KR100807039B1 - Asymmetric multiprocessing system and method thereof - Google Patents

Abstract

The present invention relates to an asymmetric multiple processing system and method for processing a plurality of work groups consisting of sequential tasks. More specifically, in a multi-processing system for processing a plurality of work groups consisting of sequential tasks using a plurality of central processing unit, the task of the first task group is sequentially identified to select a specialized job and the specialized job Is processed using the first central processing unit, and during the processing of the specialized job, the general job not selected as the specialized job of the second working group is processed using the second central processing unit, and when the specialized job is processed, It relates to a multi-processing system, characterized in that for sequentially processing the general task of the next order of the specialized task in the first group of tasks using a second central processing unit.

Asymmetric multiprocessing, AMP, multiprocessor systems,

Description

Asymmetric multiprocessing system and its method {ASYMMETRIC MULTIPROCESSING SYSTEM AND METHOD THEREOF}

1 is a diagram illustrating an example of a VPN device implemented using a symmetric multiprocessing system.

FIG. 2 is a diagram illustrating a difference in performance of executing kernel code in a symmetric multiprocessing system using two CPUs and a system using one CPU.

3 is a diagram illustrating an example of a VPN device implemented using an asymmetric multiprocessing system.

4 is a diagram illustrating an internal configuration of an asymmetric multiple processing system according to an embodiment of the present invention.

FIG. 5 is a diagram illustrating a configuration of using a queue in assigning a specialized task in an asymmetric multiprocessing system according to an embodiment of the present invention.

6 is a flowchart illustrating a flow of a specialized task processing unit when a queue is used to allocate a specialized task in an asymmetric multiprocessing system according to an embodiment of the present invention.

FIG. 7 is a diagram illustrating a configuration of using an interrupt in assigning a specialized task in an asymmetric multiprocessing system according to an embodiment of the present invention.

8 is a diagram illustrating the performance of a VPN device implemented using an asymmetric multiprocessing system according to an embodiment of the present invention.

9 is a flowchart illustrating a flow of an asymmetric multiprocessing method according to an embodiment of the present invention.

<Explanation of symbols for the main parts of the drawings>

410: specialized work sorting unit

420: specialized work processing unit

430: general job processing unit

440: asymmetric processing settings

The present invention relates to an asymmetric multiple processing system and method for processing a plurality of work groups consisting of sequential tasks. More specifically, in a multi-processing system for processing a plurality of work groups consisting of sequential jobs using a plurality of central processing unit, screening with a specialized job and a specialized job selected from the jobs of the plurality of work groups Asymmetric multi-processing system and method for efficient operation by allowing different central processing units to process non-generic tasks simultaneously.

Recently, the computational processing speed of the CPU has sharply increased, and the storage capacity of the memory has increased exponentially, which greatly improves the performance of the computer. However, unlike memory performance improvement, the processing speed of the central processing unit is difficult to increase linearly. This is because the frequency increases to the gigahertz (GHz) level, which requires circuit analysis and design that is different from the conventional megahertz (MHz) class circuit, and also requires the problem of controlling the heat.

Therefore, recently, two or more CPUs are connected to one computer or two or more cores are in charge of processing in one CPU, without relying on the processing speed of one CPU. The method using the included central processing unit has been used. In general, as described above, a system including two or more CPUs is controlled by one operating system except for a super computer or a large computer designed for a specific purpose. Symmetrically Such a configuration is called a symmetric multi processing system.

Recently, a VPN (Virtual Private Network), which can replace a private network installed to exchange data between the head office, a domestic branch office, an overseas branch office, and a factory at a lower price, is widely used in a company. have. VPN (Virtual Private Network) is a new type of network proposed to provide various functions by forming logically closed user groups regardless of physical network configuration on public switched networks such as the Internet. It is advantageous to use the existing public network rather than physically installing a new leased line, so that the same effect as installing a leased line can be obtained at a low cost.

In order to provide a VPN, a VPN device is required. The VPN device transmits data to a predetermined server through a public Internet network used by an unspecified number of users. Packet). Since a packet is transmitted using a public network, the security of a packet may be the biggest problem in a VPN device, and an encryption and decryption module for a packet is built in for security. In addition, in order to enable fast transmission and response as if a dedicated line is installed, a VPN device requires a fast processing speed, and in order to enable a fast processing speed, the implementation of a VPN device using the above-described multiple processing system has been discussed.

