JPS62233825A - Automatic optimized application allocating system for shared picture buffer - Google Patents

Abstract

PURPOSE:To improve the memory efficiency of a buffer by using the buffer after dividing it in plural processes, allocating the buffer to each process within a latest fixed time and furthermore allocating automatically the areas in response to the storing frequency of the picture data into the buffer. CONSTITUTION:A buffer using right check processing part 6 refers to a buffer using right allocation table 4 (orders of using types and area addresses) to decide a new using request or a lasting using request. If a new request is decided, a processing part 7 decides the buffer using right and sends to operation back to the part 6 again. While it is checked whether or not a shared picture buffer 2 has an idle space if the lasting request is decided by the part 6. If the buffer 2 has an idle space, data are stored at the relevant position of the buffer 2 by the function of a data store processing part 1.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an input/output method for digitized image data, and in particular, to a method for dividing and using a buffer storing a large amount of image data in multiple processes. The present invention relates to a common image buffer automatic optimization allocation method suitable for reducing buffer capacity and high-speed input/output.

[Conventional technology]

It is known that image data has a large amount of data, and in order to output image data to an external input/output device, it is necessary to transfer the data in a format that depends on the storage format of the external storage medium. On the other hand, on the side that creates image data to be output externally, either create the data in an order that takes into account the storage format, or prepare all the buffers on the data creation side and use the data creation tlct on this buffer.
It accommodates the difference between the order and the order convenient for the storage format.

In addition, when transferring a large amount of data, data access to external input/output devices should be done using D
MA transfer is generally used. The DMA transfer time (TDMA) is divided into a portion (Ta*D) proportional to the transfer data 1-(D) and a fixed portion.

T DMA = To + Td* DGenerally, T
Since o>Ta, it is more efficient to increase the transfer data fiD.

For the above reasons, a buffer is provided on the side that creates image data to be output to the outside.

This buffer is for image data, so the buffer size is large, and it can be divided into multiple buffers to accommodate multiple processes and simultaneous parallel processing to multiple external input/output devices. We have set up some.

In addition, in order to apparently absorb the time required for data transfer, it is common to use alternate buffers.

An example of the use of a replacement buffer is described in Japanese Patent Publication No. 60-22776. Now, we will briefly explain the operation using the alternate buffer.

(1) Input from a large-capacity data source is started based on instructions from the processing device.

(2) Data is first input into the first buffer.

(3) When the first buffer becomes full, the program switches the address and inputs it to the second buffer.

(4) The data once input to the first buffer in the above (2) is transferred to the large-capacity magnetic disk device using the input time of the second buffer.

(5) When the second buffer becomes full, it is switched to the first buffer again, and this time, the contents of the second buffer are transferred and stored in the large-capacity magnetic disk device using the input time to the first buffer. Ru. Data from a large-capacity data source is delivered to a large-capacity magnetic disk device through one or more repetitions K via a replacement buffer.

Next, the buffer size of the replacement buffer will be explained using FIG. 4. With alternate buffers, the time it takes to store data in the buffer is longer than the time it takes to transfer data from the buffer to a large-capacity magnetic disk drive.

It is effective when the buffer size is relatively small, as shown in FIG. 4(a), and when the buffer size is 1 for both the first buffer and the second buffer, as shown in FIG. 4(b), By increasing the buffer size to 3, it is possible to reduce the time required for transferring a certain amount of data ((■ to ■) in FIG. 4). −
If the time to store data in the buffer is relatively longer than the time to output data from the buffer, then if
As shown in C1.

Even if the buffer size is set to 4, the amount of data remains constant (■～■
) is less effective than the one shown in FIG.

[Problem that the invention seeks to solve]

As described above, even if the size of the buffer is made larger than necessary, it will not be advantageous for data transfer, but will be disadvantageous in terms of memory usage efficiency. Therefore, it is necessary to set an appropriate buffer size, but when building a system (
To determine the buffer size for each process at the installation stage, consider factors such as the frequency of use of each process, processing request speed, and image data transfer time, and make a decision based on these demarcation points (trade-offs). This requires a lot of effort, and also requires a lot of rework if the settings are changed.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-mentioned problems and provide a shared image buffer automatic optimization usage allocation method that increases buffer memory efficiency and shortens transfer time.

[Means for solving problems]

The present invention solves the above problems and achieves the above objects by repeating a large amount of image data in a certain basic unit and transferring the data to an external input/output device via alternate buffers. The amount of data stored in the other pair of buffers is detected within the time for discharging data, and if this already fills a predetermined buffer, the amount of data stored in the other buffer is If this is not the case, a separately provided usage rights allocation table is changed in order to reduce the size of the buffer.

