KR960012658B1 - Information processing system - Google Patents

Abstract

No summary

Description

System for processing temporary storage files by local processor

1 is a block diagram of the present invention.

2 is a flowchart showing the processing of a table calculation program operating on a central processing unit during sorting in table calculation, which is one of embodiments of the present invention.

3 is a flowchart showing the processing of a compiler program operating on a central processing unit in the case of a compiler which is one of the embodiments of the present invention.

4 is a memory configuration diagram of the local memory 4.

5 is a configuration diagram of a table or a work area placed in the system area 41. FIG.

* Explanation of symbols for main parts of the drawings

1 central processing unit 2 sub processing unit

3: local processor 4: local memory

5: external memory

In the information processing system, for example, table calculation, arithmetic processing in a database, processing by a sort or compiler, file conversion, and the like using a temporary storage file without degrading the performance of the central processing unit. It relates to a data processing method that can be processed efficiently.

In the information processing system, when the above-described processing is performed using the temporary storage file, conventionally, a system has been adopted in which the temporary storage file is usually stored in an external storage device such as a disk and the processing is performed by a central processing unit. However, in the case where the contents of the temporary storage file are a large amount of data, there is a problem in that the efficiency of the system gradually decreases due to the increase in the load of the central processing unit when a plurality of users start to concentrate and use them. Therefore, in the technique disclosed in Japanese Patent Laid-Open No. 63-85853, a large amount of data is processed in a separate processor to reduce the load on the central processing unit. In the technique of Japanese Patent Laid-Open No. 63-85853, the bulk data to be subjected to batch processing is copied once by copying the bulk data stored in an external storage device such as a disk storage device onto a working disk in a processing device that performs batch processing. Using a local memory in the processing device, a local processor in the processing device processed a large amount of data on the working disk, and when the processing was finished, the processing result on the working disk was copied back to the external storage device.

The technique described in Japanese Patent Laid-Open No. 63-85853 requires two data copies in which data to be processed from an external storage device such as a disc is copied to the working disk, and the processing results are copied back from the working disk to the external storage device. This copying of data requires a considerable amount of time when the data becomes large and cannot be ignored.

It is an object of the present invention to provide a means for processing data at a higher speed by shortening the time required for copying without compromising the purpose of reducing the load of the original central processing unit.

The invention is explained according to FIG. 1 is a block diagram illustrating the concept of the present invention. 1 is a central processing unit (CPU), 2 is a sub-processing unit, 3 is a local processor (LP) in the sub-processing unit 2, and 4 is a local memory (i.e., a main storage unit in the sub-processing unit 2). LM), 5 is an external memory device ES, 6 is an I / O bus which connects between the central processing unit 1, the sub-processing device 2, and the external storage device 5;

The present invention creates a temporary storage file on the local memory 4 which can be directly accessed by the central processing unit 1 shown in FIG. This temporary storage file can create / delete / read / write data in the same manner as the temporary storage file created on a normal external storage device in a program running on the central processing unit 1, and the same program. The contents can be copied in both directions between the file and the file created on the external storage device 5 in accordance with the instructions in. In addition, on the local memory 4, programs such as table calculation, arithmetic processing in a database, sorting or compiler, and file conversion can be loaded / unloaded from the external storage device 5. A plurality of programs can be loaded on the local memory 4 and are not deleted until an unload instruction is given. One or more programs operating on the central processing unit 1 designate one or more temporary storage files on the local memory 4 as processing targets and are loaded on the local memory 4. Programs can be designated as the processing means to instruct processing to the local processor 3. The local processor 3 starts the designated program according to the instruction from the central processing unit 1, processes the designated temporary storage file group, and communicates the end of processing to the program operating on the central processing unit 1. While the local processor 3 is performing the processing, the central processing unit 1 can execute other processing.

Moreover, the subprocessing apparatus 2 can be attached in multiple numbers in a system.

As described above, a program operating on the central processing unit 1 shown in FIG. 1 creates a temporary storage file directly on the local memory 4. The program may read / write the temporary gear file directly or instruct to copy the contents from the file on the external storage device 5 onto the local memory 4 as a temporary storage file. Thus, the program may simply use the local memory 4 as a temporary high speed storage device.