FIG. 1 is a diagram illustrating an example of a VPN device implemented using a symmetric multiprocessing system having two central processing units.

As shown in the figure, since the two CPUs 110 execute and share a kernel code of one operating system, the first kernel is used in the kernel code. If a critical session of the system, such as a variable, is accessed, a second CPU must be locked to prevent access to the critical part. Therefore, the operation may be delayed because the second central processing unit must wait until the first central processing unit completes the operation in order to access the important part. In particular, in the case of a VPN device, since the two CPUs 110 share the encryption library and the pattern library 120, which are the largest parts of the system load in the VPN device, the two CPUs are executed. It is hard to expect performance improvement compared to the case where it is not used.

FIG. 2 is a diagram illustrating a difference in performance of executing kernel code in a symmetric multiprocessing system using two CPUs and a system using one CPU.

As shown in reference numeral 210, in case of using Xeon 1.8GHz, which is the latest CPU of Intel, the case of using two CPUs rather than using one CPU Rather, there was a 24% performance penalty. In addition, as shown in the reference numeral 220, in the case of using the Pentium III (Pentium III) 733MHz, the existing central processing unit of Intel (Intel) showed almost the same performance as using one central processing unit. The experimental result of FIG. 2 shows that the Xeon central processing unit uses more pipelines than the Pentium 3 central processing unit, so that the overhead of the pipeline breaking due to the lock of the kernel. Because this is bigger.

As described above, in a system using a large number of kernel codes, a symmetric multiprocessing system has a problem in that performance is lower than a system using a single central processing unit. This problem occurs because two or more CPUs share one kernel code.

Therefore, only the first CPU executes the kernel code and the second CPU does not execute the kernel code. Thus, a program executed in the first CPU executes an I / O operation. If a task consisting only of arithmetic processing is assigned to the second central processing unit and the second central processing unit performs only the assigned task, the above problem may be solved. In particular, the VPN device is composed of sequential tasks for processing a packet. Among them, encryption, decryption, and pattern matching are operations in which arithmetic processing is central, and as described above, two or more CPUs are asymmetric. If you configure your system to work with Windows, you can expect to improve the system's performance. Accordingly, in the present invention, in the system for the VPN device, the first central processing unit and the second central processing unit in the multi-processing system including two or more central processing units to operate asymmetrically to operate more efficiently. We propose a new technology for the system and method thereof.

SUMMARY OF THE INVENTION The present invention has been made to improve the prior art as described above, and has a time delay due to a lock on a critical session of a kernel in using a multiple processing system having two or more central processing units. It aims to reduce the cost and make the system more efficient.

Still another object of the present invention is to provide an improved performance at a low cost by implementing a server device including two or more central processing units without purchasing expensive equipment designed to operate asymmetrically.

It is still another object of the present invention to allow a simple operation of the asymmetric processing setting unit so that the symmetric multiple processing system can be easily converted into an asymmetric multiple processing system.

It is another object of the present invention to use a system to allocate a specialized job, so that a system can be used asymmetrically even in a case of a structure classified into various modules.

It is still another object of the present invention to use an interrupt in assigning a specialized task so that the specialized task can be processed quickly in real time.

In order to achieve the above object and solve the problems of the prior art, a multi-processing system for processing a plurality of work groups consisting of sequential tasks using a plurality of central processing unit according to an embodiment of the present invention is a plurality of A specialized job sorting unit configured to sequentially check the jobs of the work group to select specialized jobs, a general job processing unit which processes general jobs not sorted by the specialized jobs by using a first central processing unit, and the selected specialized jobs; 2 The specialized job processing unit processing by using the central processing unit The general job processing unit processes the general job of the second working group while the specialized job processing unit processes the special job in the first job group, and the special job processing unit Upon completion of the specialized task in the first working group, following the completed specialized task in the first working group; It characterized by processing the normal operation of the stand by one.

