JP5825347B2 - Compression device - Google Patents

Description

  The present invention relates to compression, and particularly to a lossless compression technique.

  Data lossless compression technology is also referred to as lossless compression, and is a technology for reducing the amount of data by compressing a data sequence in a fully recoverable state. The data lossless compression technique is frequently used for the purpose of compressing a data series that needs to be restored after decompression (restoration) (for example, Patent Document 1-2).

  The algorithm of the data lossless compression technique is disclosed in, for example, Non-Patent Documents 1-4. As disclosed in Non-Patent Documents 1 to 4, in lossless compression, algorithms that search for repetitive patterns in a data series, such as Huffman coding, run-length coding, arithmetic code, LZ code, and variations thereof, are mainstream. . Non-Patent Document 5 discloses a high-speed mounting method for hardware devices in LZ.

  In addition, data stored in the main memory of a computer is known to perform hibernation, which is a process of saving data from the main memory to a non-volatile memory that is an auxiliary memory when the apparatus power supply of the computer is temporarily turned off. It has been. Patent Documents 3-5 disclose a technique for reducing the time required for saving and restoring by applying reversible compression or irreversible compression during hibernation. Further, lossless compression can be applied to programs and the like, and irreversible compression can be applied to images and the like.

  Moreover, the technique described in patent documents 3-5 is applicable also to swapping. Swapping is a process of temporarily saving a low-use process from the main memory to the auxiliary storage device and releasing the main memory area during the operation of the computer.

International Publication No. 2000/021199 Japanese Patent Laid-Open No. 10-173541 Japanese Patent Laid-Open No. 2004-038545 Japanese Patent Laid-Open No. 2004-038546 International Publication No. 2000/054133

Jacob Ziv and Abraham Lempel; "A Universal Algorithm for Sequential Data Compression", IEEE Transactions on Information Theory, 23 (3), pp. 337-343, May 1977. Rissanen, Jorma (May 1976). "Generalized Kraft Inequality and Arithmetic Coding". IBM Journal of Research and Development 20 (3): 198-203. Huffman's original article: D.A.Huffman, "A Method for the Construction of Minimum-Redundancy Codes", Proceedings of the I.R.E., September 1952, pp 1098-1101. Vitter, Jeffrey Scott, "Design and Analysis of Dynamic Huffman Coding", IEEE Foundations of Computer Science, 26th Annual Symposium on Digital Object Identifier: pp. 293-302, 1985. Sumio Morioka, Prof. Sato: "100Mbyte / s high-speed data compression engine", 9th FPGA / PLD Design Conference, Jan 2002.

  However, the above-described lossless compression method has a problem that it is difficult to maintain the compression ratio without reducing the processing speed when performing lossless compression of data. This is because the lossless compression technique has a trade-off relationship between the processing speed related to compression and the compression rate. In particular, such a problem becomes conspicuous when the data stored in the main memory of the computer is subjected to lossless compression. Therefore, it will be specifically described below.

  In order to speed up the above-described hibernation and swapping, it is effective to compress the data stored in the main memory of the computer. In particular, in recent years, the capacity of main memory has increased significantly. For this reason, the save and restore processing from the volatile storage area that is the main memory to the nonvolatile storage area that is the auxiliary memory has become longer than before. Therefore, there is an increasing need to compress data stored in the main memory.

  When compressing the contents of the computer main memory, for example, the general-purpose lossless compression method disclosed in Non-Patent Document 1-5 is used, or according to the data contents as disclosed in Patent Document 3-4, for example. It is conceivable to switch between a plurality of compression methods. By the way, the contents of the data stored in the main memory of the computer are not specified in the image or the like. Therefore, when the technique disclosed in Patent Literature 3-4 is applied, it is necessary to determine whether or not the target data can be compressed, and the load of the determination process is high. For this reason, the method of switching a plurality of compression methods according to the data contents is complicated in processing and requires a long processing time. In addition, this method is not suitable for general use because it is difficult to determine data contents.