[Effect]

The present invention has been made based on the following points.

(1) 6 Image processing often involves repeating a large amount of data in a certain basic unit. Therefore, usage requests are made for a certain period of time.
It occurs periodically, and the amount of data per time is averaged.

(2) In order to simplify data dispensing from the buffer and to be able to perform it in one dispensing process, the buffer is
It is physically continuous.

(3) Most of the payout time from the buffer is the time corresponding to the number of payouts, and the time element corresponding to the amount of data paid out per time is small.

(4) When the buffer is in the form of a replacement buffer, there is an optimal buffer size.

For this reason, in a buffer that stores image data for transfer to an input/output device among buffers that are commonly used by multiple processes, the usage allocation for each process in the buffer is not fixed in advance and new buffers are used. Allocation is changed at the time a request occurs and at the time data has been completely discharged from the buffer.

When transferring large amounts of image data between external storage devices,
This is generally done via the main memory. in this case,
A temporary buffer is required in the main memory, and this buffer has an appropriate size. This appropriate size can be determined by knowing the amount of data stored in the buffer per unit time. By allocating an appropriate buffer size to each process, memory efficiency can be increased and processing time can be shortened.

〔Example〕

Hereinafter, one embodiment of the present invention will be described based on the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. In FIG. 1, ll't data storage processing section 12 is a common image buffer, 3 is a data payout processing section, 4 is a buffer usage right allocation table, 5 is a request reception processing section, 6 is a buffer usage right check processing section, and 7 Reference numeral 8 denotes a buffer usage right division processing section, 8 a buffer usage right allocation change processing section, and 9 a buffer usage right comprehensive processing section. The operation of such an embodiment will be explained.

First, the request reception processing section 5 branches processing depending on the type of request. Types of requests include transfer processing requests, processing completion requests, and transfer completion reports. In response to a transfer processing request, the buffer usage right check processing unit 6 checks the usage right of the buffer 7, and in response to a processing end request, the dispensing processing unit 3 performs data dispensing processing for the buffer of the usage type. In addition, in response to the transfer completion report,
Perform transfer completion processing.

The buffer usage right check processing unit 6 creates a buffer usage rights allocation table (in order of usage type layer and area address) 4.
Refer to , determine whether it is a new use request or a continued use request, and
If it is a new usage request, the processing unit 7 performs a buffer usage right division process and returns the operation to the buffer usage right check processing unit 6 again. On the other hand, when the nuclear check processing unit 6 determines that the request is for continued use, it is further determined whether or not there is space in the buffer 2, and if there is space, the common image is stored by the operation of the data storage processing unit 1. Store the data in the corresponding position of buffer 2.

If there is no space in the buffer 2 and there is no data dispensing in the corresponding usage type, a data dispensing flag is set, the dispensing processing unit 3 dispenses data from the full buffer, and the replacement buffer is switched; The check processing unit 6 performs the buffer usage right check process again.

If there is no free space in the buffer and the usage type is data discharging, the buffer full flag is set and the transfer processing request source is placed in a WAIT state.

As described above, in response to a processing end request, the core dispensing processing unit 3 performs data dispensing processing in order to discharge the data in the buffer of the usage type.

In response to the transfer completion report, we will review the buffer size and buffer location for the relevant usage type, and if there is a new usage request that is waiting for a free buffer, we will allocate a buffer to it. Perform all transmission completion processing.

As mentioned above, the times when buffer usage allocation is changed are: (1) when a new buffer usage request occurs; (11) when a new buffer usage request occurs;
) This is the point in time when the data has been completely discharged from the buffer.

For explanation purposes, as shown in Figure 2 (formerly, one buffer is used), two buffers are prepared as alternate buffers as shown in Figure 2 (1), and each buffer size is Let be U.

(1) When a new buffer usage request occurs, the buffer usage right allocation table (in order of area address) 4 is
= Search sequentially from 1 to find the minimum i that satisfies the following conditions
seek.

■Occupied use 1 road - 〇■ (final pointer) - (first voice/re) + 1 ≧ Appropriate buffer size for usage type Here, the usage type is the summer type that issued the new buffer usage request, The appropriate buffer size is for usage type K in the buffer usage right allocation table (usage type) 4.

Once this more accurate i is determined, the occupied usage type is set to K, and the final pointer is changed to (head pointer) + (appropriate buffer size for usage type) - 1>. And (i+x
) and subsequent tables are moved down one by one (i+1)
against.

