seven-bit words, is held on a magnetic tape of the series fed to one of the two information registers 1 or 2, depending on which register is currently used for the Recording of information from the magnetic tape is free. An information recognition level serves the purpose, to determine whether the file was taken from the magnetic tape and stored in one of the information registers Word is a number or a character, e.g. B. a letter character, a punctuation mark, a function mark, such as B. a final character, or a recording character or the like.

Two storage units, which in the case under consideration consist of cathode ray storage tubes, are provided and form the memory or the counter memory. The in the information registers stored information words about the main line circuits or counter line circuit designated selection levels are routed to a specific memory address. Also in in the opposite direction, the information words can vcn the memory or the counter memory via the are forwarded to one of the two information registers. The information register 1 can also return information to the magnetic tape.

The storage in the cathode ray tubes requires the following selection processes:

ίο A certain tube must be selected, furthermore, the memory line must be selected in the selected tube and the relevant position.

The memory contains, for example, 100 cathode ray tubes, while the counter memory consists of only one Cathode ray tube is formed. Each cathode ray tube stores in five groups of lines each seven superimposed bits each 20 words, so that a total of 100 words of seven each Saves bits. Two tubes are paired with one tube on the left and one on the right Tube, summarized. Accordingly, the addresses of the memory get from 0 to 10,000 the address digits of the units and tens provide the group of lines and the location in each line a tube; the address digits of the hundreds, thousands and tens of thousands provide the order number the tube.

In order to select a specific tube and the address in this tube, the stages are provided that with the designations unit selector, memory left-right control and memory beam deflection Marked are.

For the counter memory comprising only one tube, this is called counter memory deflection circuits Stage provided.

The memory address supplied in binary from the data registers is used in the memory address translation stage translated into the decimal system and stored in this form in the memory address register. The memory address counter is a stage that incrementally counts those stored in the memory address register Address increased or decreased, which z. B. is the case with multiplications.

The memory ambiguity eliminator stage removes ambiguities that can result from that the address counter has certain decimal digits for complementary processes and counting processes can reproduce in different ways.

The program counter serves the purpose, e.g. B. to derive a modified address from jump commands. Corresponding levels are provided for the counter memory. The counter memory is basically the same Purposes like memory; so one takes

z. B. the memory the Augenden and the counter memory the adders, if an addition takes place in the adder.

The purpose of the regeneration counter is to keep track of the data stored in the cathode ray tubes To renew information again.

The numbers that are stored in the memory or counter memory as operands in arithmetic operations to be processed, the arithmetic unit stage is provided, which is identified by the designation comparator, Adder, regular complement is designated.

The stage, called the indication emitter, provides carry corrections made by the excess 3 code are conditional.

A timing circuit generates clock pulses and a waveform generator supplies these pulses the control pulses required for control in the machine.

The stage referred to as input-output leads from the memory to the magnetic tape and allows you to select different magnetic tape tracks.

The general control of the machine takes place from the control stage designated with Ruting, whereby this An instruction translator stage and two timer stages for the instruction and for the execution time assigned.

The embodiment of the invention discussed below uses a binary decimal encrypted excess 3 code.

From Table I it can be seen that the zone columns A and B each contain a binary 0 for numeric characters. The particular advantage of the "excess 3" key is that the nine's complement of any digit is obtained by inverting the representation, ie a 0 becomes 1 and vice versa. For example, the number 6 in the "excess 3" system is represented by 1001, ie 6 + 3. By reversing 1001, you get the binary number 0110, which corresponds to 6 or 3 + 3, ie the number 3 in the "excess 3" system. The number 3, however, is the nine's complement of the number 6 chosen above. The advantages of this system will become apparent later when the subtraction is carried out.

Another type of binary decimal representation is chosen in some parts of the arrangement, as described below, namely here the binary digits correspond from right to left to the widths ™, 2 ¹ , 2 ¹ and 2 ² according to the values 1, 2, 2 and 4. This key is hereinafter referred to as binary decimal 1, 2G, 2, 4 keys. The bits of this key are called 1-bit, 2G-bit, 2-bit and 4-bit in the ones place and 10-bit, 20 G-bit, 20-bit and 40-bit in the tens place and accordingly for the higher places . The suffix »G« is used to distinguish the second from the third bit with the same digits 2.

The magnetic tape consists, for. B. from a cellulose tape about 12 mm wide, on its magnetic Layer next to each other in the longitudinal direction of the tape seven parallel tracks are recorded. Other tapes can also be used. According to the key in Table I, six serve Lanes for displaying information and a lane for checking. The records are made from binary bits in all seven tracks simultaneously with a density of about 40 characters per centimeter of tape length. It consider a piece of the tape, and a sensing tape signal, as will be explained, senses a character representation (seven bits at a time) from the tape. The successive characters arrive alternately to character register 1 and character register 2. Each character register contains seven bistable Triggers, each of which has a binary 1 in one state and a binary 0 in the other represents. A character recognition circuit is connected to each character register, which indicates the presence of certain characters in one or both registers. While the one register at the same time takes the seven bits of a character, the other register only senses one bit at a time; h., the seven bits of a character are taken in series and delivered to the main line and arrive into main memory. The character registers act as a buffer between the asynchronously driven Magnetic tape and the synchronously operated storage unit and store in duplicate the 7-bit character representations in the same order, how they appear on the tape.

During the writing process, the recording is taken from the memory bit by bit one after the other (Series operation) and in the character register 1

ίο transferred, and after the character is completely in is stored in this register, it is written to the tape. When writing is the buffer storage only necessary in register 1, since both the memory and the register run at high speed operated opposite the magnetic tape.

The quick storage works electrostatically and

uses cathode ray tubes. This memory not only records the progressive values, but also saves the program, i. H. the commands for the machine. That way won't only the arithmetic, but also the type of arithmetic in a single record with a minimum set by human labor.

The arrangement works in series in terms of both bits and characters, which are fields (i.e. words or numbers) of different lengths. This is obviously also the case when the bits are in parallel and the characters are in serial representation. In fact, the machine can work on fields of any length programmed.