  Therefore, it is desirable to use a general-purpose lossless compression method to compress data stored in the main memory. However, an algorithm with a large amount of calculation such as an arithmetic code, which is a single general-purpose lossless compression method, takes a long time for compression and decompression, so that hibernation and swapping cannot be accelerated.

  On the other hand, as for LZ among general-purpose lossless compression methods, as disclosed in Non-Patent Document 5, for example, a high-speed mounting method for hardware devices is known. For this reason, it is considered appropriate to use LZ in order to compress data stored in the main memory.

  In LZ compression, pattern matching is performed. Specifically, paying attention to a certain data string, check whether it has appeared before, and if it has already appeared, compress it by replacing it with some code indicating that data string. Do. Of course, the code for replacing the data string is shorter than the data string.

  Here, the data string to be replaced is registered in the dictionary. It is possible to create a dictionary in advance if such data strings are already present in the data, but what kind of data strings exist like image data If it cannot be predicted, it is necessary to create a dictionary simultaneously with compression. In this case, the internal state (such as the contents of the dictionary) of the compression apparatus at a certain point reflects the past input sequence from the beginning.

  Moreover, not only LZ but lossless compression has the property that the size of the data after compression is not known until it is actually compressed. For this reason, the address in which the area for storing the compressed data corresponds to which address in the original main memory varies depending on what data exists in the main memory, and is predicted before the compression. I can't.

  A lossless compression codeword has a variable length in bits. The compressed data is a sequence of codewords, where the codewords can cross byte boundaries or word boundaries.

  Here, let us consider a case where a write error occurs when writing the compressed data to the nonvolatile memory during hibernation or swapping. Writing to the non-volatile memory is performed by designating a physical block address in units of blocks of a certain size, but when writing of a certain block fails, rewriting to a different physical block address is required. A “block” means a physical recording unit of a storage device and is also called a sector.

  In order to cope with such rewriting, for example, the entire compressed data (hereinafter also referred to as block data) to be written to the block is temporarily stored in the main memory or a buffer provided separately until the writing succeeds. It is possible to save it. In this way, if writing failure is found, rewriting may be performed with temporarily stored data.

  However, in this method, it is necessary to secure an area for storing block data on the main memory or to provide a separate buffer. Depending on the system, there may be a case where an area for this purpose cannot be secured on the main memory.

  On the other hand, if the block data is not temporarily stored, the block data is lost when it is determined that the writing has failed. In this case, it is necessary to recreate the block data, and therefore, a process (rollback) for returning the internal state of the compression device to the state before the compression of the block is required.

  By the way, as described above, the lossless compression algorithm has a characteristic that the internal state is determined by the portion of the input sequence from the beginning to the present. For this reason, the same codeword cannot be generated even if recompression is performed from the middle of the input sequence.

  Also, the codeword length is variable and its size cannot be predicted before compression. Therefore, even if the input sequence is divided and compressed every fixed amount, the code word sequence obtained by compressing the fixed amount does not fit in one block, or conversely it is too short for the block length. It can be a situation.

  Furthermore, it is not possible to delimit a codeword sequence obtained by compression for each block size because the codeword sequence may straddle between blocks. Even if the boundary between the head of the block and the code sequence coincides, it is not clear to which address in the main memory the code word at the head of the block data corresponds.

  As described above, in lossless compression, it is difficult to regenerate compressed data in units of blocks, which are compressed data writing units, and there are many cases in which compression must be performed again from the beginning of input.

  The present invention has been made in view of the above circumstances, and provides a lossless compression technique capable of regenerating data after compression in writing units.

  One aspect of the present invention is a compression device. The compression apparatus includes a compression execution unit, an input unit, a head address holding unit, a writing unit, and a control unit.

  The compression execution unit performs lossless compression on the input data and outputs compressed data.

  The input unit reads data from the first storage device, and repeats the start and stop of input to the compression execution unit.

  The head address holding unit stores the address of the first storage device at the head of the data input this time for each input by the input unit.

  The writing unit writes the compressed data output by the compression execution unit for the data input this time to the second storage device for each input by the input unit, and the success or failure of the current writing A completion signal indicating is output. In writing, the compressed data is padded with predetermined data so that the size of the data to be written becomes the writing unit of the second storage device.

