JPH05233427A - Information processor - Google Patents

Abstract

PURPOSE:To speed up interchanging process for data including a program between storage device which differ in access speed. CONSTITUTION:The information processor has a main memory MM, a subprocessor memory SM which gains high-speed access, and a low-speed access magnetic disk storage device HD; when a requested memory area RE needs to be secured in the memory MM, a main program 2 is rolled out to try to secure the area, but a roll-out to a free area UE of the memory SM by a roll- out LO1 is indicated at this time and if the roll-out LO1 can not be performed, the storage device HD is indicated by a roll-out LO3 next to totally shorten the time required to a roll-out process.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing device,
In particular, the present invention relates to an information processing apparatus that incorporates a plurality of storage devices having different access speeds and exchanges data (including programs) between the storage devices.

[0002]

2. Description of the Related Art Conventionally, a large-scale computer system or workstation has a high-speed, small-capacity primary storage device (main storage device) as a storage area in which a program is executed, and the area is assumed to have a virtually large capacity. The virtual memory system is adopted as described above. The virtual storage system uses a low-speed large-capacity secondary storage device (auxiliary storage device) such as a magnetic disk storage device. In the virtual memory system, generally, programs and data with high priority are stored in the primary storage device according to priorities such as high access frequency, and programs and data that cannot be stored are stored in the secondary storage device. When it is necessary to execute the program stored in the secondary storage device during the execution of the program, the program is called to the primary storage device and executed on the primary storage device. Calling a program from the secondary storage device to the primary storage device and storing it in the primary storage device is called loading. The division unit of the loaded data or program is a segment or page unit.

As described above, when the program is executed, the program is loaded from the secondary storage device into the primary storage device and executed as necessary, so that the same area of the primary storage device is multiplexed by a plurality of programs. Will be used. In other words, the primary storage device has an illusion as if it has a storage area capable of executing a plurality of programs, so that the capacity of the primary storage device is virtually increased. is there.

When a program is loaded during execution of the program using the virtual memory system,
A program called from the secondary storage device may be overwritten and stored so as to cover an unnecessary program among programs already on the primary storage device. This is called overlay. According to this overlay, the program existing in the original primary storage device disappears, so the roll-in / roll-out method is adopted to avoid this.

FIG. 11 is a diagram for explaining a conventional roll-in / roll-out system. The roll-in / roll-out method is performed between the primary storage device and the secondary storage device in the information processing apparatus as shown in FIG. The main programs 1, 2, and 3 are stored in the primary storage device of FIG. 11, and the main program 0 is stored in the secondary storage device. 1 from main program 0 from secondary storage
When loading to the secondary storage device, the program already on the primary storage device is saved to the secondary storage device, and then the main program 0 is placed in the primary storage device. Therefore, the program on the original primary storage device can be restarted later without disappearing. Saving the program that was on the original primary storage device to the secondary storage device is called roll-out, and reloading the rolled-out program onto the primary storage device is called roll-in.

The roll-in / roll-out method is used also when a memory area of a temporary primary storage device is emptied because a program interrupted during execution of a program has a high priority. Roll out the program being executed on the primary storage device before loading the program having a high frequency, load it afterwards, and roll in the program being executed that has been rolled out after the execution of the program having a high priority is completed. And run again.

As described above, the roll-in / roll-out method in the virtual memory system has been performed between the primary storage device and the secondary storage device.

[0008]

The secondary storage device of FIG. 11 is generally a magnetic disk storage device (hereinafter referred to as HD).
Alternatively, the data read speed and the write speed are very slow as compared with a DRAM (Dynamic Random Access Memory) or an SRAM (Static RAM), which is composed of a flexible disk storage device (hereinafter referred to as FD) and constitutes a primary storage device. .. This will be described in detail below.

FIG. 12 is a table showing a comparison of access speeds between standard storage devices.

FIG. 12 shows data per access.
Access for each memory when the amount is assumed to be 4 KB
A comparison of speeds is shown. Configure the secondary storage device in the figure
The access speed of the HD is the SR that constitutes the primary storage device.
Normal by setting the access speed of AM or DRAM to 1
When turned into 10²-10³You can see that it is twice as slow. Also,
Regarding the access speed of the FD that constitutes the secondary storage device,
Similarly, 10³-10 ^FourDouble slower primary storage and secondary storage
There is a large difference in access speed between the storage device
Is clear.

For the reasons described above, the large difference in access speed between the primary storage device and the secondary storage device causes a problem that the program exchange speed at the time of roll-in / roll-out of the program is slowed down. Furthermore, this problem is 1
This causes a problem that the processing speed in the information processing apparatus adopting the virtual storage system is slowed by providing the secondary storage device and the secondary storage device in association with the primary storage device. The slower program exchange speed becomes more remarkable as the capacity of the roll-in / roll-out target program is larger.

Therefore, an object of the present invention is to provide an information processing apparatus capable of speeding up the process of exchanging data including a program between storage devices having different access speeds in the information processing apparatus.

