JPH05204667A - Task execution control device in computer system - Google Patents

Abstract

PURPOSE:To reduce the production cost of the whole computer system by starting a task at the time of a load rate drop for a CPU when task execution requests are concentrated. CONSTITUTION:This task execution control device is provided with a load status monitoring means 4 for monitoring the current task execution status of the CPU, a load rate calculating means 2 for calculating a load rate indicated by the task execution status time rate of the CPU within a past fixed period counted from the present, a start enable load rate memory 6 for storing a start enable load rate indicating the start allowing drop rate of the load rate in each task, a start queuing task memory 7 for temporarily storing a generated task execution request, and a task judging means 3 for starting a task concerned when the start enable load rates of respective tasks stored in the memory 7 are higher than the load rate during the period of no execution of a task in the CPU. When an execution request for one task is generated, the task is executed only when the current load rate of the CPU is lower than the start enable load rate of the task concerned.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention comprises a plurality of tasks,
Involved in a computer system that executes a specified task by a CPU (central processing unit) according to a task execution request generated at random, and especially when each task execution request is concentrated in a short period of time The present invention relates to a task execution control device of a computer system that controls a computer.

[0002]

2. Description of the Related Art In a computer system used for controlling various manufacturing facilities and monitoring facilities for various plants, it is necessary to return a response to an input from a controlled object such as a plant within a fixed time.

However, a large number of task execution requests may occur only when an event occurs. In this way, priorities are assigned to the respective tasks in advance, in case a large number of task execution requests are generated within a short period of time. Then, the scheduling is performed so that the CPU having a high priority is preferentially assigned to the CPU.

However, looking carefully inside the situation where the load is high, among the many tasks requested to be executed by the CPU in the computer system, the tasks that need to be executed at the same time as the execution request is issued. Then, there is a task that does not hinder the work even if it is executed with a slight delay from the time when the execution request is generated.

Therefore, when a large number of task execution requests are generated within a short period of time, as a general rule, the input execution request of each task is once registered in the activation queue and the tasks are sequentially executed from the first task. However, when an execution request for a task with a high priority occurs, this execution request is interrupted not at the final position of the activation queue but at a position near the head. Therefore, a task that needs to be executed immediately is activated in a relatively short time even if a large number of task execution requests have already occurred.

[0006]

However, when an execution request for a task with a high priority occurs, the task execution control device that preferentially allocates a CPU to this task has the following problems to be improved. ..

That is, when an execution request for a task with a low priority occurs by chance, if a request for execution with a task with a high priority does not occur, the task with a low priority is executed immediately. In many computer systems for task scheduling, the minimum CPU unit time is given to the task once the execution is started. In addition, even if a task with low priority starts exclusive execution due to use of shared memory or a file, the task with high priority does not move and enters a waiting state.

As a specific example, in development work such as compiling a source program, it is desirable to set the processing so that it is refrained from processing while the CPU load is high. Also, if there is a function that outputs a form once a day and it happens that it is necessary to execute a task during the peak load period, this form will be output after the CPU load has decreased to some extent. It is desirable to perform the output task.

If a large-capacity CPU that can execute tasks at high speed is adopted, the probability of waiting for the execution request of the task with the high priority during the execution of the task with the low priority during the high load period is as described above. Get smaller. But,
Since the difference between the high load state and the low load state of the CPU becomes large, the functions of the CPU are not effectively used, which is very uneconomical.

The present invention has been made in view of the above circumstances, and the task execution is determined by determining the execution timing of the task with the load rate indicating the task execution time ratio of the CPU within the past certain period. When requests are concentrated, a task with a low priority is activated after a while even if the CPU is currently idle, and the load on the CPU can be leveled.
It is an object of the present invention to provide a task execution control device for a computer system that can prevent pending execution of a task with a high priority as much as possible.

[0011]

In order to solve the above-mentioned problems, the present invention comprises a plurality of tasks, and executes a task requested to be executed in response to a task execution request randomly generated.
When a plurality of task execution requests overlap with each other in the PU, the task execution control device of the computer system that schedules the execution order of these tasks,

Load status monitoring means for monitoring the current task execution status in the CPU, load ratio calculation means for calculating the load ratio indicated by the time ratio of the task execution status of the CPU within a certain past period from the present, and each task The allowable start load rate memory that stores the allowable start load rate that indicates how much the start rate is permitted when the load rate decreases, the start waiting task memory that temporarily stores the generated task execution request, and the CPU that executes the task. The task activation determining unit is configured to activate the corresponding task when the permissible activation rate of each task in the activation waiting task memory is higher than the load rate during a non-active period.

[0013]

In the task execution control device of the computer system configured as described above, the load factor indicating the time ratio of the task execution state of the CPU within the past certain period from the present is constantly calculated. That is, when the CPU is continuously executing each task, the load factor is 100%, and when each task is executed at regular time intervals,
The load factor is lower than 100%. That is, when the task processing ends, the load factor decreases with the passage of time.