In addition, the multi-processing method for processing a plurality of work groups consisting of sequential jobs using a plurality of central processing unit according to an embodiment of the present invention screens the jobs of the plurality of work groups in order to select a specialized job And a step of processing a general job not selected by the specialized job using a first central processing unit, and a specialized job step of processing the selected specialized job using a second central processing unit. In the processing of the specialized task, the general task of the second working group is processed while the specialized task in the first working group is processed in the processing of the specialized task, and the processing of the specialized task is performed in the first working group. Upon completion of the specialized task, the general task of the next order of the completed specialized task in the first task group is sequentially processed. It is characterized by.

Hereinafter, an asymmetric multiple processing system and method thereof according to the present invention will be described in detail with reference to the accompanying drawings.

3 is a diagram illustrating an example of a VPN device implemented using an asymmetric multiprocessing system.

In the VPN device implemented using the symmetric multiprocessing system shown in FIG. 1, unlike the first and second CPUs executing the same kernel code, the asymmetric multiprocessing system shown in FIG. 3 is used. The VPN device executes kernel code only in the first CPU 310, and the second CPU executes the encryption engine and the pattern engine 330.

In other words, in the VPN device implemented using the existing symmetric multiprocessing system, all the modules including the encryption engine and the pattern engine ran on the kernel, and the two central processing units rotated and processed the tasks equally. In the VPN device implemented using the asymmetric multiprocessing system according to the present invention, the encryption engine and the pattern engine 330 operate using the second CPU 320 regardless of the kernel code. Unlike modules that use kernel codes such as LAN Driver, Firewall / VPN, and IP Stack, the encryption engine and pattern engine 330 are modules that can operate without kernel code. The second CPU 320 may not operate. Therefore, since the kernel code is used only in the module using the first central processing unit, there is no problem even if there is no lock for the critical session of the kernel code, and thus the time delay caused by the lock. This eliminates the need for improved overall system performance.

4 is a diagram illustrating an internal configuration of an asymmetric multiple processing system according to an embodiment of the present invention. As shown in the figure, an asymmetric multiprocessing system 401 according to an embodiment of the present invention includes a specialized job selection unit 410, a specialized job processing unit 420, and a general job processing unit, and in some cases asymmetric processing The setting unit 440 may further include. . Hereinafter, the function of each component will be described in detail.

The specialized task selector 410 sequentially checks tasks of the first task group and selects specialized tasks. The work group is configured by connecting a plurality of jobs sequentially, and when one job is completed, the next job of the job is configured to be executed. For example, operations such as packet header analysis, encryption and decryption, pattern matching, and packet transmission for processing packets in a VPN device are to proceed to the next operation after completing one operation in order from the previous, It can be said that it is included in the working group.

The specialized task selector 410 sequentially checks the tasks in the task group one by one to select whether each task is a specialized task or a general task instead of a specialized task. For the above-mentioned screening, predetermined criteria as to whether the task corresponds to a specialized task or a general task are determined in advance, and the specialized task is selected according to the predetermined criteria based on the type of the task to be checked. .

The specialized job sorting by the specialized job sorting unit 410 is a job that is processed using at least one central processing unit that does not execute kernel code in the asymmetric multiprocessing system 301. As an example of a criterion for selecting specialized tasks, since specialized tasks are processed using a central processing unit that does not execute kernel code, the central processing unit does not include an input / output (I / O) operation using kernel code. It may be a CPU-bound job configured only with arithmetic processing. An example of a CPU-dependent operation may include encrypting data, decrypting the encrypted data, and pattern matching to determine whether the data corresponds to a predetermined pattern. . In addition, the specialized task selector 410 may operate using a separate computing device, or may operate using a first central processing unit that executes kernel code.

The specialized job processing unit 420 processes the specialized job selected by the specialized job sorting unit 410 using the second central processing unit. The second CPU is a CPU that does not execute kernel code. Since the second CPU does not share kernel code with other CPUs, the second CPU may process only a given task without being interrupted by other tasks. For example, when the specialized operation is an encryption and decryption operation, data that is subject to encryption and decryption is received and an encryption algorithm and a decryption algorithm are performed based on the data. The encryption algorithm and the decryption algorithm generally consist of mathematical operation processing, and do not include I / O operations that require the use of kernel code.