However, the program may instruct the local processor 3 to process the temporary storage file. The sequence first checks whether the required program is loaded on the local memory 4 and loads the program if it is not loaded. Since a plurality of programs can be loaded on the local memory 4, loading necessary programs in advance can save time without having to load them at runtime. Thereafter, one or more temporary storage files to be processed are designated, and one program serving as the processing means is designated to instruct the local processor 3 to process. After the instruction, the program on the central processing unit 1 may perform other processing or may wait for the end of the processing in the local processor 3. While waiting, another program runs on the central processing unit 1. When the processing on the local processor 3 ends, a termination communication is sent to the program on the central processing unit 1 which instructed the processing. Since the processing result is reflected in the temporary storage file on the local memory 4, the program on the central processing unit 1 is immediately accessible as a result. Alternatively, the local processor 3 may be instructed to perform other processing on the temporary storage file again.

As described above, in the present invention, the working disk is not a necessary component, and therefore, a plurality of data or processing results before the processing from the external storage device 5 to the working disk is transferred from the working disk to the external storage device 5. The copying process is unnecessary. Therefore, the result can be obtained by processing faster than before. In addition, while the processing is being performed on the local processor 3, the central processing unit 1 can perform other processing, thereby reducing the load on the central processing unit 1.

When there are a plurality of sub-processing apparatuses 2, the load on the sub-processing apparatus 2 can be distributed when a plurality of users simultaneously use the sub-processing apparatus 2, so that the processing performance is 1 It's even better than it is.

Hereinafter, the present invention will be described with reference to Examples.

The first embodiment is a sorting process in table calculation. This is illustrated by FIGS. 1 and 2. FIG. 2 is a flowchart showing the processing of a program operating on the central processing unit 1 in the case of sort processing in table calculation.

The table calculation program is operated on the central processing unit 1 in FIG. The program creates a temporary storage file on the local memory 4 (step 11), and stores data input from the user in the program (step 12). Alternatively, data input in the past may be plural from a file stored in the external storage device 5 to a temporary storage file (step 13). The control of the copy may be directly performed by the central processing unit 1 or may be instructed to the local processor 3. Assume that the user specifies sorting for any item in the table for the table data of the temporary storage file thus created. The program then checks whether or not the sort program is loaded on the local memory 4 (step 14), and loads the sort program if not loaded (step 15). Thereafter, if necessary, a temporary storage file to be a work area is designated, and a sort program is designated as processing means to instruct the local processor 3 to process (step 16). After the instruction, the processing on the local processor 3 is waited for to end (step 17). While the program is waiting, another program is operated on the central processing unit 1. When the end processor receives the termination notification from the local processor 3, the central processing unit 1 resumes the operation and proceeds to the next processing such as reading the temporary storage file containing the processing result (step 18).

As in the above embodiment, if a part of the functions included in the program is extracted and set as another program, and the program is executed by the local processor, the central processing unit can be processed and the temporary storage file can be processed.

The second embodiment is a compiler, which is explained by referring to FIG. 1 and FIG. 3 is a flowchart showing processing of a compiler program operating on the central processing unit 1.

The compiler is usually divided into several phases, and between each phase, what is called an intermediate language, is sent and received by a temporary storage file. Assume that the compilation process is divided into three phases.

On the central processing unit 1 in FIG. 1, the root segment of the compiler program is operating. First, the program loads each program of phases 1 to 3 onto the local memory 4 (step 21). Next, temporary storage files T1, T2 and T3 are created on the local memory 4 (step 22). T1 is used as the storage area of the source program, T2 is used as the storage area of the compilation list, and T3 is used as the storage area of the intermediate language transferred from phase 1 to phase 2. In T1, a source program to be compiled is copied from a file on the external storage device 5 (step 23). Control of copying is the same as in the first embodiment. The operating environment of phase 1 is established as above, and the compiler instructs the local processor 3 to execute the phase 1 program by designating T1 as an input file and T2 and T3 as an output file (step 24). . When phase 1 ends, T1 is no longer needed and is deleted (step 25). Then, to enter the phase 2, the temporary storage file T4 is created as the storage area of the intermediate language transferred from phase 2 to phase 3 (step 26). Then, T3 containing the intermediate language transferred in phase 1 is designated as an input file, and T2 and T4 are designated as output files, instructing the local processor 3 to execute the phase 2 program and waits for the end thereof (step). 27). At the end of phase 2, T3 is no longer necessary and is deleted (step 28). Finally, to enter the phase 3, the temporary storage file T5 is created as an area for storing the target program which is the final output of the compiler (step 29). Then, T4 containing the intermediate language transferred in phase 2 is designated as an input file, and T2 and T5 are designated as output files, and the local processor 3 is instructed to execute the phase 3 program (step 30). . When phase 3 ends, T4 is no longer needed and is deleted (step 31). This completes the compilation process. The remaining temporary storage files are T2, which is a storage area of the compilation list, and T5, which is a memory area of the target program. T5 copies to a file of an external storage device by the target program (step 32). T2 is read as a compilation list and output to the printer (step 33). As above, T2 and T5 are no longer needed and are deleted (step 34). As a post-process, the programs of phases 1 to 3 are unloaded and the compiler terminates (step 35).