Since the machine has the great advantage that it can process numbers and words of any length, there is no white space, nor is it necessary to abbreviate words, as is the case with machines with limited Field length is required. In addition, such unsightly operations as compound ones become Multiplications and divisions avoided. Furthermore, the computing time depends on the word length, so that 3 * 5 = 15 is calculated faster than e.g. B.

8683544703 7529268141 = 6538736483247207123.

This advantage is especially important because short multiplications are much more common than long ones.

To distinguish one field from the next in memory or on magnetic tape is the end of a field and the beginning of the next field marked by a »field marker«. So would using a + sign as a field marker

z. B. "JA Doe, 101 North 22nd Street" could be written as follows: + JA Doe + 101 North 22nd Street +. A minus sign can also be used as a field marker, as will be described below.
The field mark also serves as a sign for numeric terms and is always to the right of the ones digit. So z. For example, the numbers plus 123, minus 46 and plus 2007 are written next to each other with field characters as follows: +123 + 46 - 2007 +. The plus sign on the far left denotes the end of the number 123 and also serves as a sign for any number further to the left. The + sign is used as a field character for alphabetic and positive numeric terms. Only in the case of negative numbers is a —sign placed to the right of the number.

The field length can be changed by moving just one field character at one end of the field. There are no permanent markings in the

Memory or tape that would limit the field length in some way.

Any desired number of fields can be grouped together to form a record unit to build. The end of a recording unit is indicated by a recording mark like J, which is placed after the field marker, and just like the field length is also the recording unit not limited. It can consist of one, a hundred, a thousand or even more fields. With the The numbers listed above as an example, the recording unit would be written as: +123 + 46 - 2007 ^ ;.

Each digit (seven bits) of a number and each letter (seven bits) of a word are fixed by an address position (starting position, as will be explained later) in memory. Even with multi-digit Numbers and words with multiple letters are given only one address, namely that of the right field marker. As an example, let the address positions of the following digits of numbers be accepted:

Counting

Address position

0000000000000

123 + 46-2007 +

According to the agreement made above, 0351 is the address of the multi-digit number + 123, although the three digits occupy positions 0348 to 0351. (Note that the address 0347 is the Contains field characters for a number further to the left, which is not shown.) Similarly, the Address number 0354 is the address for the number -46, and the address of +2007 is 0359.

When a command arrives at address number 0351, the memory begins to be scanned at Field marker, which is at the address 0351, and from there automatically goes on to the left, i. i.e., she reads the specification in address 0350, then in 0349 etc., until finally the next field character is sensed at 0347 will. In this way, the field character on the right is sufficient to identify the complete number or the word, regardless of the length.

As already stated, the scanning of the numbers that are identified by the address takes place, from right to left, d. that is, the ones digit is sensed first. This always applies when a a single number or a single word is transferred between the memory and the arithmetic unit. It however, it should be noted that when transferring a recording between the tape and the memory through the character register, one item at a time, in the form of a complete recording unit from or onto the tape the order of sensing or writing down from right to left takes place in accordance with normal recording on the tape.

The arrangement is controlled by a command program which, as mentioned, together with the Data, constants and factors are contained in memory. The program will be like everyone else Data needed for the problem is put from tape into memory. With the programs for different problems recorded on different tapes, the arrangement Any problem can be adapted by simply copying the program from the associated tape into memory is transmitted. Since the program is contained in memory, the commands that make up the program form, are encrypted in such a way that they can also be understood by the arrangement. The program consists of groups of numbers and letters (seven bits each) that make up the command. The first seven bits for digits or letters of the command, which are read from left to right in the memory, form the operation part and designate the operation to be carried out. The next four 7-bit pieces of information are digits and form the address part of the command, which tells the machine where to place the commands has to look for the continuation. A program consists of a series of such commands.

In addition to the operation key and the four digits of the address part of a command, a field mark must be added (seven bits) can be provided to separate two adjacent commands. Therefore the field length is each instruction in memory including the field marker to the space of six address positions set. A series of commands can appear in memory in the following form:

30th O i-I CS Memory addresses in ^ o ν ψ OO 000 O SS SSS
000 000
000 field 000 1 001 7+ etc. + B 3 0 0
SS 8+ p brand 0 3 35 information .. 3 9 ν '
Address sen part Operations 40 Ψ part

With regard to the instructions of a program and the positions which they occupy in the memory, a distinction must be made, on the one hand, between the address part and, on the other hand, the address which actually contains the instruction in the memory. To avoid errors, consider the position of the command in memory as the location of the command. In the example shown, 3698 is the address part of command B 3698, while the location of command is 0006, namely the location of the field marker at the right end.

The times in which commands are taken from the memory are the command times. The first 7-bit character (operational part) of the command is sensed during a complete recording cycle (time required for sensing and / or recording a character from and / or into the memory), and this cycle is called command character cycle 1 (abbreviated ICCl) . During the following command character cycles ICC 2, ICC 3, ICC 4 and ICC 5 the thousands. Hundreds, tens and units of the address portion of the command are sensed from memory. The field marker is removed during time ICC 6.

It should be noted that, in contrast to information from the memory, commands in ascending address

sequence. In the example given, the location of the command is 0006, but the information first sensed at time / CCl is the operation part at location 0001. During the four following command character cycles, the address parts at locations 0002, 0003, 0004, 0005 and at time ICC 5 will be sensed the field marker at location 0006. This order is opposite to the convention established above, according to which the data extraction begins at the field marker and continues to the left, which corresponds to descending addresses.

After the command is sensed from memory, the machine executes this command. This time is called the execution time and consists of execution character cycles. These rounds are not set to a certain number, since the latter change with the field length and the length of the recording units. In addition, a regular Routine, namely regeneration, in which the machine does not cycle through instructions or execution executes, no information passed on, but the information contained in the memory is regenerated, such as is detailed below. It should be noted, however, that when a Command the next command is already sensed from memory and then executed; the process repeats itself until the program is completed or the further input of information stops, z. B. in the event of an error.