[0013]

An information processing apparatus according to the present invention comprises first storage means having a high access speed,
An information processing apparatus including a second storage unit having a low access speed and a large capacity, further comprising a third storage unit having an access speed higher than that of the second storage unit, and a first storage unit. When the data stored in the storage device is rolled out, the data is selectively rolled out to one of the second and third storage means.

The information processing apparatus according to the present invention is the information processing apparatus including the above-mentioned first storage means, second storage means and third storage means, and a first determination means,
It may be configured to further include a selection unit, an area securing unit, a second determination unit, a first transfer unit, a second transfer unit, and a setting unit.

The above-mentioned first determination means is configured to determine whether or not the first storage area can be secured in the first storage means in response to a request for securing the first capacity area in the first storage means. To be done.

The above-mentioned selecting means, in accordance with the inability judgment by the first judging means, of the data group previously stored in the first storing means, the capacity of each area that matches a predetermined criterion and stores them. Is configured to select at least one or more pieces of data such that the sum of the two becomes a second capacity equal to or larger than the first capacity.

The area securing means secures an area having the above-mentioned second capacity in the third storage means, and the second determining means determines whether the area securing means has secured the area. ..

The above-mentioned first transfer means transfers the data selected by the selection means to the second storage means in accordance with the determination that the second determination means cannot secure the data, and the second transfer means operates the second storage means.
A third area in which the data secured by the selecting means is secured by the area securing means in accordance with the determination that the securing is possible by the determining means.
It is arranged to transfer to the area of the storage means having the aforementioned second capacity.

The setting means is configured to set the area constituting the second capacity in the first storage means so that the area can be secured after the data transfer by the first or second transfer means.

Further, the predetermined criterion in the above-mentioned selecting means may be a criterion including that the frequency is stored in a continuous area of the first storage means and referred to infrequently.

Further, the above-mentioned data may include a program.

[0022]

In the information processing apparatus having the first storage means and the second storage means described above, the third storage means is further provided to serve as the roll-out destination of the data stored in the first storage means. , Either the second or the third storage means can be selectively designated, so that when the third storage means is selected as the rollout destination, the rollout processing speed is increased.

The first storage means, the second storage means, and the third storage means
In the information processing device including the storage means, the data stored in the first storage means selected by the selection means in accordance with the determination result of the second determination means is either the first transfer means or the second transfer means. Since it is selectively transferred to either one of the second and third storage means via the second transfer means, when the transfer processing is performed by the second transfer means, it is more than when the data transfer is performed by the first transfer means. Also, the transfer processing is speeded up, and it is possible to speed up the transfer speed of the data selected by the selection means performed in response to the request to secure the area of the first capacity in the first storage means.

[0024]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a diagram for explaining a control method when a memory area is used in an information processing apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing the area usage state and the program roll-in / roll-out status when the memory area shown in FIG. 1 is used in association with each other.

The information processing apparatus according to the present embodiment employs a method of advancing processing while using an unused portion of the memory in the apparatus as an area for rolling out programs and data.

First, an outline of a control method when the memory area is used in the information processing apparatus according to the present embodiment will be described with reference to FIG.

It is assumed that the information processing apparatus adopting the control method shown in FIG. 1 adopts a multiprocessor configuration.

In FIG. 1, the memory area of the information processing apparatus includes a main memory MM included in a main processor of the multiprocessors, a sub processor memory SM included in a sub processor, and an HD (magnetic disk storage device). The main memory MM is a memory area M1
~ M4. Sub-processor memory SM has a relatively large storage capacity of 6 MB, for example, and is configured to include memory areas M5 to M7. The magnetic disk storage device HD has a much larger storage capacity than the main memory MM and the sub processor memory SM, and stores the main programs n, m1 to 4 and the sub processor programs. Comparing the access speeds, the order is the main memory MM, the sub-processor memories SM, H
The speed is high in the order of D. As shown in FIG. 12, HD is SR
It is much slower than the memories MM and SM consisting of AM or DRAM.

FIG. 1 shows the storage state of the program during a certain period when the program is executed in the information processing apparatus. During this period, in the main memory MM, the main program 3 is stored in the area M1, the main program 2 is stored in the area M3, and the main program 1 is stored in the area M4. At this time, the program and data are not stored in the area M2, and the area M2 is in an empty state. In the sub processor memory SM, sub processor programs are stored in the area M7. At this time, the other memory areas including the memory areas M5 and M6 except the memory area M7 of the memory SM are so-called unused areas such that the programs stored therein are unlikely to be executed. Hereinafter, such an area is referred to as an idle area UE.
Further, in the main memory MM, it is required to secure a new area for executing the program during execution of the program. The area on the memory thus requested is called a requested memory area RE.

Next, the method of using the memory area in FIG. 1 will be described with reference to FIGS. In FIG. 2, states 1 to 5 are shown as control states when the memory area is used. Further, the usage status in the main memory MM and the usage status in the sub processor memory SM are shown corresponding to the statuses 1 to 5. Further, the region R is associated with each of the usage states of the main memory MM and the sub-processor memory SM.
The rollout status for programs that need E and programs that are no longer needed is shown.