In addition, each task executed in this computer system has a start permission load factor which is set in advance in the start permission load factor memory, which indicates how much the start factor is permitted when the load factor decreases according to its priority. Then, when an execution request for one task occurs, the corresponding task is executed only when the current CPU load rate is lower than the activation permission load rate of the corresponding task.

Therefore, when an execution request for a task with a low priority is input during a period when the load factor is high, the task is activated until the load factor decreases, even if the CPU is currently idle. Not done. That is, C
The load on the PU is distributed. When an execution request for a task with a high priority occurs during this period, the task is executed before the task with a low priority.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a task execution control device of a computer system according to an embodiment.

The timer 1 sends a time interrupt signal a to the load factor calculating means 2 each time the first period ΔT _{1 has} elapsed, and to the task activation determining means 3 each time the second period ΔT _{2 has} elapsed. And outputs the time interrupt signal b.

The load factor calculation means 2 is connected to a load condition monitoring means 4 and a load factor memory 5. Further, the task activation determination means 3 is connected to the activation allowable load factor memory 6, the activation waiting task memory 7, and the task memory 8. The load factor calculating means 2, the task activation determining means 3, and the load status monitoring means 4 described above are each configured by a software method using a control program.

The task memory 8 is composed of, for example, a magnetic disk storage device, and the task memory 8 stores a plurality of tasks 8a consisting of application programs for the computer system to actually execute various tasks. ing.

The load status monitoring means 4 monitors whether or not a CPU (not shown) of the computer system is currently executing a task. Further, in the load factor calculation means 2, as shown in FIG. 2, a load state memory 2a having area numbers A from 1 to M is formed. In each area, the load status of the CPU read from the load status monitoring means 4 is stored as bit data R every time each ΔT ₁ time elapses. It is set to [1] when the CPU is executing the task, and is set to [0] when the CPU is not executing the task, that is, when the CPU is empty. When the timer 1 inputs the time interrupt signal a for each ΔT ₁ , the load factor calculation means 2 calculates the current load factor L _R of the CPU according to the flowchart shown in FIG.

In FIG. 3, when the interrupt processing is started,
The area number A of the load state memory 2a is incremented by 1 (A
= A + 1). When the increased area number A exceeds the maximum number M of areas, the area number 1 at the beginning is returned (A = 1). Next, the load status of the CPU detected by the load status monitoring means 4 is checked. When the task is being executed, the bit data R _A of the area designated by the area number A is set to [1]. Also, C
When the PU is empty, the bid data R _A of the corresponding area is set to [0].

When the above-mentioned update processing of the bid data R _A for the load state memory 2a is completed, the load state memory 2
calculating a of the M-bit data R ₁ to R number of bits set to [1] of _M S (number of areas). Then, the load factor L _R is calculated by dividing the bit number S by the region number M. L _R = (S / M) × 100 (%)

Therefore, the calculated load factor L _R is (Δ
T ₁ × M) time ago to the current time (ΔT ₁
XM) shows the time ratio (%) of the task execution state of the CPP. Then, the calculated load factor L _R is written in the load factor memory 5, and the current time interrupt process is terminated. Therefore, the load factor L _R in the load factor memory 5 is updated to the latest load factor L _R for each ΔT ₁ .

In the start-up allowable load factor memory 6, FIG.
As shown in FIG. 5, the activation permission load factor indicating for each task 8a, which indicates how much the activation load is permitted when the load factor L _{R of the} CPU of the task decreases, is stored. That is, as described above, 100% or a value close to 100% is set for the high priority task 8a, and the low priority task 8a is set.
Is set to a value lower than 100%. Further, task execution requests input from the outside are registered in the activation waiting task memory 7 in time series.

Then, when the task start judging means 3 receives the time interrupt signal b from the timer 1 every ΔT ₂ , the task start judging means 3 starts or holds each task registered in the start waiting task memory 7 according to the flow chart shown in FIG. Perform judgment processing.

When the flow chart is started, the load status of the CPU detected by the load status monitoring means 4 is checked.
If the task is being executed, a new task cannot be started, so the current time interrupt process is terminated without doing anything.

When the CPU is not executing a task,
When the task is registered in the activation waiting task memory 7, the current load factor L stored in the load factor memory 5
Read _R. Then, each activation allowable load factor of each task registered in the activation waiting task memory 7 is read from the activation allowable load factor memory 6. Then, if there is a start allowable load factor higher than the current load ratio L _R of the CPU among the read start allowable load factors, the task 8a corresponding to this start allowable load factor becomes a startable task. When there are a plurality of startable tasks, the task with the highest allowable start load factor is started. If there is only one startable task,
Launch that task. If there is no startable task, the flow chart ends without doing anything in this time interrupt process. The features of the task execution control device of the computer system configured as above will be described in comparison with the operation of the conventional task execution control device.

FIG. 5A is a time chart showing the activation timing of each task in the conventional task execution control device, and FIG. 5B is the activation timing of each task in the task execution control device of the embodiment under the same conditions. 2 is a time chart showing.