The memory's 100 cathode ray tubes are divided into 50 pairs, and the two tubes are called the left and the right. The same will be applied to all 100 tubes in the storage tank Deflection potentials are applied so that the cathode ray is always on the same in all 100 tubes Surface elements of the screen hits. However, only one tube is in operation during the regeneration time, d. that is, its electron beam is on at all times. A unit selector consisting of a Decryption matrix, selects that pair from the 50 pairs that is intended for this purpose. One Memory left-right control, which consists of a bistable trigger, decides whether the right or left tube of the selected pair should be in operation.

The deflection circuits also contain triggers and current increment adders, which currents as a function generate from the sum of the receipts. These in turn generate potentials that affect the deflection plates of all cathode ray tubes in the store. So when the cathode ray hits a hits a certain area of a tube, it is shared by the unit selector and the left-right control influenced.

A program counter contains 16 triggers with the associated switching elements and can determine the location of the first command in a stored program. The 16 triggers correspond to four decimal places, and the four triggers in each digit have the weights 1, 2 G, 2, 4, as mentioned above. the 16 output lines of the program counter represent from right to left the 1-bit 2G-bit, 2-bit, 4-bit, Represent 10-bit, 20G-bit, 20-bit, 40-Bii, etc. These lines contain the binary 1 for one potential and the binary 0 for the other potential.

At the time of the command, the machine is controlled by the command timer and works during six command cycles via a routing circuit to process the information on and from the individual parts of the machine.

The potentials for the operation of the program counter reach a during the command time Memory switch for the unit selector, the left-right control and the memory deflection circuits.

The deflection circles receive signals corresponding to the units and tens. The left-right control receives signals corresponding to the first bit

ίο the hundreds, d. H. the 100-bit. The unit selector receives signals corresponding to the remaining three bits of the hundreds (200G, 200 and 400) and furthermore all bits of the thousand digit.

During the first command cycle ICC 1, information is taken from the memory and transmitted to the command translator via the output line, the main line circuits and the main line. This translator is a matrix decoder, and depending on the information received, which is given by the combination of the binary ones and zeros in their 7-bit representations, one of the 30 output lines influences one of several execution timers. The latter control the machine according to the command time in order to enable the required operation.

Assume that the program counter is sensing the command at location 0008. It follows from Fig. 4 that 0008 is the location of instruction A 0037. During the time ICC 1, a signal corresponding to the character A, namely the character at location 0003, is taken from the memory and transferred to the command translator, where it controls the character register 1, as will be explained below. The removal of a character from memory destroys the electronic charge of the character representation, and each time such extraction has taken place, the character must be reintroduced into memory if the memory is to be preserved. Therefore, each character cycle provides a time for extraction and / or inclusion from or into the memory. After the character has been taken from the memory and transferred to the translator and the register 1, the same character is taken from the register 1 during the time ICC 1 and placed in the memory. With a few exceptions, this transfer of a character from the memory to the register 1 and back into the memory takes place in each character cycle in which the data is extracted from the memory. Therefore, this sequence of steps always applies to the following description when removing from the memory, unless a different sequence is specified in individual cases.

During the time ICC 2, a one is added to the program counter so that the digit is transmitted at location 0004, again via the output lines, main line circuits and the main line to the memory address translator. The latter transforms the impulses, which represent a digit in the "excess 3" system, into a series of impulses, the number of which is equal to the real digit. The row pulses go on to the memory address register, which consists of 16 triggers. These triggers represent ones, tens, hundreds and thousands in the key 1, 2 G, 2, 4 like it

ög is also the case in the program counter. During the time ICC 2 , the above-mentioned impulses are now entered in the thousands of the address register. In the following times ICC 3, ICC 4 and ICC 5 , the

109 64S / 196

The program counter is incremented each time, and in the same way, as described above, the characters at the storage locations 0005, 0006, 0007 are now extracted. The address register then contains the address part of the command, that is the number 0037 according to FIG. 4. At time ICC 6, the character at location 0008 is only placed in character register 1, and the character recognition determines that it is a field character. If a character other than a + field character is found here, an error has occurred and the machine stops.

After the command time, the execution time begins, and one of several timers controls the further operations of the machine via the routing circuits. The effective timer in the example under consideration is the ADD timer, because the specification A in the command means adding. As a result, the number in the counter memory is added to the number in the memory, and it follows from FIG. 4 that the number 123456 is at location 0037.

After the start of the first execution cycle, the memory address register is included in the memory address counter transfer. The latter also contains 16 triggers, which are divided into four groups according to the valences 1, 2 G, 2 and 4 are arranged. So now the memory address counter contains the number 0037.

It should be noted that in the 1-, 2G-, 2-, 4-key some digits can be represented in two ways. For example, the number 3 can be replaced by a binary one in the first and second bit or by a one in the first and third bit. As will be explained later, one must provide means to obtain an unambiguous representation of the data. This means consists of the memory ambiguity eliminator, which includes circuitry that does one result in a consistent representation of each information, even if the address counter contains the information in should be presented in different ways.

During the execution time, the signals at the outputs of the program counter are through the Memory switch blocked, and the address number of the memory address counter controls via the ambiguity eliminator and the memory switch, the unit selector, the left-right control and the Deflection circles of the memory in accordance with the value of the number in question.

The selected information is taken from the memory at the time of the first execution cycle ECC 1. In the example under consideration, the information is at location 0037. From FIG. 4 it can be seen that it is the field mark (+) which is transferred to the character register 1 and also to the adder with comparator and regular complement unit. To complete the command, in this case ADD, the word in the counter memory must be taken character by character and bit by bit corresponding to the word in the memory.

The address system of the register memory is similar to that of the memory. In this case it will be however, only two decimal places are required to control the cathode ray deflection. Since the counter memory contains only one cathode ray tube, no selection circles are required.

The starting point counter contains eight triggers, which in the 1-, 2G-, 2-, 4-key have the units and tens represent. This counter contains the address of the word in the counter memory, i. H. the address number the location of the right field marker of this word.

During the first execution cycle ECC 1, the address counter of the counter memory is switched to this address, which is contained in the starting point counter. Therefore both counters are at the same address, ie the address number "/ i", which shows the location of the word in the counter memory.