  The control unit resets the internal state of the compression execution unit every time the input stops input, or saves the internal state of the compression execution unit every time the input unit starts input.

  In addition, the input unit stops input before the maximum size that can be obtained after compression of the data already input to the compression execution unit by the current input reaches the writing unit of the second storage device, When the writing unit outputs the completion signal, the input is resumed. Upon restarting, the writing unit responds to the address corresponding to the previous input held in the head address holding unit and the completion signal. The data to be input to the compression execution unit is determined.

  Note that a representation in which the apparatus according to the above aspect is replaced with a system or method, a program that causes a computer to execute as the apparatus, and the like are also effective as an aspect of the present invention.

  According to the technology of the present invention, it is possible to regenerate compressed data in units of writing in a storage device to which compressed data is written.

It is a figure which shows the reversible compression apparatus concerning the 1st Embodiment of this invention. It is a figure which shows the reversible compression apparatus concerning the 2nd Embodiment of this invention. It is a figure which shows the reversible compression apparatus concerning the 3rd Embodiment of this invention. It is a figure which shows the reversible compression apparatus concerning the 4th Embodiment of this invention.

  Embodiments of the present invention will be described below with reference to the drawings. For clarity of explanation, the following description and drawings are omitted and simplified as appropriate. Each element described in the drawings as a functional block for performing various processes can be configured by a CPU, a memory, and other circuits in terms of hardware, and a program loaded in the memory in terms of software. Etc. Therefore, it is understood by those skilled in the art that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof, and is not limited to any one. Note that, in each drawing, the same element is denoted by the same reference numeral, and redundant description is omitted as necessary.

  Further, the above-described program can be stored using various types of non-transitory computer readable media and supplied to a computer. Non-transitory computer readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic recording media (for example, flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (for example, magneto-optical disks), CD-ROM (Read Only Memory) CD-R, CD -R / W, semiconductor memory (for example, mask ROM, PROM (Programmable ROM), EPROM (Erasable PROM), flash ROM, RAM (Random Access Memory)). The program may also be supplied to the computer by various types of transitory computer readable media. Examples of transitory computer readable media include electrical signals, optical signals, and electromagnetic waves. The temporary computer-readable medium can supply the program to the computer via a wired communication path such as an electric wire and an optical fiber, or a wireless communication path.

<First Embodiment>
FIG. 1 shows a lossless compression apparatus 100 according to an embodiment of the present invention. The lossless compression apparatus 100 is provided in a computer, for example, and compresses an information sequence on a first storage device such as a main memory (not shown) for hibernation, swapping, etc., and a second storage device such as an auxiliary storage ( (Not shown), and includes an input unit 110, a compression execution unit 120, a writing unit 130, a size measuring unit 140, and a control unit 150.

  The input unit 110 is controlled by a control signal CTR (to be described later) from the size measuring unit 140 and a completion signal RSU (to be described later) from the writing unit 130, and reads out the information series D1 that is data to be compressed from the main memory and compresses it. Data is supplied to the compression execution unit 120 by repeating the start and stop of data input to the execution unit 120. Specifically, the input unit 110 stops data input to the compression execution unit 120 when receiving the control signal CTR, and resumes data input to the compression execution unit 120 when receiving the completion signal RSU.

  In addition, the input unit 110 includes a head address holding unit 112. The start address holding unit 112 is, for example, a register, and temporarily stores the address in the main memory of the first byte of data input this time each time the input unit 110 starts to input data to the compression execution unit 120.

  Further, when resuming data input to the compression execution unit 120, the input unit 110 sends to the compression execution unit 120 according to the address corresponding to the previous input held in the head address holding unit 112 and the completion signal RSU. Determine the data to be entered this time. Details will be described later.