Each of the tasks A, B and X has the same high priority, and the task C has a lower priority than the tasks A, B and X. Therefore, each task A, B, C, X
Has a starting allowable load factor of 100, 100, 70, 100 shown in FIG. And each task A, B, C, X
It is assumed that execution requests are input at times t _A , t _B , t _C , and t _X , respectively.

In the conventional apparatus of FIG. 5 (a), since the input time t _X of the execution request of the task X having a high priority is later than the end time of the task B having a high priority, the task C having a low priority is executed. During the period. Therefore, task X
Is executed after task C consumes a predetermined time. If exclusive processing is started during the execution of task C, a case where the vehicle cannot run even if the order of task X is changed occurs. Therefore, the waiting time T _X of the task X becomes a considerably long time.

On the other hand, in the apparatus of the embodiment shown in FIG. 5A, the CPU load condition is indicated by [1] or [0] for each first cycle ΔT ₁ . Then, in order to simplify the explanation, the number M of areas of the load state memory 2a shown in FIG.
When set to (M = 5), C for each elapse of the first period ΔT ₁
The load factor of the PU is calculated.

Therefore, the load factor of the CPU is 100% at the time t _C when the execution request for the task C having the lower priority is input. Further, the load factor of the CPU at the end time of the task B remains 100%. On the other hand, the activation allowable load factor of task C is 70%. Therefore, even if the task B ends and the CPU becomes empty, the task C is not activated.

[0033] Then, when at time t _X is task demand X is input, the load factor of the CPU at the time t _X L
_R is still 100%, and the activation allowable load factor of task X is 100%. Therefore, this task X is activated with almost no waiting time.

When the task X is completed and the CPU does not execute the task, the CPU load factor L _R is 80%.
Task C for the first time at time t _G after the change from 60% to 60%
Is started.

As described above, it is determined whether the task is to be executed immediately or to be suspended for a while, based on the magnitude relationship between the load factor L _R of the CPU at the present time and the activation allowable load factor of each task having an execution request. There is. Therefore, when a large number of task execution requests occur in a short period of time, the CPU load factor L _R remains high. Therefore, even if the execution request of the task with the high priority is finished and the CPU is in the idle state, the load request L of the CPU is input for the certain period after the execution request of the task with the low priority is input. Not executed unless _R drops. Therefore, the load on the CPU is significantly distributed. Therefore, the processing capacity of the CPU can be effectively used, and it is not necessary to use the CPU having high processing capacity, so that the facility cost of the entire computer system can be reduced.

Further, when the CPU is busy, even if the CPU has a slight free time, the free time is in a standby state in preparation for the input of an execution request of a task having a high priority. As a result, it is possible to significantly reduce the probability that an execution request of a task with a high priority is kept waiting until the task with a low priority is completed, as compared with the conventional device. Therefore, the reliability of the operation of the entire computer system in which this task execution control device is incorporated can be improved.

[0037]

As described above, according to the task execution control device of the present invention, the C within the certain period in the past from the present
When the task execution timing is determined by the load factor indicating the task execution time ratio of the PU, and the task execution requests are concentrated, the task with the lower priority order is C at the present time.
Even if the PU is empty, it does not start immediately and C
The PU is activated when the load factor decreases. Therefore, the load on the CPU can be leveled and it is not necessary to use a CPU with high processing capacity, so that the manufacturing cost of the entire computer system can be reduced. Further, it is possible to prevent the execution request of the task having the high priority from being suspended as much as possible, and it is possible to improve the reliability of the operation of the entire system.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a task execution control device of a computer system according to an embodiment of the present invention,

FIG. 2 is a diagram showing stored contents of each memory in the apparatus of the embodiment,

FIG. 3 is a flow chart showing a load factor calculation procedure of the apparatus of the embodiment.

FIG. 4 is a flow chart showing a task activation determination procedure of the device of the embodiment.

FIG. 5 is a time chart for explaining the effect of the apparatus of the embodiment in comparison with the conventional apparatus.

[Explanation of symbols]

1 ... timer, 2 ... load factor calculation means, 3 ... task activation determination means, 4 ... load status monitoring means, 5 ... load factor memory, 6 ...
Allowable activation load rate memory, 7 ... Task memory waiting for activation, 8 ...
Task memory.

Claims (1)

[Claims] 1. A plurality of tasks are provided, and a task whose execution is requested is executed by a CPU in response to a task execution request generated at random, and when a plurality of task execution requests overlap each other, these tasks are executed. In a task execution control device of a computer system for scheduling the execution order, a load status monitoring means for monitoring the current task execution status of the CPU, and a load indicated by the time ratio of the task execution status of the CPU within a certain past period from the present. A load factor calculating means for calculating a rate; a start-permissible load factor memory for storing a start-permissible load factor indicating whether to permit start-up when the load factor decreases for each task; and the generated task execution request. The activation wait task memory that is temporarily stored and the activation wait task memory during the period when the CPU is not executing the task. Li each task start allowable load factor is the task execution control unit of a computer system that includes a task activation determination means for activating the corresponding task is higher than the load rate.