BACKGROUND OF THE INVENTION

Applications of minicomputers have grown in sophistication and complexity to the point where their requirements for memory addressing capability have far exceeded the number of bits available in the computer's word length. This word length is typically 16 bits which limits the main memory of the computer to 32,768 words with infinite indirect addressing or 65,536 words if only one level of indirect addressing is allowed.

Because core memory technology had not provided economically attractive methods to provide large memory storage within a computer, past efforts were directed at mass memory devices such as magnetic discs or magnetic tape units. These mass memory devices could supply millions of bits for a small fraction of the cost of an equivalent core memory storage area. However, these devices are slow. Sometimes complex software operating systems were employed to make these devices appear to the user as an extension of the main computer memory. This technique, referred to as virtual memory, was slow and difficult to use efficiently.

Another obvious option available to the designer was to expand the word length of the computer and rearrange the architecture of the machine so that a longer word length was available for memory addressing. This was the most direct solution but also the most costly. For example, the registers, buses and the memory word length in a computer might have been expanded from 16 bits to 20 bits to increase the memory addressing capability from a maximum 65,536 to over 1,000,000 words. This option was especially unattractive when it was desirable to maintain software compatibility with previous computers because of the time and money that had been invested in building an extensive software library.

Another option that was available to the designer was bank switching. One apparatus for bank switching is described in patent application Ser. No. 360,286, entitled "Polymorphic Memory Apparatus and Method", filed by Robert J. Frankenberg on May 14, 1973. Bank switching was a cumbersome technique because of the requirement of an additional memory controller for each bank used in the system.

Also available to the designer was the use of translation tables. Prior art designs using translation tables lacked the versatility required for viable and sophisticated software operating systems. In a translation table system it is desirable to transfer words between memory areas selected by different translation tables, to execute subroutines in a memory area addressed by another translation table and to transfer program control to a program in a memory area addressed by another translation table. Furthermore, it is desirable to make some physical memory locations accessible to all users while allowing each user to also have a portion of physical memory reserved for his exclusive use.

SUMMARY OF THE INVENTION

In the preferred embodiment of the present invention, memory addresses from the central processing unit (CPU) are composed of a logical page address and a word address. The logical page address is used to address registers in a translation table containing physical page addresses. Physical page addresses differ from the logical page addresses in that physical page addresses contain more bits and address a specific area in physical memory. The physical page address is appended to the word address to form the complete physical memory address of a physical memory location that will be loaded into or read from by the computer. Any logical page may be translated to any physical page by loading the appropriate information into the translation table registers. Furthermore, since multiple translation tables are provided, the ultimate physical address selected will also depend on the particular translation table enabled.

Translation Table Enabling circuitry provides the capability of selectively altering the state of the translation tables while remembering their previous state. This circuitry is useful in executing instructions which allow a memory user to load into and read from memory location addressed through other translation tables. Other instructions permit permanent transfer of program control from a program in a memory area addressed by a first translation table to a program addressed by a second translation table. This is accomplished by disabling the first translation table, enabling the second translation table, and transferring program control to a specified address in physical memory through the second translation table. Other instructions provide for temporary transfer of program control from a main program in a memory area selected by a first translation table to a subroutine in a memory area selected by a second translation table by disabling the first and enabling the second translation table. The return to the main program is accomplished by reenabling the first translation table and disabling the second translation table upon completion of the subroutine.

The present invention also provides a register referred to as the fence register for storing a logical base page word address. Whenever the logical page addressed is the base page, the contents of this register are compared to the word address from the CPU. Based on the results of this comparison, the logical base page is translated to a physical page contained in the translation tables, or the address from the CPU is used to directly address a physical memory location. By directly addressing physical memory with the address from the CPU when indicated by the results of the comparison, a common area in physical memory is provided which is independent of any translation table base page register contents.

DESCRIPTION OF THE DRAWINGS

FIG. 1 shows the memory address format of a minicomputer for use with the preferred embodiment.

FIG. 2 shows the memory reference instruction format of a minicomputer.

FIG. 3 is a diagram showing the address translation feature of the preferred embodiment.

FIG. 4 is a diagram which describes the address translation provided by the preferred embodiment.

FIG. 5 is a block diagram constructed according to the preferred embodiment.

FIG. 6 6A-6D is a schematic diagram of logic circuits of the preferred embodiment.

FIG. 7 is a simplified diagram of circuits useful for transferring data between memory areas accessed by different translation tables.

FIG. 8 contains a flow chart of the operation of transferring data between memory areas accessed through different translation tables according to the preferred embodiment.

FIG. 9 9A- 9D is a logic diagram constructed according to a portion of the preferred embodiment.

FIG. 10 10A- 10B is a logic diagram construction according to a portion of the preferred embodiment.

FIG. 11 is a flowchart showing the logical steps for detecting a read violation in the preferred embodiment.

DESCRIPTION OF THE PREFERRED EMBODIMENT

FIG. 1 shows the memory addressing scheme of a minicomputer wherein the memory address is divided into a page address 20 and a work address 10. The size of the page and word addresses was arbitrarily decided at the definition stage of the instruction set for the computer and is not related to a physical partitioning of memory or to the scope of the invention.

Regardless of which page is selected, the word address will be between 0000₈ (0000₁₀) and 1777₈ (1024₁₀). As one increments the memory address the bits 0-9 will equal 0000 every 2000₈ or 1024₁₀ words which defines the page size for this embodiment. These page boundaries are shown by chart 30 in FIG. 1. The usefulness of the paging concept is shown by looking at the format of the memory reference instructions in FIG. 2. The operation code 53 specifies the operation to be performed on the word read from memory. Typical operations are ADD, AND, or XOR the contents of a memory location with the contents of a register. The word address 57 of the data to be read from memory is specified by bits 0-9 of the instruction. The page address of the word to be read from memory is determined by bit 55 and will be either the current page, defined as the page the instruction is on, or the zero page. Therefore, to presently available CPU' s using page address schemes, the zero page addresses are extremely important because they are directly accessible by the CPU from anywhere in memory. Because of this restriction to directly address only the current page and zero page locations, minicomputer programmers have devised various programming schemes and segmented their programs to minimize this limitation. Adapting to these limitations, the present invention makes more pages of physical memory available to be selected while leaving the immediately addressable memory space at the previous 32 logical pages. The memory user has the option to change the contents of the translation table registers or use another translation table to address the memory area not addressed by the presently enabled translation table.

Referring to FIG. 3, logical page address 60 specified by the computer is used to address one of the 32 registers in translation table 64 which provides a 12-bit word containing 10-bit physical page address 65 and bits 66 and 67. Bits 66 and 67 indicate whether the page is to be read-or write-protected when the memory protection circuitry is enabled. The 10-bit physical page address 65 is appended to the 10-bit word address 57 to form 20-bit address 69.

FIG. 4 also shows the operation of the translation table. The 5 bits of logical page address 60 address one of the 32 registers in translation table 64; the output of addressed register 64 is 10 -bit physical page address 66. Even though there are 1024₁₀ pages of memory selectable when using 10 bits for page addressing, only 32₁₀ pages are selected by a single translation table at any one time because page addressing is still limited by the 5 bits of logical page address 60 from the CPU of the preferred embodiment.

Referring to the block diagram of FIG. 5, the system is first described after initial power up. The state of 4 major control signals disables the memory expansion system, and all CPU to memory communication is carried on as if the preferred embodiment of the present invention is not installed. (For a description of the CPU and general memory operation refer to patent application Ser. No. 360,286 entitled "Polymorphic Memory Apparatus and Method", filed by Robert J. Frankenberg on on May 14, 1974. CPU 150 is also described in the HP Journal issue of October, 1974.

Signals 101, 102, and 103, are used together to command the system into one of the 8 states referred to as Q0 through Q7, which are described in Table 1. At the time of initial power up signals 101, 102, and 103 88 are all in the logical one state which is Q7 in Table I and represents the NOP, i.e., no operation, state of the system. These three signals cause enabling and addressing circuitry 105 to ignore all inputs and to maintain signal 108 in the logical zero state. (Signal 108 is preset to the logical zero state when power is first applied.) Therefore, multiplexer control gate 109 is disabled. Multiplexer control gate 109 generates signal 110 in the logical zero state which selects the B input of multiplexer 111 to be gated to physical memory address register 182. CPU 150 may only access the lowest 32K of physical memory when the B input of multiplexer 111 is selected since the 5 most significant bits of the B input of multiplexer 111 are grounded.

Still referring to FIG. 5, the translation table registers are addressed consecutively starting with 0 for the first register in translation table 130 and ending with 127₁₀ for the last register in translation table 145. The translation table registers are loaded with address information as described below. First, the address of the register to be loaded is sent by CPU 150 via s-bus 155 to table address register 106 within enabling and addressing circuitry 105. It is clocked into table address register 106 by the simultaneous execution of system state command Q5 as described in Table I. (Special configurations of data on S-bus 155 and their effect in conjunction with the execution of the state commands of Table I is shown in Table II.) Next, the physical page address and write/read protection information is placed on s-bus 155, lines 0-9, 14 and 15 respectively, and stored into the selected translation table register by the simultaneous execution of stage command Q1, as described in Tables I and II. The above procedure is repeated until all translation table registers are loaded with the desired information. Special instructions which streamline operation of the system are described in Appendix A.

Referring to FIG. 6A, it is necessary to transmit a control word to state register 268 and execute state command Q5 to begin logical page address translation by the preferred embodiment. This is accomplished by simultaneously presenting a control word on s-bus 155, the logic state of the bits selected per Table II, and executing system state command Q5, described in Table I. For instance, to begin translation of logical page addresses from the computer through translation table 130, the s-bus control word must have all bits in the logical zero state except s-bus 14. The decoding of system state command Q5 by state command decoder 201 causes flip-flop (FF) 205 to be set and FF 210 to be reset. The Q output of FF 205 is coupled to gate 215, producing a logic one output therefrom. The logic one output from gate 215 produces a zero output from gate 220. The output of gate 220 is coupled to inverter 225 to produce signal 227 in the logic one state. Signals 227 and 226 are the enable signals for translation tables 130 and 135 respectively. FF 210 is reset and generates signal 108 in the one logic state. Therefore, logical page address translation through translation table 130 is enabled.

Referring now to FIG. 5, logical page address 160 from CPU 150 is gated to translation table 130. The contents of the translation table register addressed by logical page address 160 is output to bus 115 and is coupled to the A input of multiplexer 111. The signal 108 is in the logical one state, and since the output from comparator 165, usage discussed below, is a logic one state at this time, gate 109 produces signal 110 in the logical one state. Multiplexer 111 selects the A input, the translation table register contents on bus 115, to be output to the 10th through 19th bits of physical memory address register 182. The word address 125 is coupled to the 0 through 9th bits of physical memory address register 182.

Still referring to FIG. 5, when a read command 186 or a write command 187 is output from CPU 150 to memory timing and control 180, the clock signal on line 123 will clock both the output from multiplexer 111 and word address 125 into physical memory address register 182. Also, timing and control signals 184 are output to physical memory 190. Execution of read command 186 reads the data from physical memory 190 into memory data register 192. Transfer of data from memory data register 192 to CPU 150 will be discussed below with protection circuitry 199. Execution of write command 187 is preceded by storing the word to be written in memory in memory data register 192.

Assuming that all the translation table registers are loaded with the desired physical page addresses, CPU 150 may now access memory locations through any one of the translation tables or may disable the translation tables by executing system state command Q5, described in Table I, and use logical page address 160 with word address 125 to directly access the lowest 32 pages of physical memory.

