JPH0789322B2 - Program start method in computer system - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a program start-up method in which a central control unit in a computer system sequentially starts programs that are set to be executed periodically.

(Prior Art) When a program is cyclically started in a computer system, the program in the storage device 20 is divided into an execution management program M and cyclic start programs P _{1 to} P _n as shown in FIG. The central control unit 10 periodically interrupts the execution management program M with a clock pulse, and the execution management program M counts the number of interrupts, creates an appropriate cycle, and starts the program in this cycle. Are started in sequence.

Conventionally, the activation method has a data memory as shown in FIG. 5, and the execution management program M updates the activation timer 21 each time it is interrupted by a clock pulse, and activates it with the updated value m. Timetable 22 memory m _{1 to} m _n
Determine which memory of this is to be processed in this cycle. memory
Bit b ₁ ~b _n of m ₁ ~m _n is equivalent to 1 bit is one program, if the "1" to this bit is stored in the current period corresponding program indicates that it should be started There is.
If the bit b ₁ is "1", the relative value n of the bit position causes the program P ₁ start address stored in the memory n ₁ of the start jump table 23
Start P ₁ . If bit b ₁ ~b _n are all "1" is stored, starts all programs current period, and if all "0" is stored, the current period is started anything Program Will not do.

The start cycle of the program is determined by the interrupt cycle of the clock pulse and the interval in which "1" is stored in the corresponding bit of the start time table 22. If the clock pulse is a 10 ms cycle, the memory m _{1 to} m _n If all "1" s are stored in the bit b _{1, the} program P ₁ is started in a cycle of 10ms, and "1" is stored in each bit b ₂ of the memories m _{1 to} m _n. If so, program P ₂ is started every 20 ms. When the value m of the startup timer 21 is updated to the memory m _n of the startup time table 22, the cyclic startup program can be continuously started by returning the value of the startup timer 21 to the initial value.

Although various startup methods have been used in the past, they are basically as described above.

In many cases, a large number of terminals are connected to a computer system, and it is often required to control randomly generated control processing in real time between these terminals and the central controller. The biggest problem in this computer system is the processing capacity of the central control unit 10. This is basically determined by the average time and the amount of programming for which the central control unit 10 processes one instruction of the program. As shown in FIG. 3, as the load of the central controller 10, if the time during which the cyclic start program is executed is a fixed load and the time during which input / output control processing with the terminal is executed is a variable load, The processing capacity of a computer system is determined by how much variable load can be processed.

(Problems to be solved by the invention) Therefore, if the start cycle of the cyclic start program is fixed as in the conventional case, the terminal processing capability, which is the variable load of the computer system, is increased due to the size of the fixed load. However, there is a drawback in that the processing time of the variable load must be secured for a certain amount or more, and thus the time amount of the periodic start program that is the fixed load is limited. .

(Means and Actions for Solving Problems) In the program starting method of the present invention, the execution management program is cyclically started by clock pulses in the central controller, and a plurality of periodic programs are executed by this execution management program. In the activated computer system, the execution management program is provided with means for varying the activation period of the periodic program in accordance with the variation of the load applied to the central controller.

(Example) Next, the Example of this invention is described with reference to drawings.

The memory value a of the load factor 24 in FIG. 2 is the load factor applied to the central control unit 10 in FIG. 4, and is calculated by a periodic start program dedicated to the load factor calculation at fixed time intervals and stored each time. Will be updated. This load factor is calculated by adding the fixed load and the variable load shown in FIG. 3 to the unloaded empty portion at regular intervals. Since it is a matter, I will omit the explanation.

In FIG. 2, startup timetables 22-1 to 22-n are provided, and in the execution management program M, the value a of the load factor 24 is set.
Selects the startup timetable to be processed in this cycle. Once the start timetable is determined, the procedure until the program is started is the same as in the conventional method, one memory is read from the memories m _{1 to} m _n of the start timetable by the value m of the start timer 21, and "1" is stored. the relative values of n bits b ₁ ~b _n which are, to start the appropriate cyclic program reads the corresponding program start address of the boot jump table 23.