During the time ECCl and also during all other execution cycles, the number of addresses in the totalizer memory is effective in order to control the totalizer memory deflection circuits via the ambiguity eliminator and the switch. According to the example given above, the field marker is therefore removed from the counter memory during the time ECC 1 and introduced into both the character register 2 and the adder.

During execution cycle 1 (ECC 1), a field mark is transferred from the memory to register 1 and a field mark is transferred from the counter memory to register 2. Both field marks are used together in

so the adder switched. The character recognition circles determine that field marks are present in both registers. Become the field markers not found, there is an error and the machine stops.

The parts such as the comparator, adder and regulating complement circuits (TIC) can receive their information from the memory, via the main line and / or from the counter memory via the corresponding line, in that the information is processed bit by bit according to the required mode of operation. The result is normally fed into the totalizer memory. The comparator, adder and Γ / C circles, together with other parts of the arrangement, can perform not only addition and subtraction, but also multiplication, division, collation (ie comparison of the relative sequence of characters according to Table I) etc. These circles have no memory function, but only process the individual bits during the cycle.

It should be noted that the four bits which contain the numerical position of the field marker representation (see Table I) are the same as the four bits which represent the decimal digit zero. The only difference is in the zone bits. So if two field marks are fed to the adder, its output is zero. Test circuits are provided which determine whether the sum of the binary bits is 0 during the time ECC 1. If it is not 0, there is an error and the machine stops.

In the following it is assumed that no errors have occurred. The field marker is in memory taken and entered in register 1 and is, as in most information circulations, in the Speieher saved again. The field mark, which comes from the counter memory and to register 2 is transferred back to the counter memory for new storage.

At the beginning of the second execution cycle ECC 2 , the counter memory is switched one step further to the address "+1", and the address counter of the memory is switched down by one step, ie in the case considered here to address 0036. The characters at these addresses are taken , as already described above, and fed into the character register 2 and 1 and also into the adder. (It should be noted that the starting point counter is still on the same address number,

namely the address which designates the location of the field marker in the counter memory.)

4 shows that the character in the memory at location 0036 is the numeric character 6, and it is assumed that the numeric digit 2 is in the address "" +1 "in the address""+1". For this reason, information 6 and 2 are introduced into the adder during ECC 2 , where the number 6 is also passed into register 1 and number 2 is also passed into register 2.

The character recognition circles determine that both registers contain numeric digits, and if that Should not be the case, they stop the operation because the addition of non-numeric Information is a prohibited operation.

During the recording part of ECC 2, e . when a character is written into the memory or counter memory, the character 6 is transferred from register 1 back to the same address of the memory from which the character was sensed, and at the same time the output of the adder, in which the sum of 6 and 2 equals 8, placed in the address "" +1 "of the counter memory.

To complete ECC 2 , the memory address counter goes one step down (after 0035) and the counter memory goes up one step (after address "n + 2"). Now comes the next cycle ECC 3, which runs in a similar way to ECC 2. The following cycles take place in the same way until a field marker is sensed from the memory. These field marks are determined by the character recognition circles.

Indeed, at this point the device determines that a complete number has been sensed and processed. During an »ADD« process, the numbers in the memory and totalizer memory can be of the same or different lengths. The information circulation continues until field marks are sensed from both memories. The details regarding a full _ß ADD "-Vorganges and the sequence of individual operations during these rounds is given below.

45

Commands

In order to carry out the solution to a given problem, a program with the correct sequence of instructions must be drawn up and introduced into the machine.

First, a brief description of the commands should be given. As mentioned, every command has one Operation key which is recognized by the command translator and which is used by the individual control circuits Brings effect to run the desired program.

The "ADD" command with the key character A causes the number at the special memory address to be added to the number in the counter memory. If the memory address is Olli, the complete command in the counter memory is AOlli. The machine performs the addition algebraically, and the result is obtained in the counter memory; the number in the memory remains unchanged. The number of digits in the sum is equal to the number of digits of the summand with the largest number of digits. This is explained using the following example:

Spe'c'.i; - vcr and after
addition

+ 0222 +

+0222+

+ 88 +

Counter memory
before addition after addition

+ 101090 +

+ 11-

+ 101312+

+ 0223 + +77+

If the sum exceeds the available space, e.g. B. when adding +5+ to +5+, d. H. in which case the sum has more digits than any addend, then the machine stops and indicates an error. This type of error can be caused by special storage or by shifting be avoided, as will be explained. E.g. if +5+ is added to +5+ instead of +5+ the correct result will be +10+.

The command "DELETE and ADD", abbreviated to RADD, has the key character B and causes the digit or word at the relevant memory address to be transferred to the counter memory, regardless of which value was previously in the counter memory. This command therefore means deleting the counter memory and executing the ADD command. In the case of non-numeric words, they are correctly transferred to the counter, as can be seen from the following example:

Memory before and after
addition

+ 123 +

+ 123-

-SMITH +

Count work memory
before addition | after addition

+456789 +
+9 +

+ 123 +

+ 123-

+ SMITH +

The command »DELETE, ADD and DELETE« in memory, abbreviated to RADDR, has the code character C and causes the transmission of a digit or a word from the relevant address of the Memory in the totalizer memory regardless of what was previously in the latter. Will continue the number or the word in the memory is replaced by zeros, the field marker at the right end in the memory is replaced with a positive sign, and the field marker at the left end remains unchanged. This Command means: deleting the counter memory, executing the addition and resetting the observed Storage locations to zero. Again, non-numeric words will be correct in the totalizer memory transfer. An example is given below.