In the state 1 of FIG. 2, the main memory MM
There is enough free space. At this time, if any of the main programs 1 to 3 in the memory MM requires the requested memory area RE, the area RE can be secured in the empty area of the memory MM, so that the rollout operation of the unnecessary program is performed. Absent.

In the state 2 of FIG. 2, the main memory MM
Is in a state where there is no free area (FULL), but when the sub processor memory SM has a sufficient free area, if either the main program 1 or 3 in the memory MM needs the area RE, Rollout LO1 is performed, and unnecessary main program 2 is transferred to an empty area of subprocessor memory SM, and memory M
The required memory area RE is secured on M.

In state 3 of FIG. 2, main memory MM
Is FULL, and the sub-processor memory SM is also FULL
If it is L, if either the main program 1 or 3 of the memory MM needs to secure the required memory area RE, the rollout LO3 transfers, for example, the unnecessary main program 2 to the HD, and A region RE is secured in the area.

In state 4 of FIG. 2, main memory MM
Is FULL, and the sub-processor memory SM is also FULL
If L, the main program 1 of the main memory MM
If 3 or 3 needs to secure the area RE, the program in the idle area UE of the sub-processor memory SM is first transferred to the HD by the rollout LO2. Then, the rollout LO1 causes the unnecessary main program 2 on the main memory MM to be transferred to the sub-processor memory S.
It is transferred to the area UE on M. As a result, the required memory area RE is secured in the main memory MM.

In state 5 of FIG. 2, main memory MM
Is FULL, and the sub-processor memory SM is also FULL
If it is L, if the subprocessor program of the subprocessor memory SM needs to secure the required memory area RE, the program stored in the idle area RE in the subprocessor memory SM by the rollout LO2 is HD.
Transferred to. As a result, the sub processor memory SM
The required memory area RE required by the sub processor program is secured.

In the above-mentioned method, as a memory area for rolling out unnecessary programs and data, first, the existence of the idle area UE of the sub-processor memory SM other than the main memory MM is checked. If there is an idle area UE in the memory SM, the area UE is used as an area for rollout and the HD is not used. On the contrary, when the idle area UE does not exist in the sub program memory SM, the HD is used as a storage area for rollout as in the conventional case.

In FIG. 1, the sub processor memory SM is oriented first as a storage area for rolling out a program on the main memory MM in order to secure the required memory area RE. The idle area UE in the shared memory of the processor or other high-speed access file (IC card or RAM disk device) may be directed. Although the program is rolled out, the data can be rolled out in the same manner.

Conventionally, the secondary storage device, HD, was uniformly used as the roll-out destination of programs and data, but in the present embodiment, the sub-processor memory SM that can be accessed at high speed is directed first, and then the low-speed access is performed. Since the HD is aimed at, the transfer time of the data and the program in the rollout becomes shorter as a whole than before. Further, the data transfer time including the operations of the roll-ins LI1 to LI3 performed corresponding to each of the rollouts LO1 to LO3 described above can be dramatically shortened.

FIG. 3 is a block diagram of a document processing dedicated workstation according to one embodiment of the present invention.

FIGS. 4 (a) to 4 (c) are diagrams for explaining the memory area utilization state in the workstation according to the control method of FIG. 1 when the workstation dedicated to document processing of FIG. 3 is operating.

The workstation shown in FIG. 3 has a main module X, which improves responsiveness to humans and operability.
Graphic module Y and input control module Z
Hierarchical multiprocessor configuration including is adopted.

A widely used single processor,
With the structure of a single bus, processing from a high response request such as mouse operation to heavy editor editing processing is created as software on the same operating system, so mutual processing content and performance Therefore, it is difficult to predict the behavior and responsiveness. Therefore, a system capable of handling multimedia including a printer, a keyboard, a mouse, a scanner, and the like, and having a fast response is desired. Therefore, in the workstation shown in FIG. 3, the hardware modules X to Z each including a dedicated processor are separately prepared for each function, and these modules are hierarchically stacked to meet this demand.

In FIG. 3, the main module X performs main document editing, file processing and communication processing so that the main CPU (central processing unit) 1, main memory 2 and HD can be used.
3. AI that consists of ROM and is provided for processing Japanese
It includes a dictionary 4 and a LAN (local area network) controller 5 for controlling communication with the outside. Further, the main module X is provided outside the bus conversion unit 6, the input / output control unit 7, and the workstation for converting the data format according to the difference in bus specifications between the bus connected to the main CPU 1 and the bus connecting the outside of the device. External HD attached
9 for connecting the workstation and the workstation
(Small computer system interface) 8,
The printer 10 and the FD 11 are connected to the main CPU 1 via the control unit 7. The above HD3 and external H
D9 is provided as a virtual storage device for virtually increasing the capacity of the main memory 2.

The graphic module Y is provided with a graphic CPU 12, a vector drawing unit 14, a CRT (cathode ray tube) 16 and a CRT so as to perform highly accurate display control.
A frame memory 15 provided in association with 167 is included.