By writing the structure of the compiler in this way, it is possible to compile with little load on the central processing unit. The present invention is particularly effective for a program that creates several temporary storage files and processes them as in the above embodiment.

Hereinafter, the system configuration and operation will be described in detail. First, the local memory 4 of the subprocessing apparatus 2 may be referred to through an area (window) allocated for the local memory 4 in the logical address space of the CPU 1. That is, the local memory 4 can be read / written from the CPU 1. The local memory 4 can also be referred to via the I / O bus 6 by the DMA mechanism of the external storage device 5. The reference of the local memory 4 by the DMA mechanism is made through a window allocated to the local memory 4 in the logical address space of the CPU 1. In addition, the CPU 1 and the LP 3 can interrupt each other through an interrupt signal line.

4 is a memory configuration diagram of the local memory 4. The system area 41 is an area in which various tables and the like of the system program executed by the LP 3 are stored. The program load area 42 is an area in which a plurality of programs are stored, and the program work area 43 is a work area used by these programs. The file area 44 is an area in which a plurality of temporary storage files are stored.

5 constitutes a table and a work area which lie within the system area 41. FIG. 5A shows a communication area between the CPU 1 and the LP 3. 51 is an area used for communication from the CPU 1 to the LP 3, and 52 is an area used for communication from the LP 3 to the CPU 1. The message sent from the CPU 1 to the LP 3 is as follows.

(Iii) Program start request

(Ii) file allocation request

(Iii) File deletion request

On the other hand, the message sent from the LP 3 to the CPU 1 includes the following.

(Iii) program termination notice

(Ii) file allocation termination contact

(Ⅲ) file deletion end contact

FIG. 5B shows the configuration of a table for managing a program loaded into the program load area 42. As shown in FIG. 61 is the name of the program, 62 is the head address on the local memory 4 into which the program is loaded, 63 is the last address, and 64 is the entry point where the program is first controlled.

5C is a file management table for managing the area of each file on the file area 44. FIG. The file area 44 is divided into memory blocks having a storage capacity of 128 KB, for example.

74 and 75 are queue elements corresponding to each memory block.

Memory blocks are numbered starting at 1, and the numbers are stored in the corresponding cue elements. It represents a chain of queue elements, ie which files are using which blocks of memory in which order. The head and last queue elements of each queue are pointed by each entry in the file management table. The entry (pre-memory block 71) is an entry in which a pointer of a chain of unused memory blocks is stored. Entries 72 and 73 are entries designed for each file number, and entry 72 is an entry in which a pointer of a memory block chain in use is stored. The entry 73 is also an entry corresponding to a file number to which a memory block is not allocated. 74 is a queue element corresponding to the free memory block, and 75 is a queue element corresponding to the memory block being used. In order to use the expected storage method, the subprocessing apparatus 2 maps a file to a logical space, and the entry 72 stores the logical address of the file.

The order in which the CPU 1 creates the temporary storage file on the local memory 4 is as follows. The CPU 1 checks the number of free memory blocks corresponding to the queue element 74 pointed by the entry 71 of the file management table and checks whether the temporary storage file can be allocated. Next, check whether there is an empty entry point. If the file area can be secured, the file allocation request message including the number of memory blocks required in the area 51 of the system area 41 is written, and the LP 3 is interrupted. The LP 3 updates the entry 71 by extracting as many queue elements as necessary from the queue elements 74 of the free memory block pointed by the entry 71 of the file management table. Next, the chain of queue elements 75 corresponding to the memory blocks 7 allocated to the empty entry 73 is connected. In addition, the LP 3 secures a continuous storage area in the logical space and creates an address translation table that associates the logical address with the address of the memory block. The LP 3 then interrupts the CPU 1 by writing a file allocation end contact message including the file number assigned to the area 52 of the system area 41. The CPU 1 can know the assigned file number by referring to the area 52.