The specialized job processing unit 420 may use a specialized job queue or interrupt in order to receive a specialized job from the specialized job selector 410. A configuration using a specialized work queue and a configuration using an interrupt will be described later with reference to FIGS. 5 to 7.

The general task processor 430 processes the general task not selected as the specialized task by the specialized task sorting unit 410 by using the first CPU. The general task means all tasks except specialized tasks, and may be a task including I / O tasks using kernel code as well as arithmetic processing using a central processing unit. Since the first CPU can execute kernel code, the first CPU can process tasks using kernel code.

In addition, the general job processor 430 may perform the finishing work of the specialized job processed by the specialized job processor 420 in some cases. The specialized job processing unit 420 processes the specialized job using the second central processing unit that does not execute the kernel code, and thus, it is difficult to process the I / O job using the kernel code. Therefore, when a task requiring kernel code, such as an I / O task, remains among the tasks processed by the specialized task processor 420, the general task processor 430 may perform the finishing task for the task. Assigning a job from the specialized job processing unit 420 to the general job processing unit 430 for finishing work may utilize a method of allocating a specialized job from the specialized job selection unit 410 to the specialized job processing unit 420. As one may use a queue or interrupt.

The asymmetric processing setting unit 440 sets asymmetric processing for the second central processing unit used by the specialized job processing unit 420. Since a symmetric multiprocessing system performs multiple CPUs sharing the kernel code, in order to operate the system asymmetrically, one or more CPUs do not execute kernel code and use only separate modules. Should be set to Therefore, the asymmetric processing setting unit 440 sets the second CPU to run asymmetrically with the first CPU without executing kernel code.

In order to set asymmetric processing for the second CPU in the asymmetric processing setting unit 440, a value of a predetermined registry memory in the second CPU may be changed. Generally, the central processing unit has a high speed storage device called a registry memory, and the registry memory includes a program in the central memory where a program to be processed is processed. It contains a PC (Program Counter) value indicating whether or not. When the PC value is changed, the central processing unit stops a previously executed program and operates a program stored in a main memory corresponding to the changed value. In this manner, the asymmetric processing setting unit 440 may set the second CPU to execute a separate module without executing the kernel code. In addition, any method may be possible as long as the central processing unit can execute a separate module without executing kernel code so that the central processing unit can operate asymmetrically with other central processing units. As described above, according to the present invention, there is a technical effect of allowing a multiple processing system designed to operate symmetrically to operate asymmetrically with a simple change of a memory value.

FIG. 5 is a diagram illustrating a configuration of using a queue in assigning a specialized task in an asymmetric multiprocessing system according to an embodiment of the present invention.

As shown, the specialized job selection unit 410 inserts the selected specialized jobs into the specialized job queue 510, and the specialized job processing unit 420 checks the specialized job queue 510 to perform the specialized job. If there is, it extracts and processes the said specialized work. A queue is a data structure for storing and managing one or more pieces of data, in the form of a list. Since only one end of the list can be enqueued and only the other end of the list can be dequeue, the queued data is first dequeued. Therefore, a queue may be suitable for delivering tasks that need to be processed sequentially.

For example, as shown in the drawing, in the case of inserting four specialized jobs selected in the specialized job sorting unit 410 in order, the queue stores five specialized jobs in order. When the specialized job processing unit 420 extracts the specialized job, the first inserted "specialized job 1" is extracted. In addition, when the specialized job sorting unit inserts the "specialized job 5" again, it is inserted after the "specialized job 4".

In the configuration of delivering a job using the queue, when the general job processing unit 430 processes the finishing work of the specialized job processed by the specialized job processing unit 420, the special job processing unit 420 may perform the specialization. The same may be applied to assigning a job to the general job processor 430. As described above, using a method of transferring a job using a queue has the advantage of dividing the system into several modules and simplifying the structural design.

FIG. 6 is a flowchart illustrating a flow of a specialized task processor 420 when a queue is used to allocate a specialized task in an asymmetric multiprocessing system according to an embodiment of the present invention.