Referring to FIG. 6A, circuitry of the preferred embodiment is shown which saves the enabled state of the translation table in use upon receipt of any CPU interrupt. This circuitry then selects the translation table addressing the memory area containing the program to service the interrupt. Subsequently, the translation table that was in use at the time of the interrupt may be restored to the enabled state. The contents of state register 268 are stored in save register 250 in response to signal 265 for use in subsequently restoring those contents in state register 268. Also, translation table 130 is enabled through gate 266. Signal 265 is also coupled to gate 220, which provides a logical zero to gate 225, thereby generating signal 227 and selecting translation table 130. Circuitry in CPU 150 responds to signal 265 and causes the next instruction to be fetched from the memory location address having the same binary value as the select code of the device interrupting. FF 205 is coupled to gate 215 which supplies a low logic level output to gate 200. Gate 220 will now continue to provide a low output after signal 265 goes to the logical zero state. FF 260 is also set by signal 265.

After an interrupt has been processed by the CPU, the interrupt program will then restore the original state register contents. The save register 250 is read onto s-bus 155 by executing system state command Q6, described in Table I. The state command decoder 201 provides signal Q6 to gate 286, which causes signal 285 to read the contents of save register 250 onto lines 15 and 14 of s-bus 155. (FF 279 is assumed set at this time. FF 279 is used when it is desired to read the violation register, described later, instead of the status register. This is accomplished by executing state command Q4 immediately prior to the execution of state command Q6, thereby setting FF 279 which then disables gate 286 and enables gate 272 instead.) CPU 150 then shifts the bits read from save register 250 so they are positioned on lines 9 & 8 of s-bus 155. The CPU then merges these shifted bits with a data word having a logical one at bit 14 and then reads them back onto s-bus 155 while again executing system state command Q5, described in Table I. The information on s-bus bits 8 and 9 is clocked into state register 268, and the original state of the translation tables is thereby restored. Only two translation tables are controlled by the circuit of FIG. 6, but any number of translation tables could be controlled in this way by adding additional storage elements to both state register 268 and save register 250.

Another circuit shown in FIG. 6 is particularly suited to applications where it is desirable to alternately enable two different translation tables for the transferring of blocks of data from a memory area selected by one translation table to a memory area selected by another translation table or to execute subroutines in a memory area selected by another translation table. Some elements of this circuit have been extracted from FIG. 6 and are shown in FIG. 7. Once FF 205 is set, signal 227 will be output in the logical one state from gate 225. Translation table 130 is enabled and translation table 135 is disabled. To alternately enable translation table 130 and translation table 135 it is necessary to execute system state command Q3 as described in Table I. Operation of this circuit is described below.

Still referring to FIG. 7, assuming that FF 260 is reset and FF 205 is reset, gate 220 has a logical zero output, and gate 225 has a logical one output. Therefore, translation table 130 is enabled. Execution of system state command Q3 places a logical one level from the output of gate 345 on the J input and a logical zero level on the inverting K input of FF 260. Clock signal 346 will cause FF 260 to toggle to the set state. This produces a low output from gate 215, thereby disabling translation table 130 through gate 225 and enabling translation table 135. Memory is subsequently accessed through translation table 135. Execution of system state command Q3 again will restore translation table 130 to the enabled state and disable translation table 135. The usefulness of this circuitry is shown by the flowchart in FIG. 8.

The flow chart procedure can be implemented in a software program, a microprogram, or hardware circuitry. The preferred embodiment uses a microprogram because of the speed of its execution. (All microprograms of the system instructions are listed in Appendix C. Definitions of the microninstructions are found in Appendix B.) There are two entry points. Assume for a moment that translation table 130 is enabled and a program in memory area 335 is being executed. If it is desired to read from memory are 335 through translation table 130 and write into memory area 340 through translation table 135 then entry point 2 is used. If entry point 1 is used words will be read from memory area 340 and written into memory area 335.

The above descriptions have been described in terms of accessing data in a memory area address by another translation table. However, these circuits are also effective in the execution of subroutines in memory areas addressed by other translation tables. Particular instructions are listed in Appendix A.

Referring to FIG. 5, the method by which word addresses of the logical base page are selected either for translation through the translation table, if it is enabled, or for direct addressing of physical memory will now be described. Before system operation is begun, fence register 168 is loaded by executing system state command Q5, described in Table I, while reading data, in the format described in Table II, onto s-bus 155. Thereafter, when the system is enabled and logical page address 160 addresses the base page, comparator 170 outputs an enable signal to comparator 165. Comparator 165 then compares word address 125 to the contents of fence register 168. If the word address 125 is within the logical base page area selected to directly address physical memory, the output signal from comparator 165 will be a logical zero level. This will disable gate 109 and the B input of multiplexer 111 is gated to the physical memory address register 182. If the comparison of logical work address 125 with the contents of fence register 168 does not indicate the word address is to directly address physical memory, then gate 109 will not be disabled and the translation table register contents on ME-bus 115 will be gated from the A input of the multiplexer 111 to physical memory address register 182.

Therefore, the base page word addresses selected for direct addressing provide a common area of physical memory which is accessed by every memory use regardless of the translation table register contents. The remaining word addresses on the logical base page are translated by the translation table registers and may be assigned to access any area of physical memory 190 by storing various physical page addresses in the registers of translation tables 130, 135, 140, and 145 which are addressed when logical page address 160 addresses the base page. The address chosen as the dividing point between the dedicated and common areas is chosen by the system programmer.

The protection circuitry 199 shown in FIG. 5 does not check for write violations until CPU 150 provides signal MPCND 500 in the logical one state. Any instructions which attempts a write or jump operation has a microinstruction step where it executes MPCK (described below in Appendix B) and reads the suspect address onto s-bus 155. When a protected memory area is being addressed, execution of the microinstruction MPCK by CPU 150 provides signal 500 to protection circuitry 199. Detected attempts to violate a protected area of memory will result in the generation of signal 505 to CPU 150 which will inhibit the instruction.

Referring to FIG. 9A, violation register 550 stores logical page address 160 and the type of violation that has occurred. The violation register clock is generated by signal 500, signal 514, or signal 186, depending on the type of violation. Once a violation has occurred, the generation of signal 505 sets flip-flop 506 which prevents the violation register from being altered until reset by CPU 150.

The four classes of violations are read-protect violation, write-protect violation, base page violation, and priviledged instruction interrupt and are discussed more fully in the instruction descriptions in Appendix A. These violations are indicated by signals 511, 512, 513 and 514, respectively. The state of these signals is saved in violation register 550 when the violation is detected. The format of the word saved in the violation register is shown in Table III.

The work "map" is synonomous with translation table and is used in the tables, program listing comments, and the instruction descriptions. The work mapped indicates the logical page address has been translated to a physical page address by the translation tables.

Further information regarding the state of the system is available to CPU 150 from status register 279, shown in FIG. 6B. It provides status information in the format shown in Table III in response to system state command Q6. Signal 285 enables gates 278 and 280 to read status information onto s-bus 155. Referring to FIG. 6A, the state command decoder 201 output is coupled to gate 286 which provides signal 285 only if FF 279 has been reset by the execution of system state command Q4, described in Table I.

All attempted violations of protected memory or attempts to execute prohibited instructions are detected and inhibited at the time the instruction is begun except for the read protect feature. Modern CPU architectures use look ahead read circuitry to read the memory location most likely to be requested next before a read command is issued. Therefore, the violation circuitry does not generate an interrupt until CPU 150 actually requests data. Referring to FIG. 5, this is indicated by the generation of signal 510 by CPU 150 to memory data register 192 via gate 194. When this request for data occurs and read protection is indicated by read protect bit 516 protection circuit 199 will generate signal METDIS 188 in the logical zero state. This inhibits reading the contents of the protected memory location back to CPU 150 via s-bus 155. The detailed diagram of this portion of the protection logic is shown in FIG. 6D. FIG. 11 is a flow chart which shows the logical steps performed by the circuitry to determine whether a read violation is being attempted.

                                  TABLE I__________________________________________________________________________SYSTEM STATE COMMANDSCOMMAND          SYSTEM USAGE__________________________________________________________________________Q0  1 Enable reading of SYS/USR map to S-bus (per register   106, Bit 5)  2 Store S-bus into maps (per register 106, Bit 7)  3 Relative map address specified by register 106,   Bits (0-4)Q1  1 Store S-bus into maps (per register 106, Bits 5,6)  2 Map register address specified by register 106,   Bits (0-4)Q2  1 Enable maps to S-bus (per register 106, Bits 5,6)  2 S-bus bits (10-13) are always low  3 Map register, address specified by register 106,   Bits (0-4)Q3  1 Select opposite program mapQ4  1 Set "Status Command" Flag through next microprocessor   cycle (See Q6)  2 Reset to currently selected program mapQ5  1 Store S-bus into;    a) MEM State Register (2-bits)    b) MEM Fence Register (11-bits)    c) MEM Address Register (8-bits)  2 Above registers selected by S-bus bits (15-13)Q6  1 Enable status data onto S-bus;    a) Normally reads Violation Register    b) If preceded by Q4 command, Status Register  will be Read.Q7  1 No microcode specified  2 NOP state__________________________________________________________________________

              TABLE II______________________________________STATE    S-BUSCOMMAND  BITS     SIGNIFICANCE OF S-BUS BITS______________________________________Q0  0-9      S-bus bits 0-9 represent address              data being stored into a selected              map register.    14       S-bus bit 14 represents the Write              Protect bit.    15       S-bus bit 15 reresents the Read              Protect bit.Q1           S-bus bits 0-9 represent data              being stored into a selected map              register.             S-bus 14 represents the Write              Protect bit.             S-bus 15 represents the Read              Protect bit.Q2  0-9,14,15             S-bus bits represent data being read              from the selected Map register.    10-13    Always in zero state.Q3  N/A      No special significance.Q4  N/A      No special significance.Q5  15       Allow privileged operation.    14       Load State Register 268 per S-bus              bits 8,9    13       Load Register 106, per S-bus bits             0-10Q6  0-15     Status register, per Table III or              violation register, per Table IVQ7  N/A      No special significance.______________________________________

              TABLE III______________________________________STATUS REGISTERBit     Significance______________________________________15      MEM Off/On (0/1) at least interrupt14      System/User (0/1) selected at last interrupt13      MEM Off/On (0/1) currently12      System/User (0/1) selected currently11      PM enabled currently10      Portion mapped*9       Base Page Fence Bit 98       Base Page Fence Bit 87       Base Page Fence Bit 76       Base Page Fence Bit 65       Base Page Fence Bit 54       Base Page Fence Bit 43       Base Page Fence Bit 32       Base Page Fence Bit 21       Base Page Fence Bit 10       Base Page Fence Bit 0______________________________________*Bit  10    Mapped Addresses 0     Fence ≦M<20008 1     1<M<Fence

              TABLE IV______________________________________VIOLATION REGISTERBit     Significance______________________________________15      Read Violation14      Write Violation13      Base Page Violation12      Privileged Instruction Violation11      Reserved10      Reserved9       Reserved8       Reserved7       ME-Bus enabled6       MEM maps enabled5       System/User (0/1) enabled4       Map register bit 43       Map register bit 32       Map register bit 21       Map register bit 10       Map register bit 0______________________________________

APPENDIX A

Special terms and special notes plus abbreviations useful in interpreting the instruction descriptions. "MEM" refers to the system of the preferred embodiment.

Protected Mode (PM)

A program state created by the CPU.

Protected mode is entered by executing an STC 05B instruction.

Protected mode is exited by the CPU acknowledging an interrupt.

Preset will disable mapping functions and exit the protected mode.

Mem violation (MEMV)

Any of several conditions met which are not allowed by the preferred embodiment.

Can only occur when in PM.

Signified by an interrupt request to Channel 5 with the I/O Flag set (i.e., one may programmatically detect whether CPU or MEM caused the interrupt by execution of either a SFS 05 or SFC 05).

N.b. a STF 05 or CLF 05 still enables or disables the parity-related functions of MP.

Mem violation Register (MVR)

The MVR is not loaded with any data until Control-5 is set.