The input control module Z is the input control CPU 1
7, an input / output control unit 20, an externally operable mouse 21 and a keyboard 22 connected to the input control CPU 17 via the input / output control unit 20.

Each of the modules X, Y and Z has its own system software (including an operating system). The system software of the module X includes a main memory management program (hereinafter referred to as main memory management) for controlling and managing the use of the main memory 2. The system software of module Y includes a graphic memory management program (hereinafter referred to as graphic memory management) to control and manage the use of memory in the module.

As described above, in the workstation,
The input control module Z, the graphic module Y, and the main module X are stacked in this order from the side closer to the input side, and the shared memory 1 is provided between the layers, that is, between the main module X and the graphic module Y.
3 is provided, and the shared memory 18 and the local memory 19 are provided between the graphic module Y and the input control module Z, thereby realizing an architecture for exchanging information between the modules. The memories 2, 13, 15, 18, and 19 are semiconductor memories that can be accessed at high speed. In addition, the distributed processing and the multiplexing of the buses adopting the hierarchical structure as described above can centrally deal with large-capacity data processing such as image processing to highly responsive processing such as mouse control. ..

In the workstation of FIG. 3, the graphic module Y for handling image data has a data buffer for printing (hereinafter referred to as a printing buffer).
A vast memory space is prepared in advance on the shared memory 13 in FIG. Since the print buffer is provided in association with the printer 10 and is not required except during the printing operation of the printer 10, it is normally the idle area UE. Therefore,
When the program is rolled out from the main memory 2, the main module X first directs the idle area UE of the shared memory 13 as the roll-out destination of the program,
Next, the secondary storage device HD3 or the external HD9 is directed.

When the system of the workstation of FIG. 3 is started up, the print buffer of the shared memory 13 is often the idle area UE. Therefore, immediately after the system is started up, the shared memory 13 is unconditionally directed to the first as the roll-out destination of the program in the main memory 2. Since the printing buffer stores the printing data as needed and is not in the idle area UE during the operation of the workstation, at this time, the secondary storage device is used as the roll-out destination of the program in the main memory 2. A certain HD3 or external HD9 is directed.

In the following, in order to simplify the explanation, the first roll-out destination of the program in the main memory 2 is the first.
It is assumed that the idle area UE of the shared memory 13 is pointed to and the second is the external HD 9.

FIGS. 4 (a) to 4 (c) show the usage status of the memory area in the workstation according to the control method of FIG. 1 when the workstation dedicated to document processing of FIG. 3 is operating.

FIG. 4A shows that when the module X needs to allocate a section (memory area) of xKB in the main memory 2 when the workstation is operating, an empty area EE (> xKB) is created in the main memory 2. because there was,
A situation in which a section of xKB is allocated to the main memory 2 and a section number P is given to this section is shown.

FIG. 4B shows a case where the module X rolls out the program α of the size yKB of the partition number Q stored in the main memory 2 to the idle area UE of the shared memory 13 when the workstation is in operation. .. At this time, a free area of yKB is newly secured in the main memory 2, and the shared memory 13 includes a roll-out area LOM of yKB due to the roll-out of the program α.

FIG. 4C shows a case where the module X rolls out the yKB program α in the main memory 2 to the HD 9 when the workstation is operating. At this time, a free space of yKB is newly secured in the main memory 2, and the rolled-out program α is stored in the HD 9.
The roll-out area LOM of yKB in which is stored is included.

FIG. 5 is a flow chart showing a procedure for using the memory area according to the control method of FIG. 1 when the dedicated document processing workstation of FIG. 3 is operating.

FIGS. 6 (a) and 6 (b) are diagrams showing an example of the configuration of a table referred to during the execution of the process according to the flow of FIG.

FIG. 6A shows the memory management table MT.
An example of a B configuration is shown. FIG. 6B shows an example of the structure of the free space table ETB.

Each of the memory management table MTB and the free area table ETB of FIG. 6A is provided individually corresponding to each memory.

In FIG. 6A, the table MTB shows a partition number f, a capacity cp, and a partition number f for each partition provided in the memory.
The start address hb and the state st of the area are included. Section number f is
This is a number assigned to uniquely identify the partition on the memory, and as shown in FIG. 6A, it is continuous in ascending order, for example. The capacity cp indicates the size of the section, the start address hb indicates the start address of the section on the memory, the state st of the area indicates whether or not a program or data is stored in the section, and is stored. If there is no store, it is "empty", and if it is stored, "assignment"
Is set.

The empty area table ETB of FIG. 6B is provided in association with the memory management table MTB, and the one of the areas of the memory managed by the memory management table MTB that has the area state st = “empty”. Manage centrally. That is, of the partitions registered in the memory management table MTB, the reference flag F, partition number DN, capacity CP, and
The head address HB and the state ST of the area are provided. The reference flag F is set to "1" when a program or data stored in the section is referred to (accessed). The partition number DN is a number for uniquely identifying the partition and matches with that of the memory management table MTB.
The capacity CP indicates the size of the section, the head address HB indicates the head address in the memory of the section, and the state ST of the area indicates that the section is “empty” like the state st of the area of the memory management table MTB. Indicates whether it is an "allocation". In this case, "empty" is set in the area state ST to indicate that the partition is an empty area.