As shown, the specialized job processing unit checks the size of the specialized job queue (S602), and when the specialized job queue does not contain any specialized jobs (size = 0), a predetermined time (for example, 1 / After stopping (S601) for 1000 seconds), the size of the specialized work queue is checked again (S602). Therefore, if there is no specialized work, the existence of the specialized work can be checked by periodically checking the specialized work queue at 1/1000 second intervals. In addition, when a specialized job is included in the specialized job queue (size> 0), the specialized job is taken out of the specialized job queue (S603) and the extracted specialized job is processed (S604).

FIG. 7 is a diagram illustrating a configuration of using an interrupt in assigning a specialized task in an asymmetric multiprocessing system according to an embodiment of the present invention.

In general, when interruption occurs in the computer system, an interrupt is a notification to the central processing unit to solve the problem. When an interrupt occurs, the central processing unit stops the work being processed and processes the work corresponding to the interrupt. Originally used to solve problems in the system, it is also widely used to send messages to the central processing unit.

As shown in FIG. 7, in the present invention, the specialized job selector 410 notifies the specialized job processor 420 of the selected specialized job through an interrupt. For example, the specialized job processing unit 420 using the second central processing unit is set to process a job stored in a predetermined memory address in response to a specific interrupt, and the specialized job selected by the specialized job selecting unit 410 is selected. After storing the at the memory address, if an interrupt is generated in the second CPU, the second CPU processes the specialized task. In the above-described method, when the general task processor 430 processes the finishing work of the specialized job processed by the specialized job processor 420, the specialized job processor 420 assigns the specialized job to the general job processor 430. The same applies to As described above, a method of allocating a specialized task using interrupts has an advantage in that tasks can be assigned and processed quickly in real time.

8 is a diagram illustrating how the performance of processing a packet varies according to a method of using the CPU in a VPN device implemented using a multi-processing system having two CPUs according to an embodiment of the present invention. Figure is shown.

Reference numeral 810 denotes the performance when operating using symmetric multiprocessing except for the encryption operation and the pattern matching operation, and reference numeral 820 includes the encryption operation and the pattern matching operation and performs symmetric multiprocessing. It shows the performance when operating using. Also, reference numeral 830 denotes the performance when the encryption operation and the pattern matching operation are separated into modules and operated using symmetric multiple processing, and reference numeral 840 denotes the encryption operation and the pattern matching operation separately. Figure 2 shows the performance of asymmetric multiprocessing for the central processing unit.

As shown in the figure, in the case of using asymmetric multiprocessing (840), it can be seen that there are many performance improvements such that the results are similar to those in the case of performing the encryption operation and the pattern matching operation. In addition, compared with the case of using symmetric multiprocessing, there was a 71% performance improvement in comparison with the case where the encryption operation and the pattern matching operation were not separated into modules (820). There was a 97% improvement in performance compared to the separate module (830). As can be seen from the experimental results, the system implemented using the asymmetric multiprocessing method can expect a significant performance improvement compared to the system using the symmetric multiprocessing.

9 is a flowchart illustrating a flow of an asymmetric multiprocessing method according to an embodiment of the present invention.

In step S901, the work of the plurality of work groups is sequentially checked to select the specialized work. The work group is configured by connecting a plurality of jobs sequentially, and when one job is completed, the next job of the job is configured to be executed. In order to select a specialized task, the criteria of whether the task corresponds to a specialized task or a general task may be determined in advance, and the specialized task may be selected according to the predetermined criteria based on the type of the task to be checked. . In addition, the specialized job selected in step S901 is a job that is processed using at least one central processing unit that does not execute kernel code. Since the specialized operation is processed using a central processing unit that does not execute kernel code, the central processing unit is configured only by arithmetic processing using a central processing unit, not including I / O (input / output) operations using kernel code. It may be a CPU bound job.

In step S902, when a predetermined job is selected as a specialized job, the specialized job is processed using a second central processing unit. The second CPU is a CPU that does not execute kernel code. Since the second CPU does not share kernel code with other CPUs, the second CPU may process only a given task without being interrupted by other tasks. For example, when the specialized operation is an encryption and decryption operation, data that is subject to encryption and decryption is received and an encryption algorithm and a decryption algorithm are performed based on the data. In order to receive the selected specialized task selected in step S901, a specialized task queue or interrupt may be used. In addition, while step S902 is in progress, step S901 may be simultaneously performed with respect to the second work group, thereby increasing efficiency of work processing.