It is then continuously strobed until a MEMV occurs or Control-5 is cleared.

Therefore the contents are valid following a MEMV until STC 05 is issued. Preset will clear the MVR.

Base Page Fence (BPF)

Allows a portion of the base page (00000-01777) to reference the corresponding physical address regardless of which map is enabled.

Whether the portion is above or below the fence address is program assignable when the fence is set.

Logical addresses 00000 and 00001 (i.e. A/B) are not mapped when using the standard set of assembly language instructions.

Preset will clear the BPF.

Power Fail/Auto Restart

All registers on the MEM are volatile and should be saved in main memory by the power fail routines if auto-restart is desired.

Upon restoration of power, all maps are disabled and must be loaded before enabling under program control.

The MVR and BPF are automatically cleared when power is restored.

Alternate Map

The MEM has four distinct maps.

Two are for program execution.

The program map which has been specifically enabled is referred to as the current map.

The other program map is referred to as the "alternate map".

The Port A and Port B maps are never referred to as current or alternate. These are used by other devices to communicate with memory.

For example: SJP would make the System map be the current map and the User map would then be the alternate map.

              MEM MACRO INSTRUCTIONS______________________________________CODE*  MNEMONIC   LEN    PRV  INT  DESCRIPTION______________________________________02     MBI        1      A    Y   Move Bytes Into03     MBF        1      M    Y   Move Bytes From04     MBW        1      A    Y   Move Bytes Within05     MWI        1      A    Y   Move Words Into06     MWF        1      M    Y   Move Words From07     MWW        1      A    Y   Move Words Within10     SYA/B      1      M    N   Load/Store System11     USA/B      1      M    N   Load/Store User12     PAA/B      1      M    N   Load/Store Port A13     PBA/B      1      M    N   Load/Store Port B14     SSM        2      M    N   Store Status in                             Memory15     JRS        3      M    N   Jump and Restore                             Status20     XMM        1      M    Y   Transfer Map or                             Memory21     XMS        1      M    Y   Transfer Map                             Sequentially22     XMA/B      1      A    N   Transfer Maps                             Internally24     XLA/B      2      M    N   Cross Load25     XSA/B      2      A    N   Cross Store26     XCA/B      2      M    N   Cross Compare27     LFA/B      1      A    N   Load Fence30     RSA/B      1      N    N   Read Status31     RVA/B      1      N    N   Read Violation32     DJP        2      M    N   Disable and JMP33     DJS        2      M    N   Disable and JSB34     SJP        2      M    N   Enable System AND                             JMP35     SJS        2      M    N   Enable System and                             JSB36     UJP        2      M    N   Enable User and JMP37     UJS        2      M    N   Enable User and JSB______________________________________ *Last two octal digits. Preceded by 1017 or 1057 as required for A/B register significance. Instructions not explicitly referencing A/B will receive 1057 from the Assembler.

The column headings are defined as follows:

Len - number of words to this macro.

Prv - special consideration is given this instruction when executed under protected mode (i.e. Control-5 is set). If "A" (always) then this instruction itself is illegal under PM and will always cause an MEM Violation. If M (maybe) then the particular instruction can cause a MEMV as explained in the instruction description information. If N (never) then this instruction will never generate a MEMV in any mode.

Int - indicates whether the execution routine of the instruction is interruptible or not.

Disable MEM and JUMP

Mnemonic: DJPOctal code: 105732Format:   DJP     DEF     ADR(,I)

Causes translation and protection features of MEM to be disabled. The P-register is then set to the address resolved from ADR(,I) prior to disabling. As a result of executing this instruction, normal I/O interrupts are held off until the first opportunity following the fetch of the next instruction, unless three or more levels of indirect are used. If no maps are enabled, the instruction defaults to a JMP *+1,I.

This instruction will normally generate a MEMV when executed in PM. However, if the System map is enabled, it is allowed in PM.

Disable MEM and Jump to Subroutine

Mnemonic: DJSOctal Code: 105733Format:   DJS     DEF     ADR(,I)

Causes translation and protection features of MEME to be disabled. The P-register is then set one past the address resolved from ADR (,I) prior to disabling. The return address is written into ADR(,I). As a result of executing this instruction, normal I/O interrupts are held off until the first opportunity following the fetch of the next instruction, unless three or more levels of indirect are used.

This instruction will normally generate a MEMV when executed in PM. However, if the System map is enabled, it is allowed in PM.

Jump and Restore Status

Mnemonic: JRSOctal Code: 105715Format:   JRS     DEF     STS(,I)     DEF     ADR(,I)

Causes status of MEM (i.e., enable/disable and System/User) to be restored as indicated by Bits 15, 14 of STS(,I). Bits (13-0) are ignored. The P-register is set to ADR(I,). As a result of executing this instruction, normal I/O interrupts are held off until the first opportunity following the fetch of the next instruction, unless three or more levels of indirect are used.

This instruction will normally generate a MEMV when executed in PM. However, if the system map is enabled it is allowed in PM.

Load Fence from A/B-register

Mnemonic: LFA, LFB

Octal Code: 101727, 105727

Format: LFA

Loads the Base Page Fence Register from the A-register or B-register. Bits (9-0) specify the address in page zero where shared (unmapped) memory is separated from reserved (mapped) memory. Bit 10 is used to specify which portion is mapped.

______________________________________Bit 10         Mapped Addresses (M)   0           Fence ≦M<20008   1           1 <M<Fence______________________________________

This instruction will normally generate a MEMV when executed in PM. However, if the system map is enabled, it is allowed in PM. When a MEMV does occur, the fence is not altered.

Move Bytes From the Alternate Map

Mnemonic: MBFOctal Code: 105703Format:   LDA     SRC     LDB     DST     LDX     CNT     MBF

Moves a string of bytes using the alternate program map for source reads and the currently enabled map for destination writes. The A-register contains the source byte address. The B-register contains the destination byte address. The X-register contains the number of bytes to be moved. Both the source and destination must begin on word boundaries. The instruction is interruptible on an even number of byte transfers, thus maintaining the even word boundaries in A and B. At the time of interrupt, A, B, and X are reset to allow continuation of the instruction at next entry. When completed, the X-register will always be zero; A and B will be equal to the original address plus the number of bytes moved.

This instruction can cause a MEMV only if Read or Write protection rules are violated.

Move Bytes Into the Alternate Map

Mnemonic: MBIOctal Code: 105702Format:   LDA     SRC     LDB     DST     LDX     CNT     MBT

Moves a string of bytes using the currently enabled map for source reads and the alternate program map for destination writes. The A-register contains the source byte address. The B-register contains the destination byte address. The X-reg contains the number of bytes to be moved. Both the source and destination must begin on word boundaries. The instruction is interruptible on an even number of byte transfers, thus maintaining the even word boundaries in A and B. At the time of interrupt, A, B, and X are reset to allow continuation of the instruction at next entry. When completed, the X-register will always be zero; A and B will be equal to the original address plus the number of bytes moved.

This instruction will cause a MEMV when executed in PM and no bytes will be transferred, if so indicated by the write protection bit.

Move Bytes Within the Alternate Map

Mnemonic: MBWOctal Code: 105704Format:   LDA     SRC     LDB     DST     LDX     CNT     MBW

Moves a string of bytes with both the source and destination addresses established through the alternate program map. The A-register contains source byte address. The B-register contains the destination byte address. The X-register contains the number of bytes to be moved. Both the source and destination must begin on word boundaries. The instruction is interruptible on an even number of byte transfers, thus maintaining the even word boundaries in A and B. At the time of interrupt, A, B, and X are reset to allow continuation of the instruction at next entry. When completed, the X-register will always be zero; A and B will be equal to the original address plus the number of bytes moved.

This instruction will cause a MEMV when executed in PM and no bytes will be transferred, if so indicated by the write protection bit.

Move Words from the Alternate Map

Mnemonic: MWFOctal Code: 105706Format:   LDA     SRC     LDB     DST     LDX     CNT     MWF

Moves a series of words using the alternate program map for source reads and the currently enabled map for destination writes. The A-register contains the source address. The B-register contains the destination address. The X-register contains the number of words to be moved. The instruction is interruptible after each word move, at which time A, B and X are reset to allow continuation at next entry. When completed the X-register will always be zero; A and B will be equal to the original address plus the number of words moved.

This instruction can cause a MEMV only if Read or Write protection rules are violated.

Move Words Into the Alternate Map

Mnemonic: MWIOctal Code: 105705Format:   LDA     SRC     LDB     DST     LDX     CNT     MWT

Moves a series of words using the currently enabled map for source reads and the alternate program map for destination writes. The A-register contains the source address. The B-register contains the destination address. The X-register contains the number of words to be moved. The instruction is interruptible after each word move, at which time A, B and X are reset to allow continuation at next entry. When completed the X-register will always be zero; A and B will be equal to the original address plus the number of words moved.

This instruction will cause a MEMV when executed in PM and no words will be transferred, if so indicated by the write protection bit.

Move Words Within the Alternate Map

Mnemonic: MWWOctal Code: 105707Format:   LDA     SRC     LDB     DST     LDX     CNT     MWW

Moves a series of words with both the source and destination addresses established through the alternate program map. The A-register contains the source address. The B-register contains the destination address. The X-register contains the number of words to be moved. The instruction is interruptible after each word move, at which time A, B and X are reset to allow continuation at next entry. When completed, the X-register will always be zero; A and B will be equal to the original address plus the number of words moved.

This instruction will cause a MEMV when executed in PM and no words will be transferred, if so indicated by the write protect bit.

Load/Store Port A Map per A/B

Mnemonic: PAA, PABOctal Code: 101712, 105712Format:   LDA     TBL     PAA

Transfers 32 map registers to or from memory. If bit 15 of A/B is clear, the Port A map is loaded from memory starting from the address specified in bits 14-0 of A/B. If bit 15 of A/B is one, the Port A map is stored into memory starting at the address specified in A/B. The A/B register will be incremented by 32 to allow multiple map instructions. An attempt to load any map register in PM will cause a MEMV. An attempt to store the Port A map is allowed within the constraints of Write protected memory.

Load/Store Port B Map per A/B

Mnemonic: PBA, PBBOctal Code: 101713, 105713Format:   LDA     TBL     PBA

Transfer 32 map registers to or from memory. If bit 15 of A/B is clear, the Port B map is loaded from memory starting from the address specified in bits 14-0 of A/B. If bit 15 of A/B is set, the Port B map is stored into memory starting at the address specified in A/B. This A/B register will be incremented by 32 to allow multiple map instructions. An attempt to load any map register in PM will cause a MEMV. An attempt to store the Port B map is allowed within the constraints of Write protected memory.

Read Status Register into A/B-Register

Mnemonic: RSA, RSB

Octal Code: 101730, 105730

Format: RSA

Transfers the 16-bit contents of the MEM status register into A/B. This instruction may be executed at any time. The status register format is listed in Table III.

Read Violation Register into A/B-Register

Mnemonic: RVA, RVB

Octal Code: 101731, 105731

Format: RVA

Transfers the 16-bit contents of the MEM violation register into A/B. This instruction may be executed at any time. The violation register format is listed in Table IV.

Enable System Map and Jump

Mnemonic: SJPOctal Code: 105734Format:   SJD     DEF     ADR (,I)

Causes MEM hardware to use the set of 32 map registers, referred to as the System map, for translating all programmed memory references. The P-register is then set to the address resolved from ADR (,I) prior to enabling. As a result of executing this instruction, normal I/O interrupts are held off until the first opportunity following the fetch of the next instruction, unless three or more levels of indirect are used.

This instruction will normally generate a MEMV when executed in PM. However, if the System map is enabled it is allowed and effectively executes a JMP*+1,I.

