JPH01180637A - Estimating system for instruction part - Google Patents

Abstract

PURPOSE:To improve the executing efficiency of an object program by adding position information in accordance with the position of data when a data part is allocated on a desired device type forming an object in approximation of data and estimating an instruction part based on the position information. CONSTITUTION:A front end part 2 is started by a source program 1 and then produces an intermediate language 4. In this case, the position information is added to the operand of the data on the language 4 via a relative position deciding means 3. When a back end part 5 is started, the language 4 is supplied. An instruction part estimating means 6 is started based on the position information added to the language 4. Thus the part 6 estimates an optimum instruction part and at the same time produces an object program 7. Therefore, the estimating accuracy is improved for the instruction part and the waste of a branch system object is reduced. Then the scale of the program 7 is reduced and the program executing efficiency is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a method for predicting an instruction part, and more particularly to a method for optimizing a branch system in the target program generation process of a compiler.

(Prior art) Conventionally, when generating an object code when predicting an instruction part, the data access object to be generated in the next pulse is not known, so the number of instructions that may be generated by the object code is not known. had to predict the object with the maximum value of .

(Problems to be Solved by the Invention) In the conventional instruction part prediction method described above, since the instruction part is predicted to be large, unnecessary objects are often generated when branch-related instructions are generated. This has the disadvantage that the target program becomes large and the execution efficiency of the program decreases.

An object of the present invention is to add position information according to the position of data when allocating a data part on a target model that constitutes an object in data access, and to predict a command part based on the added position information. It is an object of the present invention to provide an instruction part prediction method that eliminates the above-mentioned drawbacks and is configured so as not to reduce efficiency when executing a target program.

(Means for Solving the Problems) The instruction part prediction method according to the present invention is configured to include a relative position determination means and an instruction part prediction means.

The relative position determination means is for adding position information based on the position of data when allocating a data section on a target model constituting an object in data approach.

The command part prediction means is for predicting the command part based on the position information added by the relative position determining means.

(Example) Next, the present invention will be explained with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of an instruction part prediction method according to the present invention. In FIG. 1, 1 is an input source program, 2 is a front-end section, 3 is a relative position determination means, 4 is an intermediate language generated by the front-end section 2, 5 is a back-end section, and 6 is an instruction part prediction. Means 7 is an object program generated by the back end section 5. The front end section 2 is a relative position determining means 3
The back end section 5 includes an instruction section prediction means 6.

In general, a compiler that inputs a source program 1, translates it, and generates a target program 7 is roughly divided into two parts. One is a front end unit 2 that inputs the source program 1 and converts it into an intermediate language 4 in a format that is easy to translate; the other inputs the intermediate language 4,
This is a back-end unit 5 that generates a target program 7 by adding model-dependent information. In the embodiment shown in FIG. 1, the front end section 2 and the back end section 5 form a compiler.

In FIG. 1, when the front end section 2 is activated using the source program 1, the front end section 2 generates an intermediate language 4. At this time, position information is added to the operand of the data of the intermediate language 4 by the relative position determination means 3.

Next, when the backend section 5 is activated, the intermediate language 4 is input. Therefore, the instruction part prediction means 6 is activated based on the position information added to the intermediate language 4, and generates the target program 7 while predicting the optimum instruction part.

The above is the outline of FIG.

2 is a flowchart showing the operation of the relative position determining means 3 of FIG. 1, and FIG. 3 is a flowchart showing the operation of the relative position determining means 3 of FIG.
3 is a flowchart showing the operation of FIG.

Next, the relative position determining means 3 and the command part predicting means 6 will be explained in detail with reference to FIGS. 2 and 3.

In Figure 2, when inputting source program 1 and assigning data definitions in the source program,
In the first step, it is compared whether the address of the data is larger than the displacement available on the target model. When the address of the data is larger than the displacement, position information is added to the operand of the data to generate intermediate language 4.

In FIG. 3, the intermediate language is input, and the position information in the operand of the data is fetched in the first step. In the second step, the presence or absence of the above position information is determined, and the process proceeds to the third step. In the third step, since a data access command is generated when position information exists, the command part is predicted including this command. When position information does not exist, no instruction for data access is generated, so the target program is generated by predicting the instruction part without including this instruction for data access.

The prediction of the instruction part becomes more accurate according to FIGS. 1 to 3,
An efficient target program 7 is generated.

(Effects of the Invention) As explained above, the present invention adds position information according to the position of data when allocating a data section on a target model that constitutes an object in data access, and adds position information to the added position information. By predicting the instruction part based on the instruction part, the prediction of the instruction part becomes accurate, the branch-related objects become more efficient, the runtime efficiency improves, and the generated target program becomes smaller.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of an instruction part prediction method according to the present invention. FIG. 2 is a flowchart showing a processing example of the relative position determining means of FIG. 1. FIG. 3 is a flowchart showing a processing example of the instruction part prediction means of FIG. 1...Source program 2...Front end section 3...Relative position determination means 4...Intermediate language 5...Back end section 6...Instruction part prediction means 7...Target program patent application Person NEC Co., Ltd. agent Patent attorney Inoro Jusai Figure 1

Claims (1)

[Claims] Relative position determination means for adding position information based on the position of the data when allocating the data section on the target model constituting the object in data approach; An instruction part prediction method comprising: an instruction part prediction means for predicting an instruction part.