Since the main memory management of the workstation of FIG. 3 accesses the memory management table MTB and the free area table ETB provided for the main memory 2, the data stored therein is read and written by the main memory management. .. Similarly, the data stored in the memory management table MTB and the empty area table ETB provided for each memory in the graphic module Y are read and written by the graphic module management. In this way, the data in the tables MTB and ETB provided for each memory is updated according to the usage status of the memory when the workstation is operating.

The procedure of FIG. 5 will be described below with reference to FIGS. When the workstation of FIG. 3 is powered on and the system is booted, the workstation boots and begins data processing. It is assumed that the main module X requests, for example, a temporary buffer of a capacity xKB on the main memory 2 while the workstation is operating. In response to the allocation request for this partition, a series of processing for allocation request processing for the partition of capacity x KB in FIG. 5 is started.

When the main module X requests allocation of a section of capacity xKB, this is processed as an interrupt, and the subroutine SUB1 of step S1 of FIG. 5 (abbreviated as S1 in the figure).
The process moves to. The details of the subroutine SUB1 will be described later, but here, the main memory management processes so as to allocate the section of the capacity x KB on the main memory 2. When the section allocation processing by the subroutine SUB1 is completed, the process returns to the module X and in the next step S2, it is determined whether or not the section allocation in the memory 2 is successful. If it is determined that the partition allocation has succeeded as shown in FIG. 4A, the process proceeds to the next step S3, and the partition number (= P) allocated in the main memory is processed by the immediately preceding interrupt. To the program that was suspended and restart the program's execution.

If the partition of the capacity xKB cannot be allocated on the main memory 2 in the process of step S2, the process proceeds to the subroutine SUB2 of the next step S4.

The details of the processing of the subroutine SUB2 will be described later, but here, in order for the main memory management to secure the temporary buffer of the capacity xKB in the main memory 2, y (≧ x is not executed in the main memory 2). ) K
Select the program α of B. This selected program α is a program that is rolled out from the main memory 2. When the program α to be rolled out is selected, the process of the subroutine SUB2 ends, and the process proceeds to step S5.

In step S5, the process of the above-mentioned subroutine SUB1 is started. The subroutine SUB1 of step S5 is executed by the graphic memory management, and the graphic memory management is performed by the above-mentioned step S5.
The allocation of the yKB partition, which is the roll-out destination of the program α selected in 4, is performed on the shared memory 13. After this allocation process is completed, the process returns to the module X and moves to the next step S6.

In step S6, it is determined whether or not the allocation of the yKB section on the shared memory 13 has succeeded by the graphic memory management in step S5.
If it is determined that the allocation is successful, the next step S7
Process shifts to.

In the process of step S7, the program α in the main memory 2 is rolled out to the shared memory 13 in response to the allocation of a new partition of yKB to the shared memory 13 by the graphic memory management. To be done. FIG. 4B shows this state.
In FIG. 4B, the program α to be rolled out on the main memory 2 is a roll-out area LO divided into the shared memory 13 by the graphic memory management.
Rolled out to M. Therefore, the main memory 2
There is an empty area of the section number Q on the top. After that, the process proceeds to the process of step S9 described later.

Returning to the processing of step S6, if a new partition of the capacity yKB cannot be allocated on the shared memory 13 by the graphic memory management in step S5, the processing of step S8 is executed.

In the process of step S8, the program α to be rolled out on the main memory 2 is rolled out to the HD 9. This state is shown in FIG. In FIG. 4C, the program α to be rolled out on the main memory 2 is the rollout area LO on the HD 9.
It is shown to be rolled out to M. This allows
A free area of capacity yKB is created on the main memory 2.
Thereafter, the processing shifts to step S9.

The process of step S8 described above is unnecessary and is not performed when the program α is already stored in the HD 9.

In the processing of step S9, the main memory management is performed so that the partition (partition number = Q) used by the program α on the main memory 2 becomes an empty area. Although the details of this processing will be described later, when the processing of opening the partition of the partition number = Q on the main memory 2 is performed, the processing returns to the processing of step S1 described above, and the main module X allocates a new partition. Be prepared for the next interrupt generated by issuing a request.

As described above, when the main module X requests the partition allocation, the subroutine SUB1 in step S1 first allocates the partition on the main memory 2. If the partition cannot be allocated on the main memory 2 at this time, the processes of steps S4 to S9 are executed to first roll out the program α on the main memory 2 to the shared memory 13. If the program α on the main memory 2 cannot be rolled out to the shared memory 13, it is rolled out to the HD 9. Thus, when a new partition allocation request occurs in the main memory 2, the shared memory 13 is first directed as the roll-out destination of the program to be rolled out from the main memory 2, and the roll-out to the shared memory 13 is impossible. In that case, the rollout to the HD 9 is directed next.