Enable System Map and Jump to Subroutine

Mnemonic: SJSOctal Code: 105735Format:   SJS     DEF     ADR(,I)

Causes MEM hardware to use the set of 32 map registers, referred to as the System map, for translating all programmed memory references. The P-register is then set one past the address resolved from ADR(,I) prior to enabling. The return address is written into ADR(,I). As a result of executing this instruction, normal I/O interrupts are held off until the first opportunity following the fetch of the next instruction, unless three or more levels of indirect are used.

This instruction will normally generate a MEMV when executed in PM. However, if the System map is enabled it is allowed and effectively executes a JSB *+1,I.

Store Status Register in Memory

Mnemonic: SSMOctal Code: 105714Format:   SSM     DEF     STS(,I)

Stores the 16-bit contents of the MEM status register into the memory location pointed to by STS(,I). This instruction is used in conjunction with the JRS instruction to allow easy processing of interrupts which always enable the System map (if the MEM is enabled). The status register format is listed in Table III.

This instruction can cause a MEMV only if Write protection rules are violated.

Load/Store System Map per A/B

Mnemonic: SYA, SYBOctal Code: 101710, 105710Format:   LDA     TBL     SYA

Transfers 32 map registers to or from memory. If bit 15 of A/B is clear, the System map is loaded from memory starting from the address specified in bits 14-φ of A/B. If bit 15 of A/B is set the System map is stored into memory starting at the address specified in A/B. The MEM provides no protection (if not in PM) against altering the contents of maps while they are currently enabled. The A/B register will be incremented by 32 to allow multiple map instructions.

An attempt to load any map in PM will cause a MEMV. An attempt to store the System map is allowed within the constraints of Write protected memory.

Enable User Map and JumpMnemonic: UJPOctal Code: 105736Format: UJP DEF ADR(,I)

Causes MEM hardware to use the set of 32 map registers, referred to as the User map, for translating all programmed memory references. The P-register is then set to the address resolved from ADR (I) prior to enabling. As a result of executing this instruction, normal I/O interrupts are held off until the first opportunity following the fetch of the next instruction, unless three or more levels of indirect are used. If the User Map is already enabled, the instruction defaults to a JMP *+1,I.

This instruction will normally generate a MEMV when executed in PM. However, if the System map is enabled, it is allowed in PM.

Enable User Map and Jump to Subroutine

Mnemonic: UJSOctal Code: 105737Format:   UJS     DEF     ADR(,I)

Causes MEM hardware to use the set of 32 map registers, referred to as the User map, for translating all programmed memory references. The P-register is then set to one past the address resolved from ADR(,I) prior to enabling. The return address is written into ADR(,I). AS a result of executing this instruction, normal I/O interrupts are held off until the first opportunity following the fetch of the next instruction, unless three or more levels of indirect are used. If the User map is already enabled, the instruction defaults to a JSB *+1,I.

This instruction will normally generate a MEMV when executed in PM. However, if the System map is enabled, it is allowed in PM.

Load/Store User Map

Mnemonic: USA, USBOctal Code: 101711, 105711Format:   LDA     TBL     USA

Transfers 32 map registers to or from memory. If bit 15 of A/B is clear the User map is loaded from memory starting from the address specified in bits 14-0 of A/B. If bit 15 of A/B is set the User map is stored into memory starting at the address specified in A/B. The MEM provides no protection (if not in PM) against altering the contents of maps while they are currently enabled. The A/B register will be incremented by 32 to allow multiple map instructions.

An attempt to load any map in PM will cause a MEMV. An attempt to store the User map is allowed within the constraints of Write protected memory.

Cross Compare

Mnemonic: XCA,XCBOctal Code: 101726, 105726Format:   XCA     DEF     ADR(,I)

Compares the contents of the A/B register to a second operand in memory. If they are equal the next instruction in numerical sequence is executed. If not equal, the next instruction is skipped. The second operand is obtained by resolving ADR(,I), and using the alternate program map for the read operation. If neither the System map nor the User map map is enabled (i.e., MEM disabled state) then a compare directly with physical memory occurs.

This instruction will generate a MEMV only if Read protection rules are violated.

Cross Load

Mnemonic: XLA, XLBOctal Code: 101724, 105724Format:   XLA     DEF     ADR(,I)

Loads the A/B-register from memory. The operand is obtained by resolving ADR(,I) and using the alternate program map for the read operation. If neither the System map nor the User map is enabled (i.e., MEM disabled state) then a load directly from physical memory occurs.

This instruction will generate a MEMV only if Read protection rules are violated.

Transfer Maps Internally per A/B

Mnemonic: XMA, XMBOctal Code: 101722, 105722Format:   LDA     CTL     XMA

Transfer the entire contents (32 Map registers) of the System map or User map to the port A map or Port B map as determined by the control word in the A/B-register.

______________________________________Bit            Significance15             System/User (0/1) 0             Port A/Port B (0/1)14-1           Ignored______________________________________

This instruction will always generate a MEMV when executed in PM.

Transfer Map or Memory

Mnemonic: XMMOctal Code: 105720Format:   LDA     FST     LDB     TBL     LDX     CNT     XMM

Transfers (CNT) number of words either from sequential memory locations to sequential map registers or from maps to allow re-entry at a later time. The X-register will always be zero at the completion of the instruction; A and B will be advanced by the number of registers moved.

An attempt to load any map register in PM will generate a MEMV. An attempt to store map registers is allowed within the constraints of Write protected memory.

Transfer Maps Sequentially

Mnemonic: XMSOctal Code: 105721Format:   LDA     FST     LDB     TBL     LDX     CNT     XMS

Transfers (CNT) number of words either to sequential map registers or from maps to sequential memory locations. The A-register points to the first register to be accessed. The B-register points to the starting address of the table in memory. The X-register indicates the number of map registers to be affected. A positive quantity will cause the word found in location TBL to be used as a base quantity to be loaded into the first register. The next register will be loaded with the base quantity plus one, and so forth up to (CNT) number of registers. However, if X contains a negative quantity, the map registers are transferred to sequential memory locations on a one-to-one basis as in XMM, beginning with location TBL. Bits 0-9, 14, 15 are used as described in XMM.

An attempt to load any map register in PM will generate a MEMV. An attempt to store map registers is within the constraints of Write Protected memory. to memory. Bits 0-9 of memory correspond to 0-9 of the map and bits 14, 15 of memory relate to bits 10, 11 of the map. ##SPC1##

The A-register points to the first register to be accessed. The B-register points to the starting address of the table in memory.

______________________________________               127    Port B     96               95    Port A     64               63    User       32               31    System     0______________________________________

Maps are addressed as a contiguous space. Wrap-around from 127 to 0 can and will occur. It is the programmers responsibility to avoid this error. The X-register indicates the number of map registers to be transferred (1≦|CNT|≦128).

A positive number in X will cause the maps to be loaded with the corresponding data from memory. A negative (two complement) number in X will cause the maps to be stored into the corresponding memory locations.

The instruction is interruptible after each group of 16 registers has been transferred. A, B and X are the reset

Cross Store

Mnemonic: XSA, XSBOctal Code: 101725, 105725Format:   XSA     DEF     ADR(,I)

Store the contents of the A/B-register into memory. The destination is determined by resolving ADR(,I) and using the alternate program map for the write operation. If neither the System map nor the User map is enabled (i.e., MEM disabled state) then a store directly to physical memory occurs.

This instruction will always generate a MEMV when executed in PM.

APPENDIX B

MICROINSTRUCTION FORMATS AND EXECUTION

This section contains a description of the form of microinstruction word types, and a description of the form and effect of each microinstruction field within a word type.

A diagram of the form of each word type is in Table V. Table VI shows the binary and mnemonic form of each field of the microinstruction words.

Word Type 4

Word type 4 performs unconditional jumps or subroutine jumps to any location in the ROM address space. The only difference between JMP and JSB is that JSB causes the SAVE Register to be loaded at P2. At P2, the RAR is loaded with the 12-bit address specified in bits 16-5. However, the address will be modified before loading by one of eight mapping schemes determined by the Special Field.

Work type 4 is determined by the presence of JMP or JSB in the OP Field, and the presence of any Special Field code except CNDX. However, specifying JTAB, RTN, or any other Special Field code which ends with -110₂ or -011₂ will not result in a direct JMP or JSB, but will load the JTAB or RTN address. This is due to the nature of the address mapping scheme, discussed below.

The three low-order bits of the Special Field (RIR2-RIR0) constantly control the ROM address multiplexers which select one of eight possible sources to apply to the input of the RAR. This is designed to aid in the implementation of various types of indexed jumps and returns from sub-routines.

                                  TABLE V__________________________________________________________________________CPU MICROINSTRUCTION FORMATSWORDTYPE    OP CODE ALU      S-BUS    STORE  SPECIAL    23 22 21 20       19 18 17 16 15                14 13 12 11 10                         9 8 7 6 5                                4 3 2 1 0__________________________________________________________________________    ANY OP                    ANY    ANY NON-1   BUT JMP,       ANY ALU  ANY S-FIELD                         STORE  JUMP    JSB, IMM                  FIELD  SPECIAL                         ANY    ANY NON-2   IMM     PO UO ←OPERAND→                         STORE  JUMP       Cl Ll ←FIELD→                         FIELD  SPECIAL3   JMP     ANY      =0  ADDRESS     CNDX       CONDITION                RJS (9 BITS)                                ANY JUMP            ←ADDRESS→                                SPECIAL4   JMP JSB      (12 BITS)           BUT CNDX__________________________________________________________________________ NOTES: word type 2.  bit-19: if zero then ALU passes else ones complement.  bit-18: If zero then upper byte else lower byte. word type 3.  bit-14: if zero then reverse jump sense.

                                  TABLE VI__________________________________________________________________________MICROINSTRUCTION MNEMONICSBITS POS.→  20-23 0-4    15-19                    15-19 10-14                               5-9   18-19BINARY OP(4) SPECIAL(5)               ALU(5)                    COND(5)                          S-BUS(5)                               STORE(5)                                     IMM(2)__________________________________________________________________________00000  NOP   IOFF   INC  TBZ   TAB  TAB   HIGH00001  ARS   SRG2   OP1  ONES  CAB  CAB   LOW00010  CRS   L1     OP2  COUT  T    T     CMHI00011  LGS   L4     ZERO AL0   CIR  L     CMLO00100  MPY   R1     OP3  AL15  IOI  IOO00101  DIV   ION    OP4  NMLS  CNTR CNTR00110  LWF   SRG1   SUB  CNT8  DSPL DSPL00111  WRTE  RES2   OP5  FPSP  DSPI DSPI01000  ASG   STFL   OP6  FLAG  ADR  IR01001  READ  CLFL   ADD  E     M    M01010  ENV   FTCH   OP7  OVFL  B    B01011  ENVE  SOV    OP8  RUN   A    A01100  JSB   COV    OP9  NHOI  LDR  MEU01101  JMP   RPT    OP10 SKPF  RES2 CM01110  IMM   SRGE   OP11 ASGN  MEU  PNM01111  (BLANK)        NOP    DEC  IR2   NOP  NOP10000        MESP   CMPS NLDR  S1   S110001        MPCK   NOR  NSNG  S2   S210010        IOG    NSAL NINC  S3   S310011        ICNT   OP13 NDEC  S4   S410100        SHLT   NAND NRT   S5   S510101        INCI   GMPL NLT   S6   S610110        RES1   XOR  NSTR  S7   S710111        SRUN   SANL NRST  S8   S811000        UNCD   NSOL NSTB  S9   S911001        CNDX   XNOR NSFP  S10  S1011010        JIO    PASL INT   S11  S1111011        JTAB   AND  SRGL  S12  S1211100        J74    ONE  RUNE  X    X11101        J30    SONL NOP   Y    Y11110        RTN    IOR  CNT4  P    P11111        JEAU   PASS NMEU  S    S__________________________________________________________________________

              TABLE VII______________________________________CPU MICROINSTRUCTION                                   STATELABEL  Op     SPEC    ALU   STORE S-BUS COMMAND______________________________________         MESP          MEU   MEU   Q0         MESP          MEU   $     Q1         MESP          $     MEU   Q2         MESP          $     $     Q3         *             MEU   MEU   Q4         *             MEU   $     Q5         *             $     MEU   Q6         *             $     $     Q7______________________________________   -- Any legal code * -- Any legal code except MESP $ -- Any legal code except MEU?