This control procedure is shown in FIG. When the clock interrupt arrives, the execution management program M is executed from the head address. Here, the value m of the start timer 21 shown in FIG. 2 is added according to the flow 1. When this value m exceeds the length of the startup timetable, it is set to the initial value. Next, in a flow 2, the value a of the load factor 24 shown in FIG. 2 is read, and from this value a, the start time table to be processed in this cycle is determined from the start time tables 22-1 to 22-n, and the start thereof is started. Calculate the start address of the timetable. Next to the head address of the start time table in flow 3 adds the value m of the activation timer 21 reads the bit b ₁ ~b _n to be processed a current cycle. Next, in flow 4, the relative value n of the bit position where "1" is first stored is calculated, and the corresponding program start address of the start jump table 23 is read according to the value of the relative value n. Then, except for the bit processed this time, the remaining bits are saved in the save memory. This is because the cyclic starting program sequentially processes all the bits after the execution is completed. In Flow 5, the program is executed according to the start program start address read in Flow 4.
You will jump to _{one of} P _{1 to} P _n .

In the started periodic program, flow 6 or flow 8
Then, the processing peculiar to the program is performed, and when the processing is completed, the processing jumps to the flow 4 of the execution management program M by the flow 7 or the flow 9 and returns. Flow 4 came back
Then, the save memory in which the remaining bits are stored is read out and the program to be started next is determined. However, when all the start-ups have been completed, the execution management program M has completed the current cycle execution and the next clock pulse has been completed. Waiting for an interrupt.

However, it goes without saying that the time from the completion of the execution management program to the next clock pulse interrupt is not idle time, and the processing program for terminal control or the like is executed.

By doing so, the start cycle of the cyclic start program can be lengthened (it can be shortened) or stopped according to the change in load. Which program has a different start cycle may be determined according to the priority of the program from the viewpoint of the entire system.

If you just want to stop the start-up, there is also a method to prepare one start-stop smack bit corresponding to bits b _{1 to} b _n without installing multiple start-up timetables. There are various methods such as reading the memory of the load factor with the routine at the beginning and comparing it with the data fixed for each program to determine whether to continue the process. The present invention can be implemented by putting it in an appropriate program routine.

In addition to the load shown in the present embodiment, a value that is approximately proportional to the load factor applied to the central control unit 10, such as the number of processing requests generated from the terminal or the number of terminals waiting for processing, is converted into a load. However, the object of the present invention can be achieved.

As can be understood from the above embodiment, the processing idle time of the central control unit 10 of the computer system is effectively used by the cyclic start program having a low priority, and the load is increased during the period when the fluctuation load of the terminal control processing or the like increases. Can be linked to increase the start cycle from the one with a lower priority of the cyclic start program that is a fixed load, or can stop it to secure the time for terminal control, so depending on the combination, It is also possible to eliminate the processing idle time of the control device 10.

(Effects of the Invention) As described above, by carrying out the present invention, the processing idle time of the central control unit of a computer system can be effectively used by a cyclic start program having a low priority, and a variable load such as terminal control can be utilized. Since the processing time can be secured in tandem with the increase in the number of devices, the processing capacity of the central controller can be dramatically improved.

[Brief description of drawings]

FIG. 1 is a diagram showing a processing flow of an embodiment of the present invention, FIG. 2 is a diagram illustrating an embodiment of the present invention, FIG. 3 is a diagram for explaining the present invention, and FIG. FIG. 5 is a block diagram of an applied computer system, and FIG. 5 is a diagram for explaining a conventional embodiment. 10 ...... central controller, 20 ...... storage device, M ...... execution management program, P ₁ to P _n ...... cyclic program, 21 ...... activation timer, 22-1 to 22-n ...... launch timetable, twenty three……
Start jump table, 24 …… Load factor.

Claims (1)

[Claims] 1. A computer system in which an execution management program is periodically activated by a clock pulse in a central control unit, and a plurality of periodic programs are activated by the execution management program, wherein the execution management program is the central control. A program start-up method comprising means for changing the start-up cycle of the periodic program according to a change in load applied to the apparatus.