Storage
before Counter
Storage
before Storage
later Counter
Storage
later + 123-
-123 +
+ Z0295 + + 456789 +
+ 9 +
+ 9 + + 000 +
-000 +
+ 00000 + + 123-
-123 +
+ X0295 +

The command »SUBTRACTION«, abbreviated SUB, has the code character S and causes the subtraction of a number in the special memory address from the number that is already in the counter memory. This subtraction is done algebraically, and the result appears in the counter memory, and the number in the counter

Factory memory is retained. This process takes place as follows:

Memory before and after
subtraction

+ 111 +

+ 111-

flOOO +

Counter memory before subtraction ί after subtraction

+333 + +333 + + 999 + + 012-

+ 222+

+444+

+ 0001-

+ 013-

The command »DELETE and SUBTRACTION«, or R SUB for short, has the character Z and introduces it the number from the memory address in the totalizer memory and reversal of the sign. This command means deleting the counter memory and performing the subtraction. Normally this command can only be carried out with numbers. This command is used for mixed fields given, the non-numerical information is converted into digits. If z. B. the field in the memory is + # 0295 +, the result in the counter memory would be: + 20295-. the more detailed reasons for this will follow later. The normal handling of numbers by this command is on explained in the following example:

From the store

+ 123 +
+ 123-

Counter memory before ι after

+ 456789 +

+ 123- + 123 +

The MULTIPLICATION command (MPY) has the code character M and multiplies the number in the totalizer memory by the number at the particular memory address, and the product appears in the totalizer memory. Before executing the multiplication, one of the factors must be in the register memory. The number of digits in the product is equal to the sum of the digits of the factors. The following process is considered as an example:

multiplier product multiplicand
from memory in count
before : rkspeicher
after multiplication multiplication -2+ + 2+ -J-04 + + 9 + ^J -8 + +72+ -0+ + 8+ +00+ --0 + + 8- +00+ 4-12- + 12+ + 0144-

The command "DIVISION", or DIV for short, has the character /) and means that the value in the totalizer memory (dividend) is divided by the value in the address of the memory (divisor) and the quotient (q) appears in the totalizer memory. The division continues until the number of rows in the quotient is equal to the difference between the digits of the dividend and the divisor. The rest is lost in this type of division. However, if the remainder is needed, the product of the quotient and the divisor is subtracted from the dividend, and for this purpose the dividend is previously stored in memory.

saves. Dividend and divisor must be chosen so that a number whose digits correspond to the divisor and that taken from the left end of the dividend must be less than in length the dividend, otherwise the machine stops. These Condition can be met by adding zeros to the left end of the dividend. the The machine indicates an error by stopping if the length of the dividend and divisor are the same or

ίο if the dividend has fewer digits than the divisor, or if a division with zero is to be made. The division is done algebraically, and dividend and Divisors can have positive or negative signs. The following example serves as an explanation:

dividend
in the counter
Storage
before division divisor
from memory quotient
in the counter memory
after division + 08 + +25 + Machine stops -L 080+ +25 + + 3-4- + 1065 + +25 + + 42- +10 650 ^J - +25 + +426 + + 1065+ + 250+ +4+ + 865 + +25 + Machine stops +0865 + +25+ + 34 -'- +08650+ + 25 + + 346 ^ + 865 + +250+ Machine stops +0865 + + 250+ + 3 + +20+ + 5- + 4- + 20- +5+ --A- + 20- ι g +4+ +20+ + 0+ Machine stops + 00+ + 5- + 0+

The command »ADD in SPEICHER«, or ADD MEM for short, has the key character G and causes the Addition of the number in the counter memory to the number in the relevant memory address. The sum stands then in the memory location under consideration, while the number in the counter memory remains unchanged. It Both positive and negative numbers can be added, and the sum can have any sign to have. The number of digits in the sum is the same as the number of digits in the relevant memory address before executing the command. If the sum causes a carryover from the highest position, the machine stops and an error signal appears. This ADD MEM command can also do this are used to convert numeric fields in totalizer memory to a mixed field, consisting of numeric and non-numerical information, under the following conditions: The numerical Digits must be on the right, i.e. H. stand in the lowest places of the mixed field. The addition cannot go beyond the numerical digits, i.e. i.e. if the sum is a carry over from the numeric to the non-numeric part of the field, then the machine stops and enters Error signal. Another feature of this command is. that with a lower number of digits the digits to the right of the lowest numeric characters in the memory compared to the number of digits in the totalizer memory Excess digits in the counter memory are lost. The mixed field can be both positive and negative Have signs. The ADD MEM command is primarily intended to contain the address parts of a command in the

Change memory. Some examples are given below.

command

Storage
before Counter
Storage
before Storage
later Counter
Storage
later + 030 + + 80 + +110 + + 80 + + 030- + 80- + 110- + 80- + 30 + + 80- + 50- + 80- +30+ + 80+ Machine stops + 30+ + 020+ +50 + + 020 + +30+ + 1020+ +50+ + 1020 + + 30+ +080+ Machine stops + Λ1224 + + 6-J- + / 41230 + + 6+ + / 41224- + 6-L + / 41218- + 6 + + / 11224 + + 225- + / 40999 + + 225- + / 49999 + + 6 + Machine stops + / 40030 + +50080+ + / 40110 + +50080+

ROUND 0001
ROUND 0001
ROUND 0003
ROUND 0002
ROUND 0000

Counter memory before after

+ 16095 +

+ 16095-

+ 16095 +

+ 0032-

+923174+

+ 1610+

+ 1610-

+00+

+ 923174+

The command »RECHTSVERLÄNGERUNG«, LENG for short, has the code character L and causes as many zeros to be added to the right end of the digit or word in the counter memory as are specified in the address section of the command. In other words, the right field character has been shifted to the right by a certain amount, and the resulting empty columns are filled with zeros. If the command LENG 0000 is given, no shift takes place. An example is the following:

The command »TRANSFER«, TR for short, has the code character X and causes the next command, which is in the address section of the transfer command, to be carried out. The address part of the command is transferred to the program counter. Usually, under the control of the program counter, the program proceeds from stage to stage in the order in which the instructions are recorded in memory. However, sometimes a transfer or jump from one stage to any other stage located in a different part of memory is desired. This is achieved by the X command, and an example is given below:

command

LENG 0002
LENG 0003

Counter memory before I after

Program counter
contains place of command command 0224
0230
0156
0162
0168 (any command)
TR 0156
(any command)
(any command)
etc.