  The compression execution unit 120 performs lossless compression, and includes a dictionary buffer 122 and a matching result holding unit 124. The compression execution unit 120 inputs the data (input data D2) input from the input unit 110 into the dictionary buffer 122 to perform matching, and updates the matching result held in the matching result holding unit 124 while encoding the codeword To generate a compression. The dictionary buffer 122 and the matching result holding unit 124 are normally provided in this type of compression device that performs lossless compression, and detailed description thereof is omitted here. The compression execution unit 120 is not patented for the compression algorithm or the like except that the control unit 150 can initialize (reset) the internal state (contents stored in the dictionary buffer 122 and the matching result holding unit 124). This is the same as that of a conventionally known lossless compression apparatus such as that described in Document 5. Hereinafter, data that the compression executing unit 120 compresses the input data D2 and outputs to the writing unit 130 is referred to as “output data D3”.

  For easy understanding, the input unit 110 from the start to the stop of input is referred to as one input. Further, the compression performed by the compression execution unit 120 on all input data D2 input by a single input of the input unit 110 is referred to as a single compression of the compression execution unit 120.

  The writing unit 130 writes the compressed data (output data D3) obtained by the current compression to the auxiliary memory for each compression of the compression execution unit 120, and also sends a completion signal RSU indicating the success or failure of the current writing. Output to the input unit 110. Further, when writing, the writing unit 130 adds predetermined data (padding data) to the output data D3 so that the size of the data to be written becomes a constant size. This fixed size is the size of a block or sector that is a unit of writing to the auxiliary storage, and is hereinafter referred to as “writing unit S1”. The data that the writing unit 130 adds padding data to the output data D3 and writes to the auxiliary memory is referred to as block data D4.

  The size measurement unit 140 is, for example, a counter, counts the size (for example, the number of bytes) of the output data D3 from the compression execution unit 120, and outputs the control signal CTR every time the count value becomes equal to or greater than the threshold value S2. The control signal CTR is output to the input unit 110 control unit 150. Further, the size measuring unit 140 clears the count value for the next count after the output of the control signal CTR. That is, the size measurement unit 140 measures the size of the output data D3 output from the compression execution unit 120 from the start of the current input for each input by the input unit 110, and the measured size is set to the threshold value S2. The control signal CTR is output when the threshold value S2 is exceeded or the threshold value S2 is exceeded, and the count value is cleared.

  When receiving the control signal CTR from the size measuring unit 140, the control unit 150 resets the internal state of the compression executing unit 120. That is, in the lossless compression apparatus 100 according to the present embodiment, the control unit 150 includes the internal portion of the compression execution unit 120 every time the count value of the size measurement unit 140 becomes equal to or greater than the threshold value S2, that is, every time the input unit 110 stops input. It functions as a reset unit that resets the state.

  The threshold value S2 is a compression execution unit for the data input this time when the input from the input unit 110 is stopped when the size of the output data D3 measured by the size measurement unit 140 reaches the threshold value S2 or more. The size of all the data 120 output to the writing unit 130 is determined to be equal to or less than the writing unit S1.

For example, the threshold value S2 is determined according to the following equation (1).
Threshold S2
= Writing unit S1
-Maximum codeword bit length x (number of data words that have been input but not yet output as compressed data)-End word bit length (1)

  In Expression (1), “codeword maximum bit length” is the bit length of the longest codeword of the codewords that replace the data string when the compression execution unit 120 compresses. “The number of words of data that have been input but not yet output as compressed data” is the data D2 that the input unit 110 has already output to the compression execution unit 120 in the current data input, and is still being processed by the compression execution unit 120 This is the number of data words that are not output as compressed data. The “word” is a unit at the time of input by the input unit 110, and is “1 byte”, for example. The “end word bit length” is a bit length of a word (end word) indicating the end of the output data D3 obtained by the current compression of the compression execution unit 120. In the lossless compression apparatus 100 according to the present embodiment, the compression execution unit 120 adds an end word to the end of the output data D3 every time compression of data input by each input is completed.

  By doing so, the operations of the input unit 110 and the compression execution unit 120 are stopped before the size of the output data D3 obtained by the current compression of the compression execution unit 120 reaches the writing unit S1. Therefore, the total size of the output data D3 obtained by one compression of the compression execution unit 120 is equal to or less than the writing unit S1.