Word Type 3

Word Type 3 is distinguished by JMP in the OP Field, and CNDX in the Special Field. RAR8-RAR0 are loaded during P2 with the address specified in RIR13-RIR5 (RAR11-RAR9 unchanged) if the condition specified in the Condition and RJS (Reverse Jump Sense) Fields is met (see section on Conditional Logic). If RJS (RIR14 is 0, then the JMP is taken if the condition is not met.

Condition field mnemonics are described below. Read carefully, as some conditions are met when the signal being tested is false. The condition causing JMP to occur (for RJS=1) is defined after the condition field mnemonic. The actual signal name which is input to the conditional logic on the CPU is included in parenthesis if it is different than the mnemonic of the microinstruction field.

CONDITION MNEMONICS

TBZ (TBZF)

The T-Bus was 000000₈ after the last word-type 1 or 2 μ-instruction.

ONES (ONESF)

The output of the ALU was 177777₈ after the last word-type 1 or 2 μ-instruction.

COUT (COUTF)

There was a carry out of the ALU after the last word-type 1 or 2 μ-instruction.

AL0 (ALU0F)

Alu bit 0 was 1 after the last word-type 1 or 2 μ-instruction.

AL15 (ALU15F)

Alu bit 15 was 1 after the last word-type 1 or 2 μ-instruction.

NMLS (MLSTF)

Memory power was not lost (condition not met after a power failure where memory power is lost). Memory should be good if met.

CNT8

The counter output is 11111111₂.

FPSP

Front Panel Special Test. No jump for standard front panel. Signal grounded on the standard front panel.

FLAG (FLAGFF)

The FLAG FF is set.

E (EXFF)

The EXTEND FF is set.

OVFL (OVERFF)

The OVERFLOW FF is set.

RUN (RUNFF)

The RUN FF is set.

NHOI (HOI)

There is no interrupt pending and the RUN FF is set.

SKPF

An I/O instruction skip condition (SFS, SFC was met. Met only during I/O control-type instructions.

ASGN (ASGN)

Microinstruction skip conditions is met if ASG instruction skip conditions are not met. Meaningful only when executing ASG instructions. ASGN is low if the conditions for an ASG skip are met. For ASG instructions the IR bits which specify skip tests are:

Ir0: rss -- reverse skip sense

Ir1: sz(a/b) -- skip if A/B Register is zero

Ir2: in(a/b) -- increment the A/B Register

Ir3: sl(a/b) -- skip if least significant bit (bit 0) of A/B is zero

Ir4: ss(a/b) -- skip if sign bit (bit 15) of A/B is zero

Ir5: sez -- skip if EXTEND bit is zero ASGN tests for all skip conditions at once. So there are 2 conditions which cause the SZ(A/B) test to succeed: either IN(A/B) is to be performed and the A/B Register is all ones, or IN(A/B) is not to be performed and the A/B Register is all zeros.

Then the equation of the skip (ASGN is low) is: SKIP(ASGN low) = ((TBZF.sup.. IR2+ONESF.sup.. IR2) IR0).sup.. IR1 + (EXFF IR0).sup.. IR5 + MULTIPLEXER SKIP

The ASG multiplexer output indicates a skip according to the following combinations of IR4, IR3, and IR0:

IR4      IR3        IR0        CONDITION(SS*)    (SL*)      (RSS)      CAUSING SKIP______________________________________0        0          0          NONE0        0          1          IR5.sup.. IR10        1          0          ALU0F0        1          1          ALU0F1        0          0          ALU15F1        0          1          ALU15F1        1          0          ALU0F+ALU15F1        1          1          ALU0F.sup. . ALU15F______________________________________

IR2

Ir bit 2 is high.

Note: the following 8 mnemonics test the state of front panel buttons. The conditions are met if the buttons are not depressed.

Nldr (ibl): ibl button

Nsng(instep): instr step button

Ninc(incm) : inc m button

Ndec(decm) : dec m button

Nrt(right) : right button

Nlt(left) : left button

Nstr(store) : store button

Nrst(display) : display button ("Restore")

NSTB (STROBE)

None of the front panel buttons are depressed.

NSFP (SFP)

Non-standard front panel installed. SFP is grounded by the standard front panel.

INT

An interrupt is awaiting service.

SRGL

Ir3 is "1" and the ALU bit 0 was 1 after the last word type 1 or 2 μ-instruction. Used as part of the SRG routine to test for skip condition.

RUNE (RUNEN)

The operator key switch is not in the LOCK position.

NOP (GROUND)

Jump if RJS = 0.

CNT4

The low-order 4 bits of the counter are 1111₂.

NMEU

Special condition reserved for use by memory management.

Word Type 2

This type of microinstruction is distinguished by IMM in the OP Field. The 8-bit literal in RIR17-RIR10 is gated onto the the S-bus and stored into the register specified in the Store Field at the end of the instruction cycle.

RIR18 specifies whether the literal is to be put onto the high (0) or low (1) eight bits of the S-Bus. The other half of the S-bus is all ones, as it is not being driven by any gates. RIR19 specifies whether the S-Bus is to be ones complemented (RIR19=1) through the ALU or passed (RIR19=0). If the store field specifies a register which is loaded off the S-Bus, then the data can not be complemented before storing into the register. (The ALU will still complement however.)

The Special Field is executed as in word type 1 microinstructions.

Word Type 1

This type of microinstruction is used to perform all arithmetic, logical, I/O, and memory operations, and to manage communication with special options like memory protect and memory management.

The signals specified in the OP, ALU and Special Fields are asserted at the output of the field decoders during the microinstruction cycle. They are clocked or gated to function properly with the system.

During P0-P2, RIR10-RIR13 of the S-Bus Field are selected to address the Scratch Pad RAMS. The output of the RAMs is latched into the Holding Register at the end of P2. This data is driven onto the S-Bus if RIR14 is high. RIR14=1 in the S-Bus Field determines that the S-Bus is to be driven by the Scratch Pad Registers (S1-S12, X, Y, P, and S). RIR14=O in the S-Bus Field selects one of the discrete registers to gate onto the S-Bus. If a timing or resource conflict arises, the FRZ FF (freeze) is set at the end of P2, disabling most CPU clocks, effectively freezing the microinstruction and preventing its completion. FRZ FF is reset at the end of P2 following the end of the conflict, and the instruction may go to completion.

During Direct Memory Access cycles, DMAFRZ is low during T3. This signal inhibits the S-Bus Field decoder and prevents the CPU from driving the S-Bus, so that DCPC may use it. Unless a word type 3 or 4 is being performed at this time, the processor freezes for one cycle.

During P3-P5, RIR8-RIR5 are selected to address the Scratch Pad RAMs, to select which register to store the T-Bus into. The Store Field is disabled by JORJ if JMP or JSB occurs in the OP Field, preventing unselected register alteration.

At P5, the selected register is loaded with the data at its inputs. RAMWEN is low during P5 to load the T-Bus into the Scratch Pads if RIR9 is 1 or if PNM is specified in the Store Field. It is disabled from going low during a CPU freeze, JMP or JSB in the OP Field, or when a Memory Protect Violation (MPV) occurs and the P or S Register is selected by RIR8-RIR5.

The Special Field is decoded unconditionally as long as the microinstruction is in the RIR. Decoder outputs are used for a variety of control functions, and are used at different time periods.

Micro-orders

This section contains a detailed description of the function or effect of each microinstruction mnemonic except for the Condition Field and the Jump Specials. First the mnemonic is given, then the CPU signal name of it in parenthesis by a description of its effect.

OP FIELD

NOP

No connection on CPU. No effect.

ARS (ARSOP)

32-bit arithmetic shift. The microinstruction must be in the form

ARS     PASS      B         B       L1 OR R1______________________________________OP      ALU       S-BUS     STORE   SPEC______________________________________

a. If L1: AS1=1, AS0=0, which shifts the A-Register left 1. The B-Register is passed through the ALU, and is shifted left 1 through the shifter (TBS1-0, TBS0=1), with ALX14=ALU15 and LSI=AR15. The T-Bus is then stored into the B-Register. Overflow is set if ALU14≠ALU15 (sign ≠ original B-Register bit 14). The effect is shown below.

b. If R1: AS1=0, AS0=1, which shifts the A-Register right 1, with B-Register 0 → A-Register 15. A-Register 0 is lost. The B-Register is passed through the ALU, and is shifted 1 right through the shifter (TBS0=0, TBS1=0), with ALX16=ALU15 (sign extended shift). The T-Bus is then stored in the B-Register. The effect is shown as follows: ##SPC2##

CRS (CRSOP)

32-bit circular shift. Requirements:

CRS     PASS      B         B       L1 OR R1______________________________________OP      ALU       S-BUS     STORE   SPEC______________________________________

a. If L1: AS1=1. AS0=0, which shifts the A-Register left one, with AR0ASHI=ALU15. The B-Register is passed through the ALU and shifted left in the shifter (TBS1=0, TBS0=1) with ALX14=ALU14 and LSI=AR15. The T-Bus is stored back into the B-Register.

b. If R1: AS1=0, AS0=1, which shifts the A-Register right one, with AR15→B-Register 0., B is passed through the ALU then shifted right (TBS0=0, TBS1=0), with ALX16=AR0. The T-Bus is stored back into the B-Register. ##SPC3##

LGS (LGSOP)

32-bit logical shift. Requirements:

LGS     PASS      B         B       L1 OR R1______________________________________OP      ALU       S-BUS     STORE   SPEC______________________________________

Operation is similar to CRS, ARS above.

a. L1: As B is shifted in the shifter, ALX14=ALU14, LSI=AR15. ASHI shifts 0 into the A-Register.

b. R1: As B is shifted in the shifter, ALX16=0, A-Register 5.sub.εB-Register 0. ##SPC4##

MPY (MPYOP)

Multiply step. Normally used in a repeat loop as part of a multiply algorithm. Requirements for proper operation:

MPY     ADD       B         B        R1______________________________________OP      ALU       S-BUS     STORE    SPEC______________________________________

The A-Register is shifted right internally, AR15→ALU0. The B-Register is gated onto the S-Bus. The ALU adds the S-Bus to the L-Register if A-Register bit 0 is a 1, and passes the S-Bus if it is 0. The output of the ALU is shifted right one, with ALX16=COUT (carry out of ALU). This is stored back into the B-Register via the T-Bus. ##SPC5##

Sixteen repeats will perform the function B+A.L and leave the result in the B and A Registers, least significant bit in A-Register 0 and most significant bit in B-Register 15.

DIV (DIVOP)

Divide step. Normally used in a repeat loop as part of a divide algorithm. Requirements for proper operations:

DIV     SUB       B         B        L1______________________________________OP      ALU       S-BUS     STORE    SPEC______________________________________

The A-Register is shifted left one internally with ASHI=COUT (carry from ALU). The ALU subtracts the L-Register from the B-Register. The result is shifted left one in the shifter, with ALX14=ALU14 and LSI=AR15. If COUT=1 (no borrow), then this result is stored into the B-Register via the T-Bus. If COUT=O (borrow), then the B-Register is shifted left internally (subtraction is effectively not performed), with B-Register bit 0→AR15. Sixteen repetitions of this instruction will perform (B,A)÷L=quotient in B, remainder in A assuming (B,A) represents a positive number. ##SPC6##

LWF (LWFOP)

Link with flag. If L1 or R1 is specified in The Special Field, the FLAG FF is linked with the ALU to form a 17-bit rotate through the R/S logic. For L1, LSI=FLAGFF and FLAGFF ALU15 For R1, ALX16=FLAG FF and FLAGFF ALU0.