-! - 12 + + 12-

+ 1200+ + -12000-

The command »RIGHT SHORTENING«, SHOR for short, has the character / and causes a deletion of as much information at the right end of the number or word in the counter memory than in the Address part of the command is specified. The right field marker is moved to the left until it is next to the remaining information. Here are a few examples.

command

Counter memory before after

+ 1246 + + 1246- + 0001-

+ 12+ + 124- + 000 +

SHOR 0002
SHOR 0001
SHOR 0001
SHOR 0004

The command "ROUNDING", or ROUND for short, has the code character Q and has the effect that a number of digits according to the address part of the command are removed from the right end of the number in the counter memory and the remainder is rounded off. The rounding takes place before the digit is removed by adding 5 to the significant digits to be removed and the carries are carried out. Some examples are given below, from which it can be seen that the sign is unimportant and that the machine rounds positive and negative numbers down.

The command »TRANSFER at PLUS«, TRPLS for short, has the code character Γ and has the effect that the next command from the address part is only executed if the counter memory has a contains positive number or zero. If the counter memory has a negative value, the command causes no transmission, but the following command is then carried out.

The command "TRANSFER WITH EXCEPTION AT THE END OF THE ROW", TREEF for short, has the code character E and works exactly like the transfer command, provided that the row end key of the input-output unit has not been pressed. If the end of the line has not been reached for that unit, the next command is taken from the address part of the TREEF command. However, if the end of the row is reached, the command will not be executed and the next command will be taken from the location immediately after the TREEF command. When sensing the tape, the end of row key for a particular tape unit is actuated by the read command during which the tape mark following the last tape record is sensed. When writing to the tape, the row end button is set when the capacity of the tape reel is reached. This is done with the last write or write and delete command. The key is deleted when a new wrapping command is executed. The "STORE" command with the code character 1 causes the word or number from the counter memory to be stored in memory in the address part of the command.

109 648/196

This part of the memory must previously contain a word or a number of the same length as it also has the information to be saved. The information from the register memory then replaces the old information including the field mark (leading sign). In addition, the information remains in the register memory obtain.

A test determines that the memory has no other field marker at the left end of the Contains information and that no intervening field marks are replaced. If the test is negative fails or if the word lengths do not match, the machine stops and issues an error signal. The field marker at the left end remains unchanged because it has the sign of the one adjacent to the left Contains information. These different conditions are explained using the following example:

Counter before Storage later Storage 6-J-55 + 6 + 12 + + 12+ 6 + 55 + 6 + 12- + 12- 6-55- 6-12 + + 12+ 6 + 555 + Machine stops + 12+ 6 - "- 55 + Machine stops + 2+ 6 + 55 + Machine stops +012+ - "- 5 + 54- Machine stops + 012+

The command »SELECTION OF THE INPUT-OUTPUT UNIT«, SEL for short, has the code character Γ and selects a specific input-output unit (among the tape units or others I / O units). The selected unit is identified by the number in the address part of the command. Such a select command must follow all commands that use the I / O units reverse which, however, do not contain a means for distinguishing the units, e.g. B. Sensing, Writing, writing at the end of the recording, rewinding and transmitting except at the end the series. A selection command given once applies to all subsequent commands that use an I / O unit, until another select command is given. So is z. B. the selection command SEL 0200 is a command which selects tape unit 0200.

The command "SENSE", with the code character R, causes a recording to be sensed from a tape and inserted into the memory. The particular part of the memory is determined by the address part of the command. The tape unit is determined by the last selection command. As already mentioned, the selection command must always be given before the sensing command, ie earlier in the program. The first indication that is sensed is always a field marker and goes to the memory address under consideration, the next sensing goes to the adjacent address, and so on. This operation continues until a record marker is sensed for the end of the recording. There is no limit to the length of the recording unit until the next total capacity of the memory is reached. Since there is apparently no way of predicting when the recording unit will end and when a desired piece of information will be found, the complete program must therefore be planned. The length of each record unit is precisely known and space in memory for each field is pre-allocated. Therefore the exact location of each word is known. When the following commands are given in the program:

SEL 0200
READ 0357

so the machine senses a record unit from

ίο tape off the unit 0200 and put it in the memory, starting with the field marker in address 0357 in memory, and proceeding to the addresses 0358, 0359, and so on until the record mark is sensed from the tape and stored in memory.

The command "WRITE", with the code character W, causes a recording unit to be written from the memory to the tape. The recording is selected by the last selection command and begins at the address of the write command under consideration. The information remains unchanged in the memory. If z. B. the commands SEL 0400 and M) 129 are given, the machine writes a recording unit on the tape unit 0400, which begins (with its field mark) in address 0129 of the memory, and this process continues until a recording mark is reached in the memory is. It should be noted here that a tape from which recording units are sensed cannot be programmed for a write operation in the same pass through the sensing. If a write command for a tape unit is given immediately after one or more sensing commands without the tape having been rewound, the machine stops and an error signal is given. A write command can only be given after a rewind command in response to a sensing operation, or the tape reels must be changed after the series circuit breaker has been turned OFF.
After a complete series of recording units have been written onto a tape, a rewind command is issued which writes a tape mark after the last recording. The »REWINDELBEFEHL«, REW for short, has the code character F and causes a selected tape to be wrapped and the row end button to be reset. The wrapped tape is that specified by the particular address of the previous select command. The machine can perform other operations during the rewinding process until new commands are given for the tape. In this case the machine waits until the wrapping is completed and then proceeds according to the program. The address part of a wrap command is not important because the wrap is determined by the last select command. Any number can therefore be in the address, but REW 0000 is normally written. During a sensing operation, the wrap command is executed even if the tape is not at the end. However, a tape mark is first written during a write process. If, however, the tape has reached its end, the rewind command only deletes the row end key, whereby the rewind takes place automatically, ie without a special REW command.

The command »WRITE AND DELETE IN MEMORY«, WRER for short, has the code character 8 and causes a recording unit in the

considered, part of the memory is written to the tape with the selection of the last selection command. If, however, this part of the memory is deleted, then any information in it can be replaced by an empty symbol with the exception that the field character is replaced by a positive field character and the Recording mark remains undisturbed.

