JP2560693B2 - Program link system for data flow type computer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of use] The present invention relates to a program link method for a data flow type computer, and in particular, one or more target programs compiled separately can be linked to execute one program. Program link method

[Conventional technology]

Most of the computers in operation today operate on the principle proposed by von Neumann in 1945 and are called Neumann-type. This Neumann computer stores a program describing the calculation procedure of the computer in the memory circuit of the computer,
Since the programs are sequentially executed in the order of storage addresses, they are also called sequential processing type computers. In this Neumann computer, a single program execution control unit controls the execution of processing, and when the execution start address of the program stored in the memory circuit is given in advance, the program execution control unit writes to the execution start address. Reads and decodes the specified instruction, reads the data from the memory circuit, writes the data to the memory circuit, reads the operand, and changes the instruction execution address according to the content of the instruction. And the execution of the instruction written immediately after on the memory circuit is started.
However, when the instruction execution address is changed, the instruction at the changed execution address is executed. The program link method in the Neumann computer is such that, when a plurality of target programs that have been divided and compiled in advance are stored in a storage circuit, the storage addresses of the respective target programs do not overlap and are passed between the target programs. The storage addresses of global variables are standardized, and each target program is linked into one executable program.

On the other hand, in the data flow type computer,
1 module with processing functions such as type conversion and input / output
There are more than one, and these modules are executing processing simultaneously. The execution of this process is triggered by the data, and the process is started when all the one or more data required for the process are prepared, and the data of the process result is output.
One process ends. Then, processing is performed one after another along with the flow of data to form a series of calculation processing. Since the execution of processing is controlled by the flow of data in this way, it is called a data flow type computer.

In this data flow type computer as well, the program is stored in the memory circuit of each module, but there is no direct correspondence between the memory address and the execution order, and the address corresponding to the tag called the ID of the data flowing into the module. The program is read and executed. One ID called an arc is assigned to a series of data that performs the same processing.

[Problems to be solved by the invention]

However, the above-mentioned conventional program link method of the Neumann computer cannot be applied to the data flow computer whose operation principle is completely different from that of the Neumann computer. Moreover, the program link method in the data flow type computer does not exist yet.

Therefore, in the data flow type computer, there is a drawback that linking at the target program level is impossible.

[Means for solving problems]

The program link method of the data flow type computer of the present invention includes a main purpose program storage means, a sub purpose program storage means, a local arc ID determination means, a global arc ID determination means, an ID cumulative usage amount calculation means, and an execution. And a possible program storage means.

〔Example〕

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention.

The program link method of the data flow type computer shown in FIG. 1 is the main purpose program storage means 1, the sub purpose program storage means 2, the local arc ID determination means 3, the global arc ID determination means 4, and the cumulative ID use. Quantity calculation means 5, program combination means 6, and executable program storage means 7
It is comprised including.

The main purpose program storage means 1 is composed of a memory element or a magnetic file, and stores a main purpose program M. The sub-purpose program storage means 2 is also composed of a memory element or a magnetic file and stores two sub-purpose programs A and B. The local arc ID determining means 3 does not add an offset value to the ID of the local arc provisionally assigned by the main purpose program M, and directly uses the ID.
Of the local arc temporarily assigned by the sub-objective programs A and B, the ID cumulative value is added as an offset value to determine the ID value. Global arc
The ID determining means 4 does not add an offset value to the ID of the global arc provisionally assigned in the main purpose program M, but makes a temporary decision as the value of the ID, and then the sub-purpose program.
For the IDs of the global arcs temporarily assigned by A and B, the ID cumulative usage amount is added as an offset value to be provisionally determined as the ID value. However, in this state, since the definition side and the destination side have different IDs provisionally determined for one global arc, the IDs on the destination side are unified. That is, a global arc defined by the calling program and having a destination in the called program is given a called party ID, and is defined in the called program and the calling program has a destination in the global arc. Is the ID of the calling party. The ID cumulative usage amount calculation means 5 integrates the ID usage amounts of the main purpose program M and the sub-purpose programs A and B to obtain the ID cumulative usage amount. The program combining means 6 combines the main-purpose program M and the sub-purpose programs A and B, whose arc IDs have all been determined by the local arc ID determining means 3 and the global arc ID determining means 4, to create an executable program. . The executable program storage means 7 is composed of a memory element or a magnetic file and stores an executable program.

FIG. 2 is a program diagram showing a part of the main purpose program and the sub purpose program used in the embodiment shown in FIG.

Lines 1 to 12 are the main program header, and 13
Lines 17 to 17 are the headers of sub-object program A, and lines 18 to 22 are the heads of sub-object program B. 1 line, 1
Lines 3 and 18 show the ID usage of each program. Lines 3 and 4 are the parts for storing the names of the sub-object programs A called from the main object program M, and the names A are written. Lines 5 to 7 show temporary allocation IDs of global arcs that straddle the main purpose program M and the sub-purpose program A. EOD is a code indicating a data delimiter. Lines 15 to 17 show temporary allocation IDs of global arcs that span the sub-purpose program A and the main-purpose program M.

FIG. 3 is a flow chart showing the data flow of the object program shown in FIG.

Between the main purpose program M and the sub-purpose program A,
There are four global marks, and between the main program M and the sub-program B, there are three global arcs.

In the above embodiment, the case where the number of target programs is two has been described, but the number of target programs may be one or three or more.

[Effects of the Invention] The program link method of the dataflow type computer of the present invention links one or more target programs of the dataflow type computer which are divided and compiled,
There is an effect that it can be made into one executable program.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a program diagram showing a part of an object program used in the embodiment shown in FIG. 1, and FIG. 3 is an object program shown in FIG. 3 is a flowchart showing the flow of FIG. 1 ... Main purpose program storage means, 2 ... Sub purpose program storage means, 3 ... Local arc ID determination means, 4 ... Global arc ID determination means, 5 ... ID cumulative usage calculation means, 6
...... Program coupling means, 7 ... Executable program storage means, M ... Main purpose program, A, B ... Sub purpose program.

Claims (1)

(57) [Claims] 1. The ID of a local arc provisionally assigned by the main purpose program in the data flow type computer is determined as the value of the ID as it is without adding an offset, and the local arc provisionally assigned by the sub-purpose program is assigned. For the ID, the local arc ID determining means for determining the ID value by adding the ID cumulative usage amount of the main object program and the sub-object program already combined up to that point as an offset value, and the defining side and the destination side are different. For a global arc that is a program, a global arc ID determination means that unifies according to a predetermined rule for the ID of either the defining side or the preceding side, and the main purpose program and the sub-purpose program combined up to that point Accumulate the usage of each ID
The cumulative ID usage calculating means for obtaining the cumulative ID usage, the local arc ID determining means and the global arc ID determining means
A program linking method for a data flow computer, comprising: a program connecting means for simply connecting an ID-determined main object program and one or more sub-object programs to create an executable program.