WRTE (WRTEOP)

If memory is busy when WRTEOP occurs (REFRESH or MSRDY or DMALO low) the CPU will freeze until memory is free. Then the WRITE FF is set at the end of next P5. The WRITE FF is reset at the following P2. Initiates a write cycle in memory. The T-Register should be loaded in the same instruction in which WRTE is specified, as DCPC could destroy the T-Register contents if it was loaded any earlier.

ASG (ASGOP)

Used during ASG instruction routines. Sets, clears, complements the EXTEND FF according to the combination of IR6, IR7. Also clears the L-Register during P5 so the Overflow/Extend logic will operate correctly during increments.

READ (READOP)

If memory is busy when READOP occurs (REFRESH or MSRDY or DMALO low) the CPU will freeze until memory is free. Then the READ FF will be set at the end of the next P5. It will be reset at the following P2. This initiates a read from memory. The M-Register must be loaded prior to or during the instruction with READ in the OP Field. Data must be removed from the T-Register exactly two instruction cycles after READ, or else DCPC could destroy the contents of the T-Register, and memory disables T after that time.

ENV (ENVOP)

Enables the Overflow logic for the current ALU operation. OVER FF is set at the end of P5 if the L-Register and S-Bus have the same sign bit (bit 15) and ALU15 is different. Caution is advised so that the L-Register sign bit is set properly.

ENVE (ENVEOP)

Enables both the Overflow (see above) and Extend logic for the current ALU operation. The EXTEND FF is set at the end of P5 if COUT=1 (carry from the ALU).

JSB (JSBOP)

Specify jump or subroutine jump to new location

JMP (JMPOP)

in microcode. See sections on word types 3, 4. The AND of JSBOP and JMPOP (=JORJ) prevents storing into the scratch pads, disables the Store Field decoder, disables clocking of the Status Flag Register, and enables the RAR loading logic.

IMM (IMMOP)

Specifies word type 2. IMMOP enables the RIR onto the S-Bus, specifies PASS through the ALU or CMPS if RIR19=1, and prevents the Holding Register from driving the S-Bus.

SPECIAL FIELD

The Special Field is used to control special computer options like memory protect and memory management, to control interrupt recognition, to perform special arithmetic/logical operations, to initiate special CPU control features, and to specify jump schemes.

IOFF (ISOFFSP)

Clear the INTEN FF at the end of current instruction. This prevents recognizing of interrupts from devices with select codes greater than 5 (i.e., only memory protect and power fail or HALT mode may force RAR to 4 when a jump to 0 is attempted). Used during the JMP, I and JSB, I machine instruction routines to hold off interrupts until after one more instruction is executed.

SRG2 (SRG2SP)

Enables IR bits 0, 1, 2, 4 to the SRG shift/rotate decoder. This sets up a shift or rotate of the ALU onto the T-Bus as required by the SRG instruction.

L1 (LISP)

Sets up a left Shift (TBS0=1, TBS1=0) in the shift multiplexors. Without a qualifying OP Field command (LWF, ARS, CRS, LGS, MPY, DIV), this command shifts as shown below. ##SPC7##

L4 (L4SP)

Sets up a circular left shift of 4 bit positions (TBS0=0, TBS1=1) in the shift muliplexors. Rotates the ALU four positions before sending it onto the T-Bus.

R1 (R1SP)

Sets up a right shift (TBS0=0, TBS1=0) in the shift multiplexors. Without a qualifying OP Field command (LWF, ARS, CRS, LGS, MPY, DIV), this command shifts as shown below. ##SPC8##

ION (IONSP)

Turns on the INTEN FF at the end of current microinstruction. Allows normal interrupts to force the RAR to 4₈ when a jump or RTN to address 0₈ is attempted.

SRG1 (SRG1SP)

Enables IR bits 9,8,7,6 to the shift/rotate decoder. This sets up a shift or rotate of the ALU onto the T-Bus as required by SRG machine instructions.

RES2 (XCHSP)

Not offered to the microprogrammer. Exchange Flag FF with Extend FF.

STFL (STFLSP)

Set the CPU Flag FF.

CLFL (CLFLSP)

Clear the CPU Flag FF.

FTCH (FTCH)

Interface signal to Memory Protect. Results: latch the Violation Register from the M-Bus during P5; clear MPV at the end of P5; reset the indirect counter. To be used while the address of the current machine instruction is on the M-Bus prior to its execution. Used to initialize the memory protect error detection logic.

SOV (SOVSP)

Set the Overflow FF.

COV (COVSP)

Clear the Overflow FF.

RPT (RPTSP)

Set the Repeat FF. The RIR clock is disabled after P5 and the RAR increment is disabled after the following P2. The next microinstruction is repeated and the counter is incremented each succeeding P5 until the low 4 bits of the counter are 1111₂. Then the Repeat FF is cleared at P4 and normal control is restored. The microin struction after RPT will be repeated the two's complement of the value in the low 4 bits of the counter (with 0000₂ = 16 times).

SRGE (SRGESP)

Clear the Extend FF if IR5= 1.

NOP

No special signals generated.

MESP (MESP)

Special purpose signal for use by memory management.

MPCK (MPCK)

Memory Protect Check. Interface signal to Memory Protect. Must be specified while the address of an impending memory reference is on the S-Bus. This value is compared against the Fence Register to determine if a violation occurred. If the S-Bus < Fence Register then MPV will go low if memory protect is enabled. If MPV is low, then memory references will not affect memory and memory data will appear as 0, no I/O signals may be generated, and the P and S Registers may not be altered. FTCH or IAK will clear the MPV condition.

IOG (IOGSP)

The processor will freeze until T2. At the end of T2, the IO Group Enable FF will be set, enabling I/O signal generation for one I/O cycle.

ICNT (ICNTSP)

Increment the Counter

SHLT (SHLTSP)

The Run FF will be cleared at the end of the next microinstruction.

INCI (INCISP)

The indirect counter is incremented on the Memory Protect board. Used after an indirect address level is detected to keep infinite loops from preventing detection of interrupt requests. The INTEN FF is set after 3 increments.

RESI

No Signal.

SRUN (SRUNSP)

Set the Run FF at the end of the current microinstruction.

JTAB (JTABSP)

Load the RAR at P2 from the JTAB MAP

RTN (RTNSP)

Load the RAR at P2 from the Save Register and clear the Save Register at P5.

STORE AND S-BUS FIELDS

Below is a description of the mnemonics for the S-Bus and Store fields. The scratch registers are S1-S12, X, Y, P, S. Some of the other mnemonics have special meanings and applications, and are described below. The scratch pads are addressed directly by the RIR. The signal name out of the field decoders are indicated in parentheses.

The Store field decoder is disabled by JMP, JSB, or RIR19=1. The S-Bus field decoder is disabled by JMP, JSB, IMM, or DMAFRZ (T3 of a DCPC cycle).

TAB

Store (TABST) and S-Bus (TABEN). Selects the T, A, or B Register, depending on the setting of AAFF and BAFF which are set according to the value of the T-Bus whenever the M-Register is changed. Allows A, B to be accessed instead of locations 0, 1 of main memeory. See description of T below for timing associated with T.

______________________________________T-BUS      RESULTANT         REGISTERAT STORE   VALUES            SELECTEDINTO M     AAFF       BAFF       BY TAB______________________________________0          1          1          T1          0          1          A2          1          0          B>2         1          1          T______________________________________

CAB

Store (CABST) and S-Bus (CABEN). Selects the A-Register if IR-11 is 0, or the B-Register if IR-11 is 1.

Store (TST) and S-Bus (TREN). If from Store field, TST is sent to memory if the processor is not frozen. Used to clock the S-bus into the T-Register. If from S-Bus field, it freezes the processor until memory is ready.

CIR (CIREN)

Freeze the processor until T6, then during P3-P5, load the CIR from the Interrupt Address Bus, issue IAK and gate the CIR onto the S-Bus, high order 10 bits = 0.

(LST)

Load the L-Register from the S-bus.

IOI (IOIEN)

Drive the S-Bus from the source determined by the Select Code Bus, as shown below. Note: IOIEN will generate IOI on the I/O system (to cause the output buffer to be dumped onto the I/O Bus) only during T4 or T5 and if the IO Group Enable FF is set.

______________________________________SELECTCODE BUS      SOURCE SELECTED BY IOI______________________________________00         I/O Bus (=0)01         Front Panel Display02         DCPC Channel 1 Word Count Register03         DCPC Channel 2 Word Count Register04         Central Interrupt Register05         Memory Protect Violation Register06         Nothing = 177777807         Nothing = 1777778108 -778      I/O Bus (loaded from output       buffer of I/O device)______________________________________

IOO (IOOST)

This is independent of the IOO signal generated by the I/O signal generators. Gates the S-Bus onto the I/O Bus if the IO Group Enable FF is set.

CNTR

S-bus (CNTREN) and Store (CNTRST). CNTRST stores the low order 8 bits of the S-Bus into the counter. CNTREN enables the 8-bit counter onto the low 8 bits of the S-Bus. The high 8 bits of the S-Bus are all ones.

DSPL

S-bus (DSPLEN), Store (DSPLST). Selects the front panel Display Register.

DSPI

S-bus (DIEN): Selects the Display Indicator Register of the front panel onto bits 5-0 of the S-Bus. Higher order bits are ones. Store (DIST): Stores the low 6 bits of the S-Bus into the Display Indicator Register on the front panel. High order bits ignored. Note: Bits which are low correspond to LED indicators which are lit on the Display Indicator, as shown below:

S-BUS BITS LOW WITH DSPI GIVEN                5     4     3   2   1   0______________________________________REGISTER INDICATOR LIT                S     P     T   M   B   A______________________________________

ADR (ADREN)

Enable bits 9-0 of the M-Register onto bits 9-0 of the S-Bus. If IR10=0, gates 0's onto bits 15-10 of the S-Bus. If IR10=1, gate M-Register bits 15-10 onto the S-Bus. Performs zero/current page addressing for MRG-type machine instructions.

IR (IRST)

Load IR from the S-Bus.

The M-Register is only 15 bits. When enabled onto the S-Bus, bit 15 is low.

LDR (LDREN)

Enable the complement of the contents of the loader ROM selected by IR15, IR14, and addressed by counter bits 7-0 onto the S-Bus bits 3-0. Bits 15-4 will be pulled high.

MEU (MEST)

Memory expansion unit loaded from S-Bus.

RES2

No register drives S-Bus. S-Bus is 177777₈.

CM (CMST)

Store S-Bus into M-Register if and only if the IR contains an MRG-type instruction but not jump direct. In hardware, M is loaded if CMST is low and (IR12+IR14+IR13(IR15+IR11)=1.

MEU (MEEN)

Memory expansion unit is gated onto S-Bus.

PNM (PNMST)

Load the S-Bus into M-Register and the T-Bus into the P-Register (part of the scratch pads).

S1-S12

X, Y, P, S

These registers are all in the four 16×4 bit scratch pad registers and are addressed by the RIR through a multiplexer.

ALU FIELD

The ALU field feeds directly into the ALU without decoding. The ALU continuously performs the operation specified by RIR19-15, except that this may be overridden by a IMM or MPY microinstruction to do a PASS or CMPS instead. The operations possible are shown below (S = S-Bus, L L-Register). Arithmetic is 2's - complement, + = logical OR.