  Since the writing unit 130 adds padding data to the output data D3 so that the writing unit S1 becomes the writing unit S1 at every writing, the size of the block data D4 is always the same as the writing unit S1.

  This will be described using a specific example. For example, the writing unit S1 is “1000 bytes”, and 1000 bytes of data D2 has already been output to the compression execution unit 120 by this input. It is assumed that 9990 bytes of the 1000-byte data D2 are already compressed and output to the writing unit 130 as compressed data (output data D3). At this time, “10” of 10 bytes out of the 1000 bytes is “the number of data words that have been input but not yet outputted as compressed data” in the above formula (1).

  Assume that the size of 9990 bytes after compression is 995 bytes. A 5-byte margin remains until the writing unit S1. Therefore, if the last 10 bytes can be compressed to “5 bytes-end word bit length”, the compression execution unit 120 can suppress the size of the output data S3 obtained by the current compression to the writing unit S1 or less. . When the size of the output data S2 is less than the writing unit S1, the size of the block data D4 becomes the writing unit S1 due to the padding of the writing unit 120.

  When receiving the completion signal RSU, the input unit 110 performs different operations depending on whether the writing is successful or unsuccessful. Specifically, when the completion signal RSU indicates successful writing, the input unit 110 updates the address held in the head address holding unit 112 and then reads the data to be supplied to the compression execution unit 120 next. Start output. As a result, the compression execution unit 120 starts to compress the data of the next block.

  On the other hand, when the completion signal RSU indicates a write failure, the input unit 110 reads the main memory from the address held in the head address holding unit 112 and outputs the read data to the compression execution unit 120. As a result, the compression executing unit 120 regenerates the block data for which writing has failed.

  Thus, according to the lossless compression apparatus 100 of the present embodiment, every time the size of the output data D3 from the compression execution unit 120 exceeds the threshold S2, the data input by the input unit 110 and the compression by the compression execution unit 120 are performed. Is stopped, the internal state of the compression execution unit 120 is reset, and the writing unit 130 writes block data D4 of a certain size (writing unit S1) to the auxiliary storage. That is, all compressed data obtained by one compression can be written in one block. Since the head address holding unit 112 holds the address in the main memory of the first bit of the data input to the compression execution unit 120 this time, when the block data D4 has been successfully written, the next block data D4 If the writing fails, the failed block data D4 can be regenerated, and there is no need to redo the information series D1 from the beginning. Further, there is an advantage that a memory for temporarily storing the block data D4 is not required.

<Second Embodiment>
FIG. 2 shows a lossless compression apparatus 200 according to the second embodiment of the present invention. The lossless compression apparatus 200 has the same configuration as the input unit 110 except that the input unit 210 is provided instead of the input unit 110. The input unit 210 is the same as the input unit 110 except that an input queue 214 is provided.

  The lossless compression apparatus 200 can obtain the same effect as the lossless compression apparatus 100, and by providing the input queue 214, the information series D1 read from the main memory can be accumulated to some extent, so that data transfer is performed at high speed. At the same time, the control of the start and stop of input is simplified.

<Third Embodiment>
FIG. 3 shows a lossless compression apparatus 300 according to the third embodiment of the present invention. The lossless compression apparatus 300 is the same as the lossless compression apparatus 100 except that the lossless compression apparatus 300 includes a buffer 320 and the control unit 350 is different from the control unit 150 in the lossless compression apparatus 100. Here, only the control unit 350 and the buffer 320 will be described.

  The control unit 350 copies the internal state of the compression execution unit 120, that is, the contents of the dictionary buffer 122 and the matching result holding unit 124 to the buffer 320 at the start of each compression by the compression execution unit 120.

  The buffer 320 includes a dictionary saving buffer 322 that stores a copy of the contents of the dictionary buffer 122 and a matching result saving buffer 324 that stores a copy of the contents of the matching result holding unit 124. The control unit 350 overwrites and copies the contents of the dictionary buffer 122 to the dictionary saving buffer 322 and overwrites the contents of the matching result holding unit 124 to the matching result saving buffer 324 when the compression execution unit 120 discloses each compression.