In step S903, when the predetermined job is a general job that is not selected as a specialized job, the job is processed using the first central processing unit. The general task means all tasks except specialized tasks, and may be a task including I / O tasks using kernel code as well as arithmetic processing using a central processing unit. Since the first CPU can execute kernel code, the first CPU can process jobs using kernel code. In step S903, the processing of the specialized job in step S902 and the general job of another work group can be processed, so that the system can be efficiently operated without idle resources.

In addition, in step S903, the finishing operation of the specialized work processed in step S902 may be optionally performed. In step S902, because the specialized job is processed using the second CPU that does not execute the kernel code, it is difficult to process the I / O job using the kernel code. Therefore, when a task requiring kernel code, such as an I / O task, remains among the tasks processed in step S902, the finishing operation may be performed in step 903.

The asymmetric multiple processing method according to the present invention can be implemented in the form of program instructions that can be executed by various computer means and recorded on a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. The medium may be a transmission medium such as an optical or metal wire, a waveguide, or the like including a carrier wave for transmitting a signal specifying a program command, a data structure, or the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above-described embodiments, which can be variously modified and modified by those skilled in the art to which the present invention pertains. Modifications are possible. Accordingly, the spirit of the present invention should be understood only by the claims set forth below, and all equivalent or equivalent modifications thereof will belong to the scope of the present invention.

According to the present invention, in using a multi-processing system having two or more central processing units, it is possible to reduce the time delay caused by the lock on the critical session of the kernel so that the system can be used more efficiently. There is.

In addition, according to the present invention, by implementing a server device including two or more central processing units without purchasing expensive equipment designed to operate asymmetrically, it is possible to provide improved performance at a low cost.

In addition, according to the present invention there is an effect that the operation of the asymmetric processing setting unit can be easily converted to a symmetric multiple processing system.

In addition, according to the present invention, by using a queue in assigning a specialized task, there is an effect that the system can be used asymmetrically even in a structure classified into various modules.

In addition, according to the present invention, by using an interrupt in assigning a special task, the special task can be processed quickly in real time.

Claims (8)

In the multiple processing system for processing a plurality of work groups consisting of a plurality of jobs using a plurality of central processing unit, A specialized job selection unit configured to sequentially check the jobs of the plurality of work groups and to select a specialized job; A general task processor configured to process a general task not selected by the specialized task using a first central processing unit; A specialized job processing unit configured to process the selected specialized job using a second central processing unit; And An asymmetric processing setting unit for setting asymmetric processing for the second central processing unit Including, The specialized job selection unit selects the specialized job according to the setting of the asymmetric processing, The general job processing unit The general task processing unit processes the general task of the second task group while processing the specialized task in the first task group, When the specialized task processing unit completes the specialized task in the first task group, sequentially processing the general task of the next order of the completed specialized task in the first task group; Multiple processing system, characterized in that. delete The method of claim 1, And wherein said specialized job is a CPU Bound Job. The method of claim 1, The specialized task includes at least one of an encryption task, a decryption task, and a pattern matching task. The method of claim 1, Specialized job queue for storing the selected specialized jobs in order More, The specialized job selection unit inserts and stores the selected specialized job into the specialized job queue, The specialized job processing unit may process the stored specialized job when there is a stored specialized job by periodically referring to the specialized job queue. The method of claim 1, The specialized task selector assigns the selected specialized task to the specialized task processor using an interrupt, The specialized task processor is configured to process the selected specialized task in response to the interrupt. In the multiple processing method for processing a plurality of work groups consisting of a plurality of jobs using a plurality of central processing unit, Setting up asymmetric processing for the second central processing unit; Selecting the specialized jobs by sequentially checking the jobs of the plurality of work groups according to the setting of the asymmetric processing; Processing a general job not selected as the specialized job by using a first CPU; And Specialized work step of processing the selected specialized job using a second central processing unit Including, In the processing of the general task, the general task of the second task group is processed while processing the specialized task in the first task group in the processing of the specialized task Processing a special task in the first task group in the processing of the special task, and sequentially processing a general task of a next sequence of the completed special task in the first task group Multiple processing method characterized in that. A computer-readable recording medium having recorded thereon a program for executing the method of claim 7.

Images