______________________________________INC         S PLUS 1OP1         (S+L) PLUS 1OP2         (S+L) PLUS 1ZERO        ALL ZERO OUTPUTOP3         S PLUS (S.sup.. L) PLUS 1OP4         (S+L) PLUS (S.sup.. L) PLUS 1SUB         S MINUS LOP5         S.sup.. LOP6         S PLUS (S.sup.. L)ADD         S PLUS LOP7         (S+L) PLUS (S.sup.. L)OP8         S.sup.. L MINUS 1OP9         S PLUS S (LOGICAL LEFT SHIFT)OP10        (S+L) PLUS SOP11        (S+L) PLUS SDEC         S MINUS 1______________________________________

Note: The following are logical operations.

______________________________________CMPS        NOT SNOR         NOT (S OR L)NSAL        (NOT S) AND LOP13        ALL ZERONAND        NOT (S AND L)CMPL        NOT LXOR         S (EXCLUSIVE -- OR) LSANL        S AND (NOT L)NSOL        (NOT S) OR LXNOR        NOT (S (EXCLUSIVE -- OR) L)PSAL        LAND         S AND LONE         ALL ONESSONL        S OR (NOT L)IOR         S OR LPASS        S______________________________________

M-REGISTER OPERATION

The M-Register must be loaded with the address of main memory The M-Register must be loaded with the address of main memory to be read before the READ micro-order occurs, or concurrent with it. The M-Register will be loaded at the start of P4. If the CM micro-order is used, M will be loaded only if an MRG-type instruction is in the IR. M may be altered after a reference is initiated, as the M-Bus is clocked into a holding register in memory after 200 μsec into the memory cycle. The M-Register must be loaded prior to using a WRTE command, because write data is being loaded into T when WRTE is given.

TAB Logic

Whenever M is loaded, the A-Addressable and B-Addressable flip-flops (AAFF, BAFF) are set according to the value of the T-Bus as shown below.

______________________________________T-BUS       AAFF          BAFF______________________________________1           1             02           0             1OTHER       0             0______________________________________

These flip-flops determine whether the A, B, or T-register will be used when the TAB micro-order is specified (because the A, B registers are addressed as locations 0, 1 of memory although they are actually hardware registers). If M could be receiving 0 or 1, the INC micro-order should be used in the ALU field. This T-Bus scheme with INC is used to simplify operand fetches during instruction execution. The standard sequence is shown below:

READ  INC     PNM    P    M←P, P←P+1, set TAB logic,                     initiate READ (wait) PASS    S1     TAB  Get data from T, A, or B-Register,                     according to TAB logic.

READ OPERATIONS

Freezes MSRDY (Memory Soon Ready) is high by P2 if memory is ready for another reference by the end of the next P5. If it is not high by P2, or DMALO is low, or REFRESH is low (memory refreshing), then a freeze will occur if READ is specified, until it is safe to proceed.

Initiation

A memory read cycle is initiated by the READ FF at the end of P5 when the READ micro-order is specified. READ is low from P5 through P1, but the falling edge initiates the cycle.

Data Retrieval

Data must be retrieved from memory exactly two microinstructions past the READ. After this time, memory disables the T-Register. If the T or TAB (if AAFF=BAFF=0) micro-orders are in the S-Bus field, and MSRDY is low, a freeze will occur until MSRDY is high. If the TAB logic specifies A or B, then the T-Register is not referenced at all, and no freeze will occur. Location 0, 1 of main memory may be referenced if the TAB logic is not used as prescribed.

WRITE OPERATIONS

Freezes

The WRTE micro-order requires the same freeze operation as READ.

Initiation

A write cycle is initiated by the WRITE FF at the end of P5 when the WRTE micro-order is given. WRITE is low from P5 through P1, but the falling edge of P5 initiates the cycle.