FIG. 7 shows the subroutine SU shown in FIG.
It is a processing flow figure of B1. FIG. 8 is a process flow diagram of the subroutine SUB2 shown in FIG.

FIGS. 9A to 9D are diagrams showing examples of the main memory utilization state obtained when the procedure of FIG. 5 is followed.

FIG. 10 is a subroutine S shown in FIG.
It is a processing flow figure of UB3. Next, the procedure for using the memory area shown in FIG. 5 will be described with reference to FIGS.

In FIG. 5, when the main module X requests allocation of a section of capacity xKB, this is processed as an interrupt, the process of step S1 is executed, and the subroutine SUB1 of FIG. 7 is executed.

The subroutine SUB1 in FIG. 7 has a capacity x KB.
In response to the allocation request for the section, the program is executed under the control of the main memory management.

When there is a request from the main module X to allocate a new section of the capacity xKB on the main memory 2, the main memory management starts execution of the subroutine SUB1. First, the free space table ETB relating to the main memory 2 is searched from the beginning while repeating the loop processing including steps S20 to S24 (capacity CP ≧ x
Identify the partition that is KB). At this time, as a result of searching the table ETB from the beginning to the end by repeating the loop processing, if it is determined that there is no partition of (capacity CP ≧ xKB), a new partition of capacity xKB on the main memory 2 is determined. Since the assignment is not possible, the process of step S25 is performed after exiting this loop process, and the subroutine SU
The process of B1 is ended, and the data indicating that the allocation is impossible is passed to the main module X.

On the other hand, when the free area table ETB is searched from the beginning and one partition having (capacity CP.gtoreq.xKB) can be specified, this loop processing is immediately exited and step S
The processing of 26 to S31 is executed.

In the processing of steps S26 to S31, the empty area table ET of the main memory 2
B is searched and stored in the main memory 2 (capacity CP ≧ xK
Table ETB regarding the specified section in the main memory 2 in response to the determination that there is a section that is B).
And the data updating process of the table MTB is performed. That is, if the specified partition is (capacity CP = xKB), all the data in the table ETB is emptied (sets meaningless data) for the partition specified in the process of step S26. If (capacity CP> xKB), the data in the table ETB is x for the partition specified in the processing of steps S27 to S29.
Since it is updated using KB, the capacity CP of the partition is (CP
-X) KB, and the head address HB becomes (HB + x).

After updating the data of the free area table ETB for the above-mentioned specific partition, the partition number DN of the specified partition is set to the partition number P. In addition, the main memory 2
Since a new partition allocation of the capacity xKB has been made in the above, the state st of the area of the table MTB for the partition number P is set to "allocation".

As described above, the empty area table E is used so that the new partition allocation of the capacity xKB is performed on the main memory 2.
The TB and the memory management table MTB are updated, and the partition number P newly allocated on the main memory 2 is passed to the main module X, and the subroutine S is executed.
The process of UB1 ends.

Returning to the processing of step S2 in FIG. 5, the data passed from the above-mentioned subroutine SUB1 is discriminated, and when the partition number P is passed, it means that the partition allocation of the capacity xKB on the main memory 2 has succeeded. To the next step S
Then, the process returns to the main module X.
At this time, the new partition allocation state on the main memory 2 is as shown in FIG.

When the data indicating that allocation is impossible is passed from the subroutine SUB1, it is determined in the process of step S2 that the partition of the capacity xKB cannot be allocated on the main memory 2, and the subroutine SUB of step S4 is executed.
The processing of B2 is executed.

FIG. 8 shows the processing of the subroutine SUB2. This processing is executed under the control of main memory management.

The subroutine SUB2 of FIG. 8 is one of the programs stored in the memory 2 when the main memory 2 does not have a free area for partitioning the capacity x KB in the processing of the above-mentioned subroutine SUB1. In this process, an unnecessary program is selected and rolled out to newly create a free area in the memory 2 that allows partition allocation of the capacity xKB.

More specifically, the free area table ETB related to the main memory 2 is sequentially searched from the beginning, and it is first determined whether or not the reference flag F (F = 1) is satisfied (step S42). If (F = 1), the program stored in the partition has been recently referenced (accessed), so it is excluded from the program candidates to be rolled out, and then it is determined whether or not the program is being executed. (Step S49). At this time, if it is determined that the program is being executed, the next section of the free area table ETB is searched.
On the contrary, if it is determined that the program is not being executed, the reference flag F is reset to (F = 0) (step S50), and the next partition of the free space table ETB is searched.

Returning to the processing of step S42 described above, if the reference flag F (F = 0), it is determined to be a program candidate to be rolled out, and whether (capacity cp of the partition ≧ capacity x) is satisfied or not. It is determined (step S4)
3). At this time, if it is determined that the condition is satisfied, the partition of the capacity x can be allocated in the partition, so the program stored here is selected as the program α to be rolled out (step S44). Then, the processing of the subroutine SUB2 ends. Conversely, step S43
In the determination process of (the capacity of the partition cp <the capacity x)
If so, the next area of the free area table ETB is searched to determine whether or not the reference flag is (F = 1). If (F = 1), the processes from step S49 onward are executed, and if (F = 0), the capacity of the section is added to the capacity for allocating the capacity x (step S49).
48). Then, the process returns to step S42.

