JPH0298740A - Do implied list processing system - Google Patents

Abstract

PURPOSE:To reduce the number of actual running steps and to perform an output processing at high speed by reducing the call of an RTP by changing an input/output list to a linear DO implied list. CONSTITUTION:A source WRITE(1) [1(K)*J(K), K=1,N] is executed in a source program 1. At a DO loop developing part 211 in an object program generating means 21, the part of a DO implied list is developed in a DO text, and an object program 31 to store a value of I(K)*J(K) in work(K) according to a DO variable K is generated. At an output call part 212, an object program 32 to call the RTP is generated after information with respect to the type of the work, etc., is generated. As a result, the DO implied list linear to the DO variable can be generated by converting the input/output list from [I(K)*J(K), K=1,N] to [work(K), K=1,N], and a processing can be performed only one time of call of RTP according to a FORTRAN compiler. In such a way, it is possible to reduce the number of actual running steps and to perform the output processing at high speed.

Description

[Detailed Description of the Invention] [Industrial Field of Application] The present invention applies to output sentences of FORTRA and N languages;
cD.

It concerns the processing method when a shape sequence is specified.

[Conventional technology]

In general, operating system/guastem (hereinafter abbreviated as O8) uses FORTRA, N language (published by Japanese Standards Association [[
Electronic computer programming language FORTRA, N, , l
15C-6201-1982J) and functions for processing corresponding data.

For this reason, a FORTRAN runtime input/output control mechanism (hereinafter abbreviated as RTP) exists for the purpose of bridging the functional difference between the FORTRAN grammar and the OS functions. When processing output statements, the FORTRAN controller creates information such as type, length, storage address, etc. for each output list item, and then
Create a target program that passes control to RTP. for example.

■rTE(D A, B in the case of.

A's information creation RTP call B's information creation RTP call Process as follows.

The same applies when output and DO sequence appear. VIIRIT
In the case of E(1) (A(I), I=1.N), the DO type sequence is expanded into a DO loop and processed as follows.

DOI=1. N A (I) information creation RTP call END D 0 Depending on the FORTRAN controller and layout, if the DO variable is linear (the increment value is constant) as in this example, D.

This is because the addresses where all elements are stored are determined by evaluating the DO variable when evaluating the shape sequence.

Since information such as the initial address, increment value, and number of repetitions is added, the number of RTP calls is reduced to one.

Information creation for A(1) to A(N) RTP call But even in such a con guira.

WRITE(1) (I(K) * J(K), K
= 1, N) when an expression is specified in a DO type array.

or.

V/RITE(1) (A(I(K)), K= 1
, N), if it is not linear in the DO variable,
D.

If only the DO variable is evaluated during the evaluation of the shape arrangement, the addresses where all the elements are stored cannot be determined, so the processing as before cannot be performed, and each is expanded into a DO loop as shown below.

DOK=1. N L = I(K) * J(K) L information creation RTP call END D 0 DO K=1. N L = A (I (K)) L information creation RTP call END DO [Problem to be solved by the invention] In general, as the number of RTP calls increases, the number of outputs increases due to information settings, etc. Output processing becomes slower because the number of travel steps increases.

In view of the above-mentioned drawbacks, the technical problem of the present invention is to provide a DO system that can reduce the number of actual driving steps and perform output processing at high speed.
The purpose is to provide a format processing method.

[Failure to solve the problem]

According to the present invention, in a computer system equipped with a FORTRAN language glossor, when evaluating a DO type sequence as an output sequence, the values of all elements specified in the DO type sequence cannot be determined from only the DO variables.

In the process of specifying a format sequence, first the output sequence is transferred to the work area prepared by the translation department.

By expanding to Runo, passing information about the work area created there to the output control means, and calling the output control means, the values of all elements specified in the DO type arrangement can be determined just by evaluating the DO variables. It is characterized by converting into a DO-shaped arrangement.

A form ordering processing method is obtained.

〔Example〕

Next, one embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. 1st
Referring to Figure 9, one embodiment of the present invention is shown.

It consists of a source program 1, a translation section 2, a target program 3, an RTP 4, and a main storage device 5. DO Lu during Translation Department 2! There is a target program creation means 21 consisting of an expansion section 211 and an output calling section 212. The main storage device 5 consists of arrays I, J, arrays WORK prepared as work areas by the translation section 2, and information on WORK.

Next, the operation of this embodiment will be explained with reference to FIG. ■ITE(1) (1(K) * J(K), K=
1, N).

In this source, the output sequence is an expression.

When evaluating a DO type arrangement, the values of all elements cannot be determined from the DO variable alone. No, purpose program creation means 2
The DO loop expansion unit 211 in 1 expands the DO type sequence into a DO statement and converts the DO variable KK and therefore the I (K) tree J (K
) Purpose program 3 to store the value of work(K) K
Create 1. The output calling unit 212 creates information such as the work type, and then creates a target program 32 that calls RTP.

As a result, the input/output sequence becomes (I(K)*J(K),=1).

N) to (work(K), K = 1, N)
By VC conversion, the DO variable becomes a linear DO type arrangement, and depending on the FORTRAN controller, one RT
It can be processed by calling P.

〔Effect of the invention〕

As explained above, by changing the linear or o-type arrangement, it is possible to reduce the number of RTP calls, which reduces the number of actual running steps.

Output processing can be performed at high speed. In addition, as a ripple effect, in a computer system with a vector mechanism, D
If the part expanded into an O-loop can be vectorized, it will be possible to process it faster.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention. 1... Source program, 2... Translation department, 21...
・Objective Grodarum creation means, 211... DO loop expansion unit, 212... Output calling unit, 3... Objective program, 31... DO sill-creation objective program, 3
2...RTP calling purpose program, 4...RT
P, 5... Main memory device, 51... Array engineering, 52
...Array J, 53...Array work, 54...
・Work information.

Claims (1)

[Claims] 1) Processing when a DO type sequence is specified as an output sequence in a computer system equipped with a FORTRAN language processor, in which the values of all elements specified in the DO type sequence cannot be determined from only the DO variables when evaluating the DO type sequence. In,
First, it is developed into a DO loop that transfers the output sequence to the work area prepared by the translation department, and the DO variable is evaluated by passing the information about the work area created there to the output control means and calling the output control means. A DO-type array processing method characterized by converting the DO-type array into a DO-type array in which the values of all elements specified in the DO-type array can be determined by using the following methods.