■Occupied use type = 0 ■Start pointer = (last pointer for 1) + 1 ■Final pointer = previous (final pointer at i) Set to .

In addition, the buffer usage right allocation table (usage type Jil
ff) Set the start pointer and end pointer for usage type ■( in 4) to the values of the start pointer and end pointer corresponding to the first case in the buffer usage right allocation table (area address order) 4, respectively, and store. The final pointer is set to O.

The above procedure is performed for the first and second buffers of the replacement buffer, and finally, the usable buffer A, full flag, and data dispensing flag for usage type K in the balance usage right allocation table (usage type) 4 Set each to 0.

If i cannot be determined as described above, the full flag is set to 1 for the usage type K in the buffer usage right allocation table (usage type) 4, and O is cleared except for the appropriate buffer size. Then, it is placed in a WAIT state.

(2: J: The point in time when the data output from the buffer is completed.

For the sake of explanation, it is assumed that in usage type K, data dispensing has been completed in buffer j (Noni 1 or 2) among the spare buffers.

First, if the case in which the usage type K occupies the corresponding buffer in the buffer usage right allocation table 4 (area address order) is 1, then this occupancy usage type is set to O.

Next, check the aptitude factorization. If the full flag is O, the current data storage amount of the buffer (3-j),
In other words, (last storage pointer) - (start pointer) +
Set 1 to the appropriate buffer size.

If the full flag is 1, double the appropriate buffer size.

Then, for the buffer j in the buffer usage right allocation table (in order of area address) 4, a buffer of the appropriate buffer size is newly allocated to the empty area according to the usage type. If the required free area is not available, allocate the largest area that can be secured.

Also, buffer usage rights allocation table (use type)
For usage type K in 4, set full flag to 0, cross 1>
The head pointer, last pointer, and last storage pointer of buffer j are set, the number of usable buffers is set to ``,'' and the replacement buffer (3-j) node is issued. The data payout flag is left at I.

Finally, a new usage request is issued, but there is no free area, so the full flag is set, the buffer usage right allocation table (usage type) 4 is checked, and if there is, the processing for the new usage request described above is performed again. conduct.

These processing results will be explained using Example 2 shown in FIG.

According to the processing type (■) in FIG. 3, it is as follows.

- Transfer data amount per process is 10', buffer storage request frequency per second is 100 times, buffer storage data 1 per time is 1000.1 time (sofa payout time is 0.03 seconds, Assume that the appropriate buffer size is 2000.

Here, if the buffer size remains unchanged at 2000, it will take approximately 15 seconds to transfer 106 pieces of data, since the time it takes for the buffer to become full is 0.02 seconds, which is shorter than the time taken for discharging one buffer, 0.03 seconds. On the other hand, according to the method of the present invention, the other one of the alternate buffers becomes full during the first buffer dispensing.
Set it to 000. This is the optimal buffer size of 10
It takes about 10 seconds to transfer the data of 6, which can shorten the time.

In the same manner, the transfer time for one process is the same between the fixed type and the present method for the process types (1) to (4) in FIG.

〔Effect of the invention〕

As described above, according to the present invention, in a buffer that temporarily stores image data for transfer to an external input/output device, this buffer is divided and used by a plurality of processes,
Allocate buffers to each process within a certain amount of time,
Moreover, since the C area is automatically allocated according to the frequency of storing image data in the buffer, the memory efficiency of the buffer is improved and the transfer time can be shortened.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is an explanatory diagram showing an example of the configuration of a common image buffer and a buffer usage right allocation table, and FIG. Explanatory diagram 1 shown in . Figure J4 is a time chart shown to explain the data payout process from the replacement buffer. l...Data storage processing unit, 2...Common image buffer, 3...Data dispensing processing unit, 4...Right allocation table, 5...Request reception processing unit, 6...Buffer use right check processing unit, 7... buffer usage right division processing unit, 8... buffer usage right allocation change processing unit,
9... Buffer usage right comprehensive processing unit.

Claims (1)

[Claims] 1. Image processing that processes a large amount of image data by having means for transferring data to an external storage device, a buffer for temporarily storing data at the transfer source for transfer, and a means for storing data in the buffer. In this method, the buffer is a common image buffer that is used in a time-sharing manner in a plurality of processes, a usage rights allocation table is provided that stores the correspondence between a plurality of buffer usage rights and the physical positions of usable buffers, and a unit It has means for detecting the amount of data stored in the buffer per hour for each usage right, and changes the usage rights allocation table according to the detection result from the means to change the allocation of buffer usage rights. Features a shared image buffer automatic optimization usage allocation method.