JPS6330937A - Microprogram control device - Google Patents

Abstract

PURPOSE:To increase processing speed, by introducing a retry code register which holds a recursive code, and enabling the retry of a recursive micro- operation to be performed. CONSTITUTION:In case of selecting the micro-operation executed at just prior step, a register file, and an arithmetic circuit are controlled by using a micro code held by the retry code register 2 which holds a recursive code field used at the just prior step, and the micro code held by a non-recursive code register 4. To select the outputs of a recursive code register 3, and the retry code register 2, a code selector 5 is used. The control of the code selector 5 is performed by using a designation bit 21 which designates a recursive micro-operation designating method held by the non-recursive code register 4, and the code selector 5 selects the output of the recursive code register 3 when the designation bit 21 is reset, but when the designation bit 21 is set, it selects the output of the retry code register 2, and repeats the micro-operation executed at the just prior step.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a microprogram control device that controls a microprogram written in microinstructions and has a function of re-executing a microoperation executed in the immediately previous step.

[Conventional technology]

As research into knowledge information processing progresses, the functions required of computer systems also become more sophisticated. for example,
One of these is unification processing, which is a basic function of the programming language Prolog, which is attracting attention in the field of knowledge information processing.

The microprogram control method, which is widely used as a means for realizing computer systems, is suitable for realizing the above-mentioned advanced functions. Improving the functionality of a computer system using a microprogram control method means converting functions that were previously realized by software in a conventional general-purpose computer system into microprograms. In order to increase the amount of microprograms and complicate the structure, the integration of advanced functions into microprograms has tended to be subdivided into a large number of functional routines, with a structure in which the main routine calls the functional routines.

In order to call a functional routine, it is necessary to exchange data between the main routine and the functional routine. This is where data is stored for main routines and functional routines (
This is because the registers or specific addresses in memory are different.

One of the conventional techniques for realizing data reception and passing between a main routine and a functional routine is a method in which a register to be used for data reception and passing is specified.

In this method, the main routine must set data in a specified register immediately before calling the functional routine. This method is called the first conventional technique.

Another conventional technique is a method in which the main routine passes the designated number of the register holding data to the functional routine. In this method, a register is specified for receiving and passing a designated register number, and when a functional routine accesses data, register access or memory access is performed by register indirect designation. FIG. 3 shows the case of register indirect specification register access. 1 is a control code memory that stores a microprogram, and 10 is a microcode register (MCR>) that holds a microinstruction being processed.
It is. The main routine uses the microcode register 10 to directly specify registers in the register file 13 and access data. When calling a function routine, a register designation number is set in the register 11 for indirect designation. In the functional routine, the indirect designation register 11 is used to designate a register in the register file 13 and access data. In this method, the main routine must set a register designation number in the specified register immediately before calling the function routine. This method is called the second conventional technique.

[The pressure point that the invention seeks to solve]

In the conventional technology, which used data transfer between the main routine and the function routine to realize a function routine call, the main routine always inputs data or a register specification number to a specific register group immediately before calling the function routine. I needed to set it up.

The operations set in specific registers by the main routine must be performed on all data received and passed between the main routine and the functional routines.

When using the prior art, the main routine's set operations on specific registers often increase the number of execution steps of the microprogram. Among these are
Pass to function routine like unification processing 2
In some cases, one function routine is selected from among a plurality of function routines using the results of comparing the data, and the process branches to that function routine. In the case of processing that uses data passed to a function routine in the step of calling the function routine, the main routine's setting of specific registers, which was done in the prior art, is done by the microprogram execution step.
This has the problem of increasing the number of knobs and reducing the processing speed of the entire computer system.

An object of the present invention is to control the execution of a microprogram written using a microinstruction that has a specific bit that specifies re-execution of a microoperation specified by a specific field of a microinstruction executed in the previous step. Data exchange between function routines solves the problem of an increase in the number of microprogram execution steps in processing that uses data passed to a function routine in a step that calls an IR function routine, as seen in unification processing. The object of the present invention is to provide a microprogram control device.

[Means for solving problems]

The microprogram control device of the present invention includes a control code memory that stores a microprogram, a re-execution code register that retains a specific field of a micro-instruction executed in the immediately previous step, and a re-execution code register that is to be retained. a repeatable code register that holds a specific field; a non-repeatable code register that holds a field other than the specific field held by the repeatable code register; and a repeatable code register that holds a field other than the specific field held by the repeatable code register; a code selector that selects either one of the output of the repeatable code register and the re-execution code register and supplies the selected code to the repeatable code register and the outside; and a specific bit held by the non-repeatable code register. By setting , the micro-operation specified by the specific field executed in the previous step can be repeated.