                                  APPENDIX C__________________________________________________________________________      MACRO JUMP POINTLINE    LABEL  AND MNEMONIC            BINARY CODE__________________________________________________________________________0001    NEUMACRO      JMP J30         JTABL  1000X0111100XXXX0002       JMP STFL        XMM    1000X011110100010003       JMP             XM*    1000X011110100100004           RTN                1000X011110100110005       JMP             XL*    1000X011110101000006       JMP             XS*    1000X011110101010007       JMP             XC*    1000X011110101100008       JMP             LF*    1000X011110101110009    RS*            PASS                  MEU MEU    1000X011110110000010    RV*        RTN PASS                  CAB MEU    1000X011110110010011       JMP             DJP    1000X011110110100012       JMP             DJS    1000X011110110110013       JMP             SJP    1000X011110111000014       JMP             SJS    1000X011110111010015       JMP             UJP    1000X011110111100016       JMP             UJS    1000X011110111110017    JTABL  JMP             XMM    1000X011110X00000018           RTN                1000X011110000010019       JMP             MBI    1000X011110000100020       JMP             MBF    1000X011110000110021       JMP MESP        MBW    1000X011110001000022       JMP             MWI    1000X011110001010023       JMP MESP        MWI    1000X011110001100024       JMP MESP        MWW    1000X011110001110025       JMP             SY*    1000X011110010000026       JMP             US*    1000X011110010010027       JMP             PA*    1000X011110010100028       JMP             PB*    1000X011110010110029       JMP             SSM    1000X011110011000030       JMP             JRS    1000X011110011010031           RTN                1000X011110011100032           RTN                1000X011110011110033       JSB             OPGET   ****UTILITY****LINE    LABEL  MICROINSTRUCTIONS       COMMENTS__________________________________________________________________________0034    XMM            PASS                  S3  P      S3 <= P; SAVE P0035               CMPS                  S2  X      S2 <= ONE'S COMP OF                             COUNT0036       JMP CNDX              ONES    RTN*   TEST FOR ZERO COUNT0037       IMM     LOW L   %200   L <= 11111111100000000038               SANL                  S1  A      MASK LOW 7 BITS OF A-REG0039       IMM     HIGH                  L   %337   L <= 11011111111111110040               SONL                  S1  S1     ADD CONTROL BIT (13)0041               PASS                  MEU S1     MEM ADDR REG <= S10042               PASS                  P   B      P <= B(TABLE ADDRESS)0043               PASS                  CNTR                      S2     CNTR <= S20044       JMP CNDX              FLAG    XMS    TEST FOR XMS INSTRUCTION0045       JMP CNDX              AL15                  RJS READMAP                             TEST FOR NEGATIVE COUNT0046       READ    INC PNK P      READ FIRST WORD P<=P+10047    MELOOP1        PASS                  S4  TAB    S4 <= MAP DATA0048           MESP              PASS                  MEU S4     MAP REG <= DATA0049       READ          ICNT              INC PNM P      READ NEXT WORD) INC                             CNTR AND P0050       JMP CNDX              CNT4                  RJS MELOOP1                             LOOP FOR 16X0051       JMP CNDX              CNT8    XMM,RTN                             IS TOTAL LOOP FINISHED0052       READ                   RESTART THE READ                             AT SAME ADDR0053       JMP CNDX              INT RJS MELOOP1                             TEST FOR NO INTERRUPT0054               DEC S3  S3     RESET P REGISTER FOR                             RESTART0055       JMP             SMM,RTN                             ELSE SERVICE INTERRUPT0056    XXM,RTN        PASS                  8   M      RESET B-REG0057    XMS,RTN        INC S1  CNTR   S1 <= REMAINING COUNT                             (2'S COMP)0058               PASS                  L   X      L <= ORIGINAL COUNT                             (POSITIVE)0059               CMPS                  X   CNTR   X <= REMAINING COUNT                             (POSITIVE)0060               ADD S1  S1     S1 <= ORIGINAL -                             REMAINING0061               PASS                  L   S1     L <= WORDS COMPLETED0062               ADD A   A      A <= A + TOTAL COMPLETED0063    P.RTN      RTN PASS                  P   S3     P <= NEXT INSTRUCTION0064    XMS    JMP CNDX              AL15                  RJS P.RTN  TEST FOR X 0 . . . NOP0065    MEL00P2    MESP              PASS                  MEU B      MAP REG <= DATA0066           ICNT              INC B   B      B <= B + 1; INC CNTR0067       JMP CNDX              CNT4                  RJS MELOOP2                             LOOP FOR 16X0068       JMP CNDX              CNT8    XMS,RTN                             IS TOTAL LOOP FINISHED0069       JMP CNDX              INT RJS MEL00P2                             TEST FOR NO INTERRUPT0070               DEC S3  S3     RESET P REGISTER FOR                             RESTART0071       JMP             XMS,RTN                             ELSE SERVICE INTERRUPT0072    READMAP        DEC S4  X      S4 <= X-10073               PASS                  CNTR                      S4     CNTR <= CNT+1 (TWO'S COMP)0074    MELOOP3    MPCK              INC PNM P      M.P. CHECK: P <= P+10075           MESP              PASS                  S1  MEU    S1 <= MAP REG0076       WRTE          ICNT              PASS                  TAB S1     WRITE DATA INTO TABLE0077       JMP CNDX              CNT4                  RJS MELOOP3                             LOOP FOR 16X0078       JMP CNDX              CNT8    XMM,RTN*0079       JMP CNDX              INT RJS MELOOP3                             TEST FOR NO INTERRUPT0080               DEC S3  S3     RESET P REGISTER FOR                             RESTART0081    *      JMP             XMM,RTN*                             ELSE SERVICE INTERRUPT0082    *0083    XMM,RTN*       PASS                  B   P      RESET B-REG0084               PASS                  L   X      L <= ORIGINAL COUNT                             (NEGATIVE)0085               INC X   CNTR   X <= REMAINING C0UNT                             (2's COMP)0086               SUB S1  X      S1 <= ORIGINAL - REMAINING0087               PASS                  L   S1     L <= WORDS COMPLETED0088               ADD A   A      A <= A + TOTAL COMPLETED0089           RTN PASS                  P   S3     P <= NEXT INSTRUCTION0090    XM*    IMM     CMHI                  SL  %337   S1 <= 00100000000000000091       LWF L1  PASS    CAB    T-BUS <= A/B; FLAG                             <= A/B(15)0092    PA.PB  JMP CNDX              ALO RJS SY.US  TEST FOR PORT A MAP0093       IMM     LOW L   %177   L <= 11111111011111110094               SONL                  S1  S1     S1 <= 00100000100000000095    SY.US  JMP CNDX              FLAG                  RJS XFER   TEST FOR SYSTEM MAP0096       IMM     LOW L   %337   L <= 11111111110111110097               SONL                  S1  S1     S1 <= 00100000X01000000098    XFER           PASS                  MEU S1     MEM ADDR REG <= S1(7-0)0099       IMM RPT LOW CNTR                      %000   CNTR <= 0; SET REPEAT                             FF FOR 16X0100           MESP              PASS                  MEU MEU    MEM PORT REG = MEM PROG                             REG0101       IMM RPT LOW CNTR                      %000   CNTR <= 0; SET REPEAT                             FF FOR 16X0102           MESP              PASS                  MEU MEU    MEM PORT GET <= MEM                             PROG REG0103    RTN*       RTN                RETURN0104       JSB             OPGET  GET OPERAND ADDR FROM                             INSTR + 10105           MESP              INC P   P0106       READ    INC M   M      SWITCH MAPS; GET REAL                             OPERAND0107               PASS                  MEU MEU    RESET MAP STATE0108           RTN PASS                  CAB TAB0109    XS*    JSB             OPGET  GET OPERAND ADDR FROM                             INSTR + 10110           MESP              INC P   P      SWITCH MAP STATE0111           MPCK              INC M   M0112       WRTE    PASS                  TAB CAB0113           RTN PASS                  MEU MEU    RESET MAP STATE0114    XC*    JSB             OPGET  GET OPERAND ADDR FROM                             INSTR + 10115           MESP              PASS                  L   CAB    L <= A/B; SET ALTERNATE                             MAP0116       READ    INC M   M      GET REAL OPERAND0117           MESP              INC P   P      P <= INSTR + 1;                             RESET MAP0118               XOR     TAB    COMPARE A/B WITH MEMORY0119       JMP CNDX              TBZ     RTN*   RTN-DON'T SKIP IF EQUAL0120           RTN INC P   P      P <= INSTR + 2; RETURN0121    LF*    IMM     HIGH                  L   %007   L <= 00000111111111110122               AND S1  CAB    S1 <= A/B(10-0)0123               PASS                  MEU MEU    SEND "FENCE" DIRECTIVE0124           RTN PASS                  MEU S1     MEM FENCE <= S1; RETURN0125    DJP    IMM     HIGH                  S2  %100   S2 <= 01000000111111110126       JMP             JP*0127    SJP    IMM     HIGH                  S2  %102   S2 <= 01000010111111110128       JMP             JP*0129    UJP    IMM     HIGH                  S2  %103   S2 <= 01000011111111110130    JP*    JSB IOFF        OPGET  GET OPERAND ADDR FROM                             INSTR + 10131    SETSTAT        PASS                  MEU S2     MEM STATUS IS SET HERE0132           MPCK              PASS    M      S-BUS <= ADDRESS; CHECK                             TARGET0133           RTN PASS                  P   M      P <= TARGET ADDRESS;                             RETURN0134    DJS    IMM     HIGH                  S2  %100   S2 <= 01000000111111110135       JMP             JS*0136    SJS    IMM     HIGH                  S2  %102   S2 <= 01000010111111110137       JMP             JS80138    UJS    IMM     HIGH                  S2  %103   S2 <= 01000011111111110139    JS*    JSB IOFF        OPGET  GET OPERAND ADDR FROM                             INSTR + 10140               INC S3  P      S3 <= RETURN ADDRESS0141               PASS                  MEU S2     MEM STATUS IS SET HERE0142           MPCK              PASS    M      S-BUS <= ADDRESS;                             CHECK TARGET0143       WRTE    PASS                  TAB S3     WRITE RETURN ADDR                             AT TARGET0144           RTN INC P   M      P <= TARGET + 10145    MBF    IMM MESP              HIGH                  L   %000   L <= 0000000011111111;                             SET ALT MAP0146    MBI        R1  PASS                  A   A      A <= SOURCE WORD ADDRESS0147           R1  PASS                  B   B      B <= DESTINATION WORD                             ADDRESS0148       LWF R1  PASS                  X   X      X <= WORD COUNT; FLAG                             <= ODD BYTE0149       JSB             X.LOOP-1                             MOVE BYTES IN PAIRS0150               PASS    X      T-BUS <= X0151       JMP CNDX              TBZ RJS B.RESET                             TEST FOR INTERRUPTED MOVE0152       JMP CNDX              FLAG                  RJS B.RESET                             TEST FOR NO ODD BYTE                      +10153           MESP              PASS    ADR    ALO <= IR(O); SET                             ALTERNATE MAP0154       JMP CNDX              ALO     *+2    TEST FOR MBF INSTRUCTION0155       IMM MESP              HIGH                  L   %000   L <= 0000000011111111;                             SET ALT MAP0156       READ    INC M   A      M <= SOURCE ADDRESS0157           L1  PASS                  A   A      FORM BYTE ADDRESS IN A0158               SANL                  S2  TAB    S2 <= AAAAAAAA000000000159       JMP MESP        MB*0160    MBW        R1  PASS                  A   A      A = SOURCE WORD ADDRESS0161           R1  PASS                  B   B      B <= DESTINATION WORD                             ADDRESS0162       LWF R1  PASS                  X   X      X <= WORD COUNT; FLAG                             <= ODD BYTE0163       JSB             W.LOOP-1                             MOVE BYTES IN PAIRS0164           MESP              PASS    X      T-BUS <= X; SELECT                             ALTERNATE MAP0165       JMP CNDX              TBZ RJS B.RESET                             TEST FOR INTERRUPTED MOVE0166       JMP CNDX              FLAG                  RJS B.RESET                             TEST FOR NO ODD BYTE                      +10167       IMM     HIGH                  L   %000   L <= 00000000111111110168       READ    PASS                  M   A      M <= SOURCE ADDRESS0169           L1  PASS                  A   A      FORM BYTE ADDRESS IN A0170               SANL                  S2  TAB    S2 <= AAAAAAAA000000000171    MB*    READ          MPCK              INC M   B      M <= DESTINATION ADDRESS0172           L1  PASS                  B   B      FORM BYTE ADDRESS IN B0173               AND S1  TAB    S1 = 00000000BBBBBBBB0174               PASS                  L   S1     L <= S10175               IOR S2  S2     S2 <= AAAAAAAABBBBBBBB0176       WRTE    PASS                  TAB S20177               PASS                  MEU MEU    RESET SELECTED MAP0178               INC A   A      A <= A + 10179           RTN INC B   B      B <= B + 10180    B.RESET      LWF L1  PASS                  X   X0181           L1  PASS                  A   A0182           L1  PASS                  B   B0183           RTN PASS                  MEU MEU    RESET SELECTED MAP; RETURN0184    MWI            PASS    X      T-BUS <= X0185       JMP CNDX              TBZ     MW*    TEST FOR X=00186    X.LOOP READ    INC M   A      READ SOURCE WORD0187           MESP              INC A   A      INCR. SOURCE ADDR;                             SWITCH MAPS0188               PASS                  S2  TAB    S2 <= DATA0189           MPCK              INC M   B      M.P. CHECK; M <= DEST                             ADDRESS0190       WRTE    PASS                  TAB S2     WRITE DATA INTO                             DESTINATION0191               INC B   B      INCREMENT DESTINATION                             ADDRESS0192           MESP              DEC X   X      DECREMENT COUNT; SWITCH                             MAPS0193       JMP CNDX              TBZ     MW*    TEST IF MOVE COMPLETE0194       JMP CNDX              INT RJS X.LOOP TEST FOR NO INTERRUPT 0195    DEC P P P                             <= INSTR ADDR0196           RTN PASS                  MEU MEU    RESET SELECTED MAP; -       RETURN0197    MWW            PASS    X      SET ALTERNATE MAP;                             T-BUS <= X0198       JMP CNDX              TBZ     MW*    TEST FOR X=00199    Y.LOOP READ    INC M   A      READ SOURCE WORD0200               INC A   A      INCREMENT SOURCE ADDRESS0201               PASS                  S2  TAB    S2 <= DATA0202           MPCK              INC M   B      M.P. CHECK; M <= DEST                             ADDRESS0203       WRTE    PASS                  TAB S2     WRITE DATA INTO                             DESTINATION0204               INC B   B      INCREMENT DESTINATION                             ADDRESS0205               DEC X   X      DECREMENT COUNT0206       JMP CNDX              TBZ     MW*    TEST IF MOVE COMPLETE0207       JMP CNDX              INT RJS W.LOOP TEST FOR NO INTERRUPT0208               DEC P   P      P <= INSTR ADDR0209    MW*        RTN PASS                  MEU MEU    RESET SELECTED MAP;                             RETURN0210    SY*    IMM     CMHI                  S1  %337   S1 <= 001000000000000                             -0211  JMP    MAPMOVE0212    PA*    IMM R1  CMHI                  S1  %176   S1 <=0100000100000000213           R1  PASS                  S1  S1     S1 <= 00100000010000000214       JMP             MAPMOVE0215    PB*    IMM     LOW S1  %237   S1 <= 11111111100111110216       JMP             US*+1  L <= 11011111111111110217                              S1 <= 00100000011000000218    US*    IMM     LOW S1  %337   S1 <= 11111111110111110219       IMM     HIGH                  L   %337   L <= 11011111111111110220               XOR S1  S1     S1 <= 00100000001000000221    MAPMOVE        PASS                  MEU S1     MEM ADDR REG <= S10222       IMM     LOW CNTR                      %337   CNTR <= 11011111 (-41B)0223               PASS                  S3  P      S3 <= P0224               PASS                  P   CAB    P <= A/B0225       JMP CNDX              AL15    MELOOP5                             AL15=1 => READ MAPS0226       READ    INC PNM P      READ FIRST WORD; P.                             <= P + 10227    MELOOP4        PASS                  S4  TAB    S4 <= MAP DATA0228           MESP              PASS                  MEU S4     MAP REG <= DATA0229       READ          ICNT              INC PNM P      READ NEXT WORD; P <= P + 10230       JMP CNDX              CNTB                  RJS MELOOP4                             LOOP FOR 32X0231               PASS                  CAB M      A/B <= A/B + 320232           RTN PASS                  P   S3     P <= INSTR +  10233    MELOOP5    MPCK              INC PNM P      M.P. CHECK; P <= P + 10234           MESP              PASS                  S1  MEU    S1 <= MAP REG0235       WRTE          ICNT              PASS                  TAB S1     WRITE DATA INTO TABLE0236       JMP CNDX              CNT8                  RJS MELOOP5                             LOOP FOR 32X0237           RTN PASS                  CAB P      A/B <= A/B + 320238           RTN PASS                  P   S3     P <= INSTR + 10239    SSM    JSB             OPGET  GET OPERAND ADDR FROM                             INSTR + 10240           MPCK              PASS    M      M.P. CHECK BEFORE WRITE0241               PASS                  MEU MEU    SEND "STATUS" DIRECTIVE0242       WRTE    PASS                  TAB MEU    WRITE STATUS WORD INTO                             MEMORY0243           RTN INC P   P      P <= INSTR + 2; RETURN0244    JRS    JSB IOFF        OPGET  GET OPERAND ADDR FROM                             INSTR + 10245       READ    INC M   M      READ THE STATUS WORD0246       IMM     HIGH                  S2  %103   S2 <=  01000011111111110247       LWF L1  PASS                  S1  TAB    FLAG <= STAT(15); S1                             (15) (= STAT(14)0248       READ    INC M   S3     READ JMP TARGET0249       JSB             OPGET+2                             GET TARGET ADDR FROM                             INSTR + 20250    ON.OFF JMP CNDX              FLAG    SY.USR TEST IF MEM WAS ON0251       IMM     HIGH                  S2  %101   IF OFF, S2 <=                             010000001111111110252    SY.USR LWF R1  PASS                  S1  S1     S1 <= STAT; AL15                             <= STAT(14)0253       JMP CNDX              AL15    SETSTAT                             TEST STAT(14) FOR USER                             SELECTED0254       IMM     HIGH                  L   %102   IF SYS, L <= 01000010                             111111110255               AND S2  S2     THEN S2 <= 010000X0                             111111110256       JMP             SETSTAT                             SET STATUS OF MEM: ALSO                             SET P0257    OPGET  READ    INC M   P0258               INC S3  P      S3 <= P + 1; S3                             <= INSTR + 20259               PASS                  M   TAB    M <= NEXT ADDR0260       JMP CNDX              AL15                  RJS RTH*   TEST FOR NO INDIRECT0261       JMP             INDLEVEL0262       LISTING OF INDLEVEL ROUTINE FOR REFERENCE ONLY0263    INDLEVEL      READ    INC M   M      READ NEXT LEVEL0264       JMP CNDX              NHOI                  RJS IND2   HALT OR INTERRUPT0265    INDIRECT   CNDX              PASS                  M   TAB    M(=T/A/B; INCR INDIRECT                             COUNTER0266       JMP CNDX              AL15    INDLEVEL                             CHECK FOR ANOTHER LEVEL                             OF INDIRECT0267       READ          RTN INC M   M      READ EFFECTIVE ADDRESS,                             RETURN0268    IND2       INCI              PASS                  M   TAB    M<=T/A/B/; INCR INDIRECT                             COUNTER0269       JMP CNDX              NSNG                  RJS INDIRECT                             JUMP BACK FOR SINGLE                      +1     INSTRUCTION0270               DEC P   P      RESET P0271       JMP             HORI   HALT OR INTERRUPT__________________________________________________________________________