The command "SET FELDMARKE UND SPRINGE", SETFM for short, has the code character V and causes the movement of the left field marker in the counter memory, so that the final number of entries in the field is equal to the number in the address of this command.

If non-significant digits (not zero) in the totalizer memory are deleted by this command, is automatically jumped in the next step. The command »SET FIELD MARKER AND SPRINGE «draw attention to the deletion of important digits, either by stopping them the machine or by a transfer to another part of the program or by both processes. With this command, the significant digits cannot be accidentally lost, and the Calculation cannot continue without the operator of the machine or the programmer it knows. An example of this is given below:

location

0084
0090
0096

command

SET FM 0003.

TR (-) any address

STORE any address

If the program counter is in the command address 0084, the following can occur:

Counter memory
before i after

+ 12 +
4-00123 +
+ 12345 +
+ 10000-

+012+
+ 123 +
+345 +
+ 000+

Program goes to

Command at 0096 (jump 090) Command at 0096 (jump 090) Command at 0090
Command at 0090

The command »KEINE OPERATION«, or NO OP for short, has the code character 0 (zero) and nothing is carried out. The program counter only goes to the next program step. Anyone can use this command Address as it is irrelevant. You use this command when a new Command given in a program and the old command should be removed.

The command “COMPARE AND JUMP”, COMP for short, has the character K and compares the word or the absolute value of the number (M) at the relevant address in memory with the word or number (A) in the counter memory; the program counter jumps:

1. by no command if (M) is greater than (A) ,

2. a command if both sizes are the same, and

3. by two commands if (M) is less than (A) .

In the case of non-numerical data, the comparison corresponds to the collation sequence according to Table I. For When executing this command, the numbers or words in the totalizer memory must be positive. With negative Values are compared with the tens complement. This is happening because negative numbers in the form of tens's complement are stored in the counter memory. But when If there is a negative number in the memory, the K command neglects the sign and compares only the absolute amounts. The data in the totalizer memory and memory must have the same number of digits otherwise the machine will stop giving an error signal. The command is used to to compare the result in the counter memory with a number in the memory and then to close the program to another part, depending on the three conditions above. The following Example serves to explain:

location Operation part Address part 0108
0114
0120
0126
0123 RADD
COMP
TR
TR
TR 0216
0224
0138 (M)> (A)
0168 (M) = (A)
0018 (M) < (A)

The command »EMPTY«, with code character M, affects the number in the totalizer memory in the following way: If the address part of the empty command is 0000, the command replaces every zero to the left of the highest significant digit in the totalizer memory with the empty symbol. The number of digits in the remaining number, including spaces, remains unchanged.

If the address part of the blank command is a number other than 0000, -35, the operation is as follows: The symbol for a comma is between two adjacent digits of the number in the counter memory introduced. The address part of the empty command specifies the position of the decimal part by designating the Digits to the right of the decimal point. The number of characters in the remaining field increases by one. Each zero to the left of the highest significant digit is replaced by the symbol for a blank column, provided that that the zeros are to the left of the decimal point. The places to the right of the comma remain unchanged.

In the special case that all digits to be replaced by spaces are zeros, the entire Number replaced by space including any commas and zeros to the right of it. the The number of blanks is one greater afterwards than there were zeros before.

If the numerical value in the address part of the empty command is greater than the number of digits in the counter memory before the empty command (without field marks), then the machine stops and signals one Failure.

The purpose of this command is to prepare for printing or writing. This can be carried out by an input-output unit. The prepared information can can be sensed from memory or tape, or if the prepared information is on tape the spool of tape can be used to actuate the printing device. This empty facility is important when describing checks where the decimal point is necessary and requires empty columns before the first significant digits in

Mark amounts are written. In the example explained below, the letter b shows the Empty columns. These are of course not written on the check form.

address, stops the machine and issues an error signal. The following examples should explain this:

command Counter memory before later + 00120 + + bbl20 + Empty 0000 + 00120- + bbl20- Empty 0000 + 00120 + + bbl. 20 + Empty 0002 + 00120- + bbl.20- Empty 0002 + 00120-L + b.0120 + Empty 0004 + 001204- + .00120+ Empty 0005 + 00120 -J- Machine stops Empty 0006 + 00000 + + bbbbb + Empty 0000 + 00000+ + bbbbbb + Empty 0001

Counter before Storage later 5 memories 53 + 361 + 53 + 243 + + 01243 + 53 + 361 + 53 + 243- + 01243- 53-361 + 53-243 + + 01243 + 53 + 361 + Machine stops ¹⁰ +243 + 53 + 361 + Machine stops +43 + (selected Address)

The command »STERNCHEN«, abbreviated ASTK,

The command »SAVE WITHOUT FIELD MARKERS«, abbreviated to ST NFM, has the code character 5 and causes the word or the absolute amount of a number in the counter memory to be converted into a specific location of the store to be transferred with

whose code key is 9, causes the same operation 20 exception of the field markers, which in the absence of the empty command with the exception that the character. This command is similar to the command »SAVE for an asterisk in place of the symbol for CHERN AND CHANGE THE FIELD MARKERS«, an empty column appears. This command cannot only stay here in memory for those positions where negative numbers are used. This characteristic is normally the field markers would be, unimportant if the calculated information is immediately disturbing. The address of a field without a field mark is before the first significant digit of the individual calculated position, at which the field mark would appear, neten amounts appear, is converted into printed form, in which asterisk is desirable
are. This measure prevents the forgery of
Checks, etc. 30

The command »SAVE AND CHANGE THE FIELD MARKERS«, abbreviated to STAFM, is its code character 2 causes the character in the counter if the command ST AFM has been given instead were. With this command you can delete field marks in the memory, as the following example shows:

command

Counter memory

Save before after

+ / 41320 + + / 41620-

+ 45 + 78 + + 45678- ^ + A1320- ¹ - + £ 1320-5- + / 41320-; + / 44567-

memory and its right field mark (sign), STNFM 0124 + 6-which is stored in the memory, identifies the address part 35 STNFM 0124 + 6— of this command. At the left end of the ST NFM 0122 + B + newly stored word in the memory, a positive ST NFM 0126 +4567+ tive field marker is set with the exception of the cases in
which already have a field marker at this point. The command »SAVE POSITIVE FIELD

Then this field marker is left unchanged. 40 BRANDS «, abbreviated to ST PLS, with the code This Command changes the field marker arrangement in character 6 creates a positive field marker in the beSpeicher. Examples are the following: appropriate memory address. You can do that with this

Change the sign of a number or word in memory. There is accordingly the command »SAVE NEGATIVER FELDMARKEN «, short ST MIN with the code character 7, and this saves a negative field marker in the relevant memory address.

