JPH0378834A - Data processing system - Google Patents

Abstract

PURPOSE:To maximize the executing speed in a data processing system by selecting the output of a high function instruction or a basic instruction according to the output conditions of the high function instruction. CONSTITUTION:An instruction which carries out a subroutine initial process is investigated as the conditions since the number of registers serving as the boundaries are different according to reservation or non-reservation of a local area at selection of the instruction. At non-reservation of the local area, the executing speed of a basic instruction is higher than that of a high function instruction with 3 registers to be saved and vice versa with 4 registers to be saved. At securing of the local area,the executing speed of the basic instruction is higher than the high function instruction with 5 registers to be saved and vice versa with 6 registers to be saved respectively. Therefore, the register value is set according to the conditions and the number of registers to be saved is checked. If the number of these registers is smaller than the conditional register value, an object code of the basic instruction is generated. Then an object code of the high function instruction is generated if the number of registers larger than the conditional register value. As a result, the executing speed is minimized in a data processing system.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data processing method in an automatic translation device that can efficiently execute a program written in a high-level language.

[Conventional technology]

In high-level languages, those that are frequently used and whose processing is fixed are provided in microprocessors as high-performance instructions. In conventional automatic translation devices, these high-performance instructions are used to generate instructions when generating object code from these high-level languages.

As an example of operation, when initial processing of a procedure is performed, a high-performance instruction is generated that performs tasks such as saving registers, setting a frame pointer, and securing a local area in one instruction.

[Problem to be solved by the invention]

Conventional data processing methods use high-performance instructions as described above. However, in some cases, there are many redundant parts of the instructions, resulting in a disadvantage that the execution speed is slower than if the basic instructions were used to perform an operation by omitting the redundant parts from the high-function instructions.

The present invention has been made in order to solve the above-mentioned drawbacks, and it is possible to select whether to output high-function instructions or basic instructions depending on the conditions when outputting high-function instructions, so that the instructions can be executed more easily. The purpose is to obtain fast object code.

[Means to solve the problem]

The data processing method according to the present invention includes means for determining conditions, means for outputting a high-performance command, and means for outputting a basic command when outputting a high-performance command.

[Effect]

Since the data processing method according to the present invention is provided with a conditional check 1 conditional branch, an instruction can be selected and the execution speed can always be kept at the shortest possible time.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a flowchart showing one embodiment of the present invention. When outputting an instruction, check conditions for branching and set boundary conditions. Next, the program branches based on this condition, selects whether to execute a high-function instruction or a group of basic instructions, and generates object code.

Next, FIG. 2 is a flowchart showing condition investigation and conditional branching in subroutine initial processing as one of the embodiments shown in FIG. FIG. 3 is a diagram showing a condition investigation of the flowchart of FIG. 2, where the dotted line represents the execution speed with respect to the number of registers when the local area is not secured, and the solid line represents the execution speed when the local area is secured.

In FIG. 2, when selecting an instruction to perform subroutine initial processing, the number of boundary registers differs depending on whether or not a local area is secured as shown in FIG. 3, so this is investigated as a condition. When the local area is not secured, the basic instructions are faster until the number of registers is 3 (and when the number of registers is 4, the high-function instructions are faster). Basic instructions are faster until the number of registers reaches 6, and high-function instructions are faster at execution speed when the number of registers reaches 6. Therefore, register values are set according to the condition.Next, check the number of registers to be masked, and if the number of registers to be masked is higher than the register value of the condition, the execution speed is faster. If the number of registers is small, if the number of basic instructions is large, object code of high-performance instructions is generated.

In this manner, by providing the condition check 1 conditional branch, this system can select an instruction and minimize the execution speed.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to minimize the execution speed by determining whether to divide an instruction into a high-function instruction and a corresponding basic instruction.

[Brief explanation of drawings]

FIG. 1 is a flowchart of a data processing method showing one embodiment of the present invention, FIG. 2 is a flowchart showing condition investigation and conditional branching in subroutine initial processing as one of the embodiments of FIG. 1, and FIG. FIG. 2 is a diagram showing a condition investigation related to FIG. 2;

Claims (1)

[Claims] For microprocessors with high-performance instructions, an automatic translation device that generates object code for the microprocessor from a high-level language generates high-performance instructions or a group of basic instructions configured to correspond to them, and speeds up their execution. A data processing method characterized by generating object code so that it is the fastest.