[Effect]

The microprogram control device of the present invention includes the introduction of a re-execution code register that holds a specific field of the micro-instruction executed in the immediately previous step, a repeatable code register that holds the output of the control code memory in the micro-instruction, and a re-execution code register that holds the output of the control code memory in the micro-instruction. By adding a specification bit that specifies whether to specify the micro-operation for the specific field with either output of the code register, setting the bit allows re-execution of the micro-operation for the specific field that was executed in the previous step. enable.

By re-executing some of the micro-operations executed in the step where the main routine called the function routine,
Functional routines can directly manipulate data that is passed to and from the main routine. For example, in the case of unification processing, the processing contents differ depending on the combination of data types of two pieces of data exchanged between the main routine and the functional routine. In this example, in the step of calling a function routine, the main routine checks the combination of data types of the two data types, determines one function routine from a plurality of function routines based on the combination of data types, and Jump to function routine. Therefore, in the first step of the function routine, by re-executing the micro-operation specifying the access to the two data that was performed in the previous step (the step where the main routine called the function routine), the access to the two data can be performed. For example, if a register holds the above two pieces of data, the functional routine can perform the above two pieces of data by re-executing the micro-operation that specifies the register.
data can be accessed.

〔Example〕

Embodiments of the present invention will be described with reference to the drawings.

FIG. 2 is a format diagram of microinstructions processed by the embodiment of the microprogram control device of the present invention which will now be described. 21 is for selecting either re-execution of the micro-operation specified by the specific field of the micro-instruction executed in the previous step or execution of the micro-operation specified by the specific field of the micro-instruction executed in the relevant step. This is a designated bit. Reference numeral 22 is a repeatable code field (RMC) that specifies a micro-operation that allows the re-execution of the micro-operation executed in the previous step using the specification bit 21. 20 is the specified bit 2
This is a non-repeatable code field (NMC) that specifies a micro-operation that cannot be re-executed in the previous step.

FIG. 1 is a block diagram showing an embodiment of the present invention.

In the figure, 1 is a control code memory that stores a microprogram. 3 is a repeatable code register that holds the repeatable code field of the microinstruction read from the control code memory 1;

Reference numeral 4 denotes a non-repeatable code register that holds a non-repeatable code field of a microinstruction read from the control code memory 1 and a bit specifying execution of a repeatable micro-operation.

In conventional microprogram controllers, it is not possible to re-execute the micro-operation executed in the previous step, so there is no bit in the micro-instruction that specifies the re-execution of a repeatable micro-operation. There is no distinction between regular code fields and non-repeatable code fields. That is, conventional microprogram controllers provide the same control over all fields of a microinstruction as the non-repeatable code field.

When selecting the micro-operation executed in the previous step as a repeatable micro-operation, the micro-code held by the re-execution code register 2 which holds the repeatable code field used in the previous step and the non-repeatable code register 4 are selected. The stored microcode is used to control register files and arithmetic circuits.

A code selector 5 is used to select the outputs of the repeatable code register 3 and the rerun code register 2. The code selector 5 is controlled by a designation bit 2 held in the non-repeatable code register 4 that designates a repeatable micro-operation designation method.
1, the code selector 5 selects the output of the repeatable code register 3 when the specified bit 21 is reset, but selects the output of the rerun code register 2 when the specified bit 21 is set. This will repeat the micro-operation that was executed in the previous step.

〔Effect of the invention〕

The microprogram control device of the present invention introduces a re-execution code register that holds repeatable code that controlled the repeatable micro-operation executed in the previous step,
By making it possible to re-execute the repeatable micro-operations executed in the previous step, data passing between the main routine and the functional routine as seen in the unification processing example can be done by re-executing the register-specified micro-operations. This has the effect of preventing an increase in the number of microprogram execution steps due to function routine calls and improving processing speed.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a format diagram of microinstructions controlled by the microprogram control device of the present invention, and FIG. 3 is a block diagram of a conventional microprogram control device. . 1... Control code memory, 2... Re-execution code register (LMCR), 3... Repeatable code register (
RMCR>, 4...Non-repeatable code register (NMC
R), 5...Code selector, 10...Microcode register (MCR>, 11...Register for indirect specification, 12...Register specification number selector, 13...Register file, 20... ...Non-repeatable code field, 21...Specified bit, 22...Repeatable code field. $ / Figure $ 2 side

Claims (1)

[Claims] A control code memory that stores a microprogram, a re-execution code register that retains specific fields of the micro-instruction executed in the immediately previous step, and a repeatable code register that retains specific fields that the re-execution code register is intended to retain. and a non-repeatable code register that holds fields other than the specific field held by the repeatable code register, and the repeatable code register and the re-execution code register depending on the value of the specific bit held by the non-repeatable code register. a code selector that selects one of the outputs of the repeatable code register and supplies the selected code to the repeatable code register and the outside;
A microprogram control device characterized in that by setting a specific bit held in the non-repeatable code register, it is possible to repeat the micro-operation specified by the specific field executed in the previous step.