If the machine is to be stopped automatically, then the right field marker can be any Command can be made negative. In this case the machine reads the command and stops before executing same. There is no further command for this

In this example it is assumed that this is necessary, you only need a negative field marker The selected address of this command is the address 55 to the right of the right digit of the address part of the which contains the right digits, d. H. to set the address of the command.

of 7. All of the operations described above are performed in

The command »SAVE BETWEEN THE encrypted form given in the machine so FELDMARKEN «, abbreviated to ST BFM, with which they can be recognized. The machine processes Code character 3 causes the right part of the line to 60 ten numeric characters, 26 alphabet characters and chens in the counter memory between the field marks a number of special characters, such as point, etc. stored at the selected memory address Every operation that the machine performs is will. The character in the counter memory must be determined by a character longer. As already mentioned, as the space that is pre-fed in the memory address, the command translator has a selected time frame is. If the number of characters in the counter is 65 according to the key characters received, and Werkspeicher with the exception of the field markers this gives control impulses to the machine to the or less than the number of characters available between the sequence of operations during execution of the desired to monitor the field markers at the selected stored command.

Counter Storage later Storage before | 3 + 12+ + 12 + 34567 3 + 12- + 12- 34567 3-12 + + 12- 2-567 (selected Address)

Table I shows all of the key characters and how they work. The programs on the tape need to be in the form of a recording unit with a recording mark at the end of the program. Then this can be introduced into the memory.

It has already been mentioned that the machine works in series and therefore the fields any Can have length. This means that the counter memory must also be adapted to each number of digits. the effective length of the counter memory adapts itself to the number of digits received. For example, if a Add two- and three-digit numbers, then the sum is either three or four-digit. If several If three-digit numbers are added, the sum can be five or more digits. So there are some Rules for adapting the totalizer memory to the number of digits must be observed by the programmer must know exactly in order to provide a corresponding space in the memory for it. When counting the Digits are all zeros to the left of the significant digits to be counted.

The rule the machine follows when adding or subtracting two numbers is to that the result in the counter memory has a number of digits equal to that of the larger individual number. If the result exceeds the number of digits in the individual numbers, the machine stops.

To avoid this, one has to add enough zeros to the left of the numbers for ensure that there is sufficient space in the totalizer memory. Then you can do any addition you want or perform subtraction. So stops z. B. the addition 9 + 6 the machine, while 09 + 6 or 9 4- 06 delivers the correct result 15. Fields that are too large do no harm: 009 + 6 = 015.

Adjusting to the correct number of digits can be done when recording the original values. On the other hand, zeros can be added to the left of the digits in the counter memory using the »SET FIELD MARKS «add.

With the commands R ADD or R SUB you do not need to observe this rule because the result has the same number of digits in the register memory (with the exception of the sign for R SUB) as it was from Memory was sensed.

Similar simple rules apply to multiplication and division, as do programming allow these types of calculations.

In the described embodiment of the invention, the machine has a capacity of 100 characters. This allows 98-digit numbers (without field marks) to be processed for ADD and SUB and 47-digit numbers for MULT and DIV. This capacity is usually sufficient.

Claims (3)

PATENT CLAIMS: 1. Data processing calculating machine with main storage unit, accumulator memory, program unit and arithmetic logic unit, the latter consisting of a plurality of character positions Processed information words, whereby the in- Formation words have different lengths (number of characters), characterized in that in the main storage unit and in the secondary storage unit the end of an information word is characterized by a trailing character which occurs after the lowest character position of the same, which forms the sign of important information words for numerical values and just like the other character positions of the information word a binary multiplet, and control means responsive to this final character are provided which, based on this final character, control the extraction of the bits of the lowest character position and then the bits of the further character positions from the storage units and, when the final character of the information word preceding in the memory is reached, the extraction end of bits and, depending on the length of the extracted words, limit the duration of the computation cycle carried out in series with regard to bits and character positions, during machine commands le are processed as information words of a given number of characters. 2. Machine according to claim 1, characterized in that the final character as a characterization the memory address of an information word and the words in the main storage unit in the sense of decreasing and in the accumulator storage system in the sense of ascending storage locations are stored and in an arithmetic operation in which the one operand is the main storage unit and the other operand can be taken from the accumulator storage unit, the result at the points of the one to be removed Operands in the accumulator storage unit is stored in series. 3. Machine according to claim 1 or 2, characterized in that the accumulator storage unit has twice the number of characters of the average word. Considered publications:
“Proc. Inst, of Electrical Eng. ", Part. III, 99, 1952, No. 68, pp. 99-106; “Proc. of the I. R. Ε. ”, 1948, No. 12, pp. 1452 to 1460; 1953, October, pp. 1275 to 1287, 1262 to 1280; “Proc. IRE ", 1952, January, pp. 12-29;
"Automatic calculation plan production with program-controlled machines", Birkhäuser, Basel, 1952, pp. 6 to 7; "High Speed Computing Devices," McGran Hill Book Comp., New York, 1950, pp. 268-269, 289 to 294; "The Annals of the Comp. Lab. of Harvard University ", Vol. XXVII, Cambridge, 1951, p. 195 up to 230; "Mathematic tables and other aids to computation", Vol. 3, 1948, No. 24, pp. 286 bis. 295; 1950, p. 164 up to 175. 1 sheet of drawings