JPS60112151A - High speed logging method of data - Google Patents

Abstract

PURPOSE:To obtain a data logging method which can follow data generated due to interruption with a shorter cycle by switching a buffer at a point of time when data more than reference quantity prescribed beforehand are accumulated in a buffer. CONSTITUTION:Data buffer switching is executed only at the time of little load such as a head or the middle of a telegram. Accordingly a reference value at buffer switching is determined so that data and a status will not overflow in a buffer area even when the data, which are converted into a code at the time of processing of telegram termination, and a status are transfered to a buffer unconditionally. It is assumed that a reference value equals S-nXm, where S is a buffer size, (n) is the number of bits of one telegram and (m) is the number of bits when one-bit data are code-converted. By comparing the reference value at buffer switching with accumulated data, whether buffer switching is necessary or not is decided, and data logging becomes possible in an interruption cycle T1.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a high-speed data logging processing method, and in particular, data h1- fluctuates periodically, and all data generated in a fixed short period is logged. The present invention relates to a high-speed data logging processing method suitable for.

[Background of the invention]

FIG. 1 shows an example of a data logging processing system. The computer IVi data generation interrupt 2 starts the data logging processing program. This program converts the collected data into a visible code, sends an output command to a display device 3 such as a display or printer, and sends an output command to an auxiliary storage device 5 when it is necessary to save the data.
Output command 6 is issued to. There is no problem if the data cycle generated here is relatively long and the above series of processes can be executed within that cycle, but for example, in logging of data flowing over a line, the data generation cycle is generally short, so (1) multiple It is necessary to consider the disparity in processing speed between the 19 machines and the display device and auxiliary storage device, and the installation of multiple-sided data buffers to absorb noise.

FIG. 2 shows an example of a logging processing system for data flowing on a line. This example is a multi-system where functions are divided into calculation i1A and IB, and computer IA is responsible for code conversion to make the data collected by data passing interrupt 2 on the line into a visible form, and is installed in computer R machine IB. It is in charge of transferring data to the data buffer 1a and lb and filling the buffer. The d1 computer iB functions to output data in the buffer to the display device 3 or to the auxiliary storage device 5 in response to a data buffer full (la or lb) interrupt from the computer IA. I'm sharing my share.

A conventional processing program in the machine 1A is shown in FIG. Generally, depending on the transmission speed on the line, a passing interrupt period for 1-byte length data is allowed, and code conversion processing (blocks A and G) is performed within this period.

H) and data buffer management processing (blocks 13 to F).

The structure is such that it can perform both processes. The example shown in Figure 3 shows an example in which the l task performs both code conversion processing and data buffer management processing, but these processing may be performed in separate tasks. As the transmission speed increases), multitasking methods are less likely to be used.

In Figure 3, interrupt processing when the data buffer is full is performed at two locations, blocks D, E, and L, but the size of the data buffer is generally much larger than the data length, so Data buffer switching processing is never performed twice in one interrupt processing. Therefore, the load is heaviest when data buffer switching processing occurs during data and status processing at the end of a message, but there is no problem when there is a relatively large margin in the -interrupt cycle. However, problems arise when the transmission speed on the line is high.

FIG. 4 is a time chart of a conventional program. The example shown here is an example of all possible loads for -interrupt period T, and the horizontal direction in the figure shows the processing time within one interrupt period, and the upper direction is for each load. An example is shown.

(2) is the processing time in the case of data only, which consists of data code conversion time 7 and data transfer time 8 to the data buffer. (2) is the processing time when the message ends, and the status code conversion time 9 and the data transfer time 101 are added to (2). G) This is the case where buffer switching occurs during the processing of (2) and the processing time 11 is added to it, and (2) and (2) are the cases where the buffer switching occurs during the processing of a). In this example, when the message ends, processing time t2 of ■
Even if the interrupt period is larger than T, there is no problem as long as (n11tl+12T, > 11 message busy n bytes) and T, > , 2-. but,
Buffer switching processing occurs at the end of the message ■■4 Taking into consideration, at least the interrupt period T2 is T, > ,
4. Because t4>2 T2
In this case, (1) the father will be interrupted twice (within 1 hour) while processing (2), resulting in data loss.

In this way, for buffer switching processing that occurs infrequently, if the interrupt cycle is set to T + Tt, the processable transmission speed is restricted. [7] To provide a data logging processing method that allows tracking of generated data.

[Egg of invention, essential]

In order to completely limit the timing of data buffer switching processing, the present invention predefines a certain reference amount in the buffer, and when the load is heavy, buffer switching processing is performed regardless of the amount of data accumulated in the buffer. However, when the load is light and there is more data in the buffer than a predetermined reference amount, four buffer switching processes are performed.

[Embodiment of the Invention] FIG. 5 is a processing flowchart showing an embodiment of the present invention. In this embodiment, the code conversion process is completely the same as the conventional one, and the data buffer switching process k't1. i,
This is done only when the load is low (no status processing required) at the beginning or middle of a sentence.

Therefore, in order to prevent the buffer area from overflowing even if the code-converted data and status are unconditionally transferred to the buffer during message termination processing, [buffer switching reference value] == S-n x mHere, , S; Buffer size Number of bytes of logo 1 message 4 01' The number of bytes when 11 bytes of data is code converted is used as the reference value for buffer switching, and in block N, compare this value with the amount of accumulated data. By doing so, it is determined whether buffer switching is necessary or not.

With the above measures, the problematic case ``■'' in FIG. 4 is eliminated, and data logging processing is possible even with an interrupt period T, as shown in FIG. 6.

According to an embodiment of the present invention, the conventional disadvantage that processing IJ performance and transmission speed is limited due to extremely rare cases where buffer switching occurs during processing of end-of-telegram data can be overcome.

Further, by implementing the present invention, the buffer switching timing becomes slightly earlier than the conventional one, but no practical trouble occurs.

〔Effect of the invention〕

According to the present invention, when logging data whose amount of data fluctuates periodically and is generated in a fixed short period, the influence ttg' caused by buffer switching can be minimized, so that the period is shorter. It is possible to realize a high-speed data logging processing system that can also process data generated in .

[Brief explanation of the drawing]

1 and 2 are schematic configuration diagrams showing an example of a data logging processing system, FIG. 3 is a processing flowchart for conventional code conversion processing and data buffer management processing, and FIG.
The figure shows a time chart of the conventional data logging processing method.
Fifth [ii<+tj: Flowchart of one embodiment of the present invention, FIG. 6 is a processing time chart of the present invention. 2... Data generation interrupt, 7, 8... Code conversion and data transfer time of generated data, 9.10... Code conversion and data transfer time of generated status, 11 Section 4, Figure 5

Claims (1)

[Claims] 1. The data rate fluctuates synchronously, the data generated in a constant short cycle is accumulated in a buffer, and when the buffer becomes full, it is switched to another buffer. , a data logging processing method in which data in a buffer is stored in an auxiliary storage device, etc., characterized in that the buffer is switched when data on a predefined board is accumulated in the buffer. Logging processing method.