As described above, in the subroutine SUB2, among the programs stored in the main memory 2, an area (capacity in which the sections for storing the programs that have not been referred to recently and have not been executed are continuously allocated (capacity). yK
B) is specified and the program stored therein is determined as the program α to be rolled out. When the program α to be rolled out from the main memory 2 is determined in the subroutine SUB2, the process proceeds to the process of the subroutine SUB1 shown in step S5 of FIG. 5, and the control proceeds to the graphic memory management.

For graphic memory management, the subroutine SUB1 of FIG. 7 is executed in the same manner as described above, and the main memory 2
An attempt is made to allocate a partition of the capacity yKB capable of storing the program α rolled out from the shared memory 13 to the shared memory 13. If the partition SUB1 can allocate the partition of the capacity yKB on the shared memory 13, the partition number is passed to the main module X, but if the partition allocation cannot be performed, data indicating that the allocation is impossible is passed to the main module X. ..

The partition allocation result of the capacity yKB on the shared memory 13 in step S5 is determined in step S6, and if the partition number is passed, the process proceeds to step S7, and conversely the data indicating that partition allocation is impossible. If is passed, the process proceeds to step S8.

In the processing of step S7, the main module X rolls the program α on the main memory 2 as shown in FIG. 4B in response to the partition allocation of the capacity yKB being completed on the shared memory 13. Out area LOM
Roll out to. This rollout area LOM is
In step S5 described above, the graphic memory management is a section allocated on the shared memory 13 by the size of the capacity yKB. In step S8, the roll-out area LOM could not be allocated to the shared memory 13, so that FIG.
As shown in (c), the program α on the main memory 2 is rolled out to the HD 9.

In this way, after the program α on the main memory 2 is rolled out, the processing of the next step S9 is executed.

In the processing of step S9, the processing for releasing the partition storing the program α rolled out to the shared memory 13 or the HD 9 on the main memory 2 is performed by the main memory management. That is, FIG.
Since the request for releasing the partition number Q on the main memory 2 shown in (b) or FIG. 4 (c) is issued to the main memory management, the main memory management is shown in step S9 in response to the release request. The processing of the subroutine SUB3 is executed.

The main memory management responds to the request to release the partition number Q, and the subroutine SUB3 shown in FIG.
Start executing.

Now, the partition of the partition number Q in which the program α is stored exists in the memory 2 in any of the states shown in FIGS.

FIG. 9A shows a state in which the partition adjacent to the partition of the partition number Q on the memory 2 is not empty.
Similarly, (b) and (c) show a state in which one of the adjacent sections is empty, and FIG. 9 (d) similarly shows a state in which both adjacent sections are empty.

In the main memory management, the subroutine SUB3 of FIG. 10 is executed to first determine in which state of FIG. 9A to 9D the partition of the partition number Q is allocated in the main memory 2. Determine. Then, based on the determination result, when the empty partitions including the partition of the partition number Q on the main memory 2 are adjacent to each other, the memory management table MTB and the empty area table ETB are continuously set to one empty partition. Perform data update processing.

First, the main memory is managed by the main memory 2
Table MTB regarding the partition number Q and the capacity CP and the head address hb of the partition corresponding to the partition number Q are read (variable C
1 = capacity cp, variable B1 = start address hp), and the state st of the area of the table MTB is set to “empty” (steps S61 and S62).

Then, it is determined which of the states in FIGS. 9A to 9D the section with the section number Q is assigned in the memory 2 (step S63).

As a result of the discrimination, in the case of the allocation state of FIG. 9A, a series of data regarding the section of the section number Q is newly written in the empty area table ETB. Therefore, in the empty area table ETB, for the partition with the partition number Q in FIG. 9A, the capacity CP = C1, the head address HB = B1, the partition number DN = Q, the area state ST = “empty” Additional settings are made (steps S64 and S6).
5).

As a result of the discrimination, in the allocation state of FIG. 9B or 9C, one adjacent empty space of the empty area table ETB is merged so that one adjacent empty space and the area Q are merged. The data regarding the section is updated (step S66).
-Step S68).

As a result of the discrimination, in the allocation state of FIG. 9D, three empty areas including the partition of the partition number Q are adjacent to each other. Free area table E so that it becomes an empty partition
TB is updated (steps S69 and S7).
0).

As described above, FIG.
According to the states of (a) to (d), the data of the free area table ETB is updated, and the partition Q on the main memory 2 is updated.
Is set to be an empty area. Then subroutine S
The processing of UB3 ends, and the processing is performed again on the main module X.
Return to.

As described above, when the main module X requires a storage capacity exceeding the capacity of the main memory 2, the program to be rolled out is selected from the main memory 2 and the roll-out destination of the selected program is first selected. Since the shared memory 13 that can be accessed at a relatively high speed is directed, the time required for program rollout is shortened as compared with the conventional case in which the rollout destination is the HD 9.