Whether or not the desired program is loaded on the local memory 4 can be known by searching the program name 61 of the memory management table in the CPU 1. When the desired program is loaded on the local memory 4, the procedure for starting the program of the subprocessing apparatus 2 in the CPU 1 is as follows. The CPU 1 writes a program start request message including a program name, an assigned file number and input data into the area 51 of the system area 41 and interrupts the LP 3. The system program of the LP (3) creates a file table containing a logical address and a file size in a logical space in which a file is stored in a file number, and controls the entry point 64 of a designated program using the address of the table as a parameter. Turn over. The input data is copied from the area 51 to the user area, and the address is passed to the program designated as a parameter. It is also passed to the program specified as the address parameter of the area storing the output data output by the program. When the program execution ends normally, the LP 3 copies the output data from the user area to the area 52, creates a program termination contact message, and interrupts the CPU 1. In this way, the CPU 1 can take data output from the program. The input data / output data may indirectly specify the file number of the temporary storage file. Since the CPU 1 can directly access the local memory 4, the input data can be prepared as a temporary storage file on the local memory 4, and the output file on the local memory 4 can be read out.

The procedure for deleting the file allocated on the local memory 4 is as follows. The CPU 1 writes a file deletion request message including the file number of the file to be deleted in the area 51 and interrupts the LP 3. The LP 3 connects all the queue elements 75 connected to the entry 72 corresponding to the designated file number to the chain of the entry 71 to form the queue element 74 corresponding to the free memory block. The queue element 75 connected to 72 is deleted to convert the entry 72 into an empty file entry 73. Also, the file area secured in the logical space is deleted. When the file deletion ends, the LP 3 writes a file deletion end contact message to the area 52, interrupts the CPU 1, and notifies the file deletion end.

The procedure when copying a file on the external storage device 5 to a file area allocated to a temporary storage file on the local memory 4 is as follows. When the CPU 1 directly transfers the file, the CPU 1 reads the file on the external storage device 5 into its main memory, and is performed through a window allocated to the local memory 4 of the CPU 1. . It is also possible to issue a command from the LP 3 to the external storage device 5 and copy the file of the external storage device 5 to the file area on the local memory 4 by the DMA mechanism. In addition, when the program on the external storage device 5 is loaded onto the local memory 4, the same as in the case of a file, the program to be loaded into the program management table of the system area 41 is registered and the program is loaded into the program load area ( Read in 42). The unloading of the program on the local memory 4 is performed by deleting the target program from the program management table of the system area 41.

As described above, in the present invention, the temporary storage file is created on the local memory 4 in FIG. 1 and directly accessed by the central processing unit 1, or the local processor 3 is the temporary storage file. To process. As described in Japanese Unexamined Patent Application Publication No. 63-85853, a working disk is not necessary, and therefore, a plurality of data before processing from the external storage device 5 to the working disk and the processing result are transferred from the working disk to the external storage device 5. There is no need for copying, which enables faster processing than in the prior art. In addition, while the processing is being performed on the local processor 3, the central processing unit 1 can perform other processing, thereby reducing the load on the central processing unit 1.

Claims (6)

Subprocessing unit comprising a local processor specializing in processing for temporary storage files and a local memory storing temporary storage files, a central processing unit directly accessible to the local memory, and transferring data to and from the local memory. A system having an external storage device, and a transfer path combining the central processing unit and the subprocessing device with the external storage device, wherein the central processing unit creates / delete a temporary storage file on the local memory. Means, means for reading / writing the temporary storage file, means for loading a program from the external storage device on the local memory, and a temporary storage file on the local memory as a processing target and the local memory The local program by designating a program on the Means for instructing each other of processing, wherein said local processor has means for operating a designated program according to an instruction in said central processing unit to process a designated temporary storage file and to inform the central processing unit of the end of processing; A system for processing temporary storage files by a local processor. The system according to claim 1, wherein the program is a sorting program. The system according to claim 1, wherein the temporary storage file is a work area used by a compiler program. A local memory for storing a temporary storage file and a program for processing the temporary storage file, a local processor connected to the local memory to execute the program, a local memory connected to the local processor, and communicating with the local processor And a central processing unit accessible to the local memory, wherein the central processing unit stores data in the temporary storage file, starts the program through the local processor, and executes the program. At the end, the contents of the temporary storage file created by the program are read, the local processor operates the program in response to the start-up at the central processing unit to process the temporary storage file, and the process ends. Notification to the processing device A system for processing temporary storage files by a local processor. The system according to claim 4, wherein the program is a program that performs a sorting process. The system according to claim 4, wherein the temporary storage file is a work area used by a compiler program.