In the above-mentioned embodiment, the description of the operation when the graphic module Y requests the allocation of the partition of the capacity xKB on the shared memory 13 is omitted, but the allocation of the partition in this case is omitted. The operation is performed in the same processing routine as that of the main module X described above.

If a partition of the same size as the partition in which the program α rolled out from the main module X was stored on the side of the graphic module Y for some data processing, the above-mentioned FIG. Processing similar to is performed. However, the graphic module Y preferentially extracts the main program in the main module X in the upper layer as a program candidate for rollout.

Although the above-described embodiment has a multiprocessor configuration, a single processor may be newly provided with a memory corresponding to the shared memory 13.

Although the above-described embodiment uses only the shared memory 13, the present invention is not limited to this, and it is also possible to use free areas of other fast accessible memories in the apparatus of FIG. ..

[0113]

As described above, according to the present invention, when the data stored in the first storage means is rolled out,
Data is selectively rolled out to either the second or third storage means. Further, when the first determination unit determines that the area cannot be secured in response to the request to secure the area of the first capacity in the first storage unit, the data selected by the selection unit is the first transfer unit. Alternatively, after the data is transferred to the second storage means or the third storage means by the second transfer means, the setting means enables the area to be secured in the third storage means. In this way, the data rolled out from the first storage means or the data selected by the selection means in response to the request to secure the area of the first capacity in the first storage means is either the second storage means or the third storage means. Since the data is rolled out or transferred to either side, the data transfer speed can be improved as compared with the case where the rollout destination or the transfer destination is uniformly set in the second storage means.

The above-mentioned speed-up related to rollout or data transfer can be achieved only by providing the information processing apparatus with the third storage means, or in the information processing apparatus provided with the first storage means, the second storage means and the third storage means. Has been realized by merely effectively using these existing storage means, and therefore, there is an effect that cost performance can be improved.

[Brief description of drawings]

FIG. 1 is a diagram illustrating a control method when a memory area is used in an information processing apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing an area usage state and a program roll-in / roll-out status in association with each other when the memory area shown in FIG. 1 is used.

FIG. 3 is a block diagram of a document processing dedicated workstation according to an embodiment of the present invention.

4A to 4C are diagrams for explaining a memory area utilization state in the workstation according to the control method in FIG. 1 when the workstation dedicated to document processing in FIG. 3 is operating.

5 is a flowchart showing a procedure for using a memory area according to the control method of FIG. 1 when the document processing dedicated workstation of FIG. 3 is in operation.

6 (a) and 6 (b) are diagrams showing an example of the configuration of a table that is referred to during processing execution according to the flow of FIG.

FIG. 7 is a process flow diagram of a subroutine SUB1 shown in FIG.

FIG. 8 is a process flow diagram of a subroutine SUB2 shown in FIG.

9A to 9D are diagrams showing an example of a main memory use state obtained when the procedure of FIG. 5 is followed.

FIG. 10 is a process flow diagram of a subroutine SUB3 shown in FIG.

FIG. 11 is a diagram for explaining a conventional roll-in / roll-out method.

FIG. 12 is a diagram showing a comparison of access speeds between standard storage devices in a table format.

[Explanation of reference numerals] 2, MM main memory, SM sub-processor memory, HD magnetic disk storage device, LO1, LO2, and LO3 rollout 13 shared memory MTB memory management table ETB free area table Indicates.

Claims (4)

[Claims] 1. An information processing apparatus comprising: a first storage means having a high access speed; and a second storage means having a low access speed but a large capacity, wherein the access speed is the same as that of the second storage means. A third storage unit having a higher speed than that is further provided, and when the data stored in the first storage unit is rolled out, the data is selectively stored in one of the second and third storage units. An information processing device characterized by being rolled out. 2. A first storage means having a high access speed, a second storage means having a low access speed but having a large capacity, and an access speed being higher than that of the second storage means. In an information processing device including a third storage means, it is determined whether or not the first storage area can be secured in the first storage means in response to a request to secure a first capacity area in the first storage means. First determining means for determining, and the first determining means for determining whether the first determining means is impossible.
At least one data of the data group stored in advance in the storage means that matches a predetermined reference and has a sum of the capacities of the respective areas for storing them that is the second capacity that is equal to or larger than the first capacity. Selecting means for selecting; area securing means for securing an area having the second capacity in the third storage means; second determining means for determining whether or not the area securing means has secured the area; 2 first transfer means for transferring the data selected by the selection means to the second storage means in response to the determination by the determination means that the data cannot be secured; Second transfer means for transferring the data selected by the selection means to the area having the second capacity of the third storage means secured by the area securing means; and the first or second transfer means. After the data transfer, further comprising setting means for setting to allow secure an area constituting the second capacitor in the first storage unit, the information processing apparatus. 3. The information processing apparatus according to claim 2, wherein the predetermined criterion includes that the predetermined criterion is stored in a continuous area and is referred to less frequently. 4. The information processing apparatus according to claim 1, wherein the data includes a program.