  Further, in the lossless compression apparatus 300 of the present embodiment, the completion signal RSU from the writing unit 130 is also output to the control unit 350. When the writing by the writing unit 130 fails in response to the completion signal RSU, the control unit 350 overwrites the contents of the dictionary saving buffer 322 and the matching result saving buffer 324 to the dictionary buffer 122 and the matching result holding unit 124, respectively. Restore by copying. In this case, as described in the description of the lossless compression apparatus 100 according to the first embodiment, the input unit 110 reads data from the address held by the head address holding unit 112 and inputs the data to the compression execution unit 120. Therefore, the block data D4 for which writing has failed is regenerated.

  On the other hand, when the completion signal RSU indicates that the writing by the writing unit 130 is successful, the control unit 350 does not return the contents of the buffer 320 to the dictionary buffer 122 and the matching result holding unit 124. As described above, in this case, the input unit 110 updates the address held in the head address holding unit 112 and then reads and inputs data to be input to the compression execution unit 120 next. The compressed data next to the block data D4 is regenerated.

  As described above, the lossless compression apparatus 300 according to the present embodiment saves the internal state of the compression execution unit 120 in the buffer 320 at the start of each compression, and writing of compressed data obtained by the current compression fails. In this case, the contents saved in the buffer 320 are returned to the compression execution unit 120. Therefore, the block data D4 can be regenerated when the writing of the block data D4 fails as in the lossless compression apparatus 100 of the first embodiment, and the compression efficiency is improved compared to the lossless compression apparatus 100. .

<Fourth embodiment>
FIG. 4 shows a lossless compression apparatus 400 according to the fourth embodiment of the present invention. The lossless compression apparatus 400 is the same as the lossless compression apparatus 300 except that the compression execution unit 120 in the lossless compression apparatus 200 is provided instead of the input unit 110. That is, the lossless compression apparatus 400 is the same as the lossless compression apparatus 300 except that the input unit 210 includes the input queue 214.

  Therefore, the reversible compression apparatus 400 can obtain the same effect as the reversible compression apparatus 300, and by providing the input queue 214, the information series D1 read from the main memory can be accumulated to some extent, so that data transfer As a result, the input start and stop control becomes simple.

  The present invention has been described above based on the embodiment. The embodiment is an exemplification, and various changes, increases / decreases, and combinations may be added to the above-described embodiment without departing from the gist of the present invention. It will be understood by those skilled in the art that modifications to which these changes, increases / decreases, and combinations are also within the scope of the present invention.

  For example, in the lossless compression apparatus 100 and the lossless compression apparatus 200, the control unit 150 is provided outside the compression execution unit 120, but the control unit 150 may be provided inside the compression execution unit 120.

  In the lossless compression apparatus of each embodiment, the head address holding unit 112 is provided inside the input unit 110 or the input unit 210. However, the head address holding unit 112 may be provided outside the input unit. Of course.

  In the compression apparatus of each embodiment described above, the size measuring unit 140 is provided, and when the size measured by the size measuring unit is equal to or larger than the threshold value S2 defined by the equation (1), the input by the input unit 110 is performed. Like to stop. Without providing the size measurement unit 140, the input unit 110 stops the current input before the maximum possible size after compression of the data already input to the compression execution unit 120 by the current input reaches the writing unit S1. You may make it do. The “maximum possible size after compression of input data” can be, for example, the product of the number of words of already input data and the maximum bit length of the code word.

  This application claims the priority on the basis of Japanese application Japanese Patent Application No. 2011-120037 for which it applied on May 30, 2011, and takes in those the indications of all here.

DESCRIPTION OF SYMBOLS 100 Reversible compression apparatus 110 Input part 112 First address holding part 120 Compression execution part 122 Dictionary buffer 124 Matching result holding part 130 Writing part 140 Size measurement part 150 Control part 200 Lossless compression apparatus 210 Input part 214 Input queue 300 Lossless compression apparatus 320 Buffer 322 Dictionary saving buffer 324 Matching result saving buffer 350 Control unit 400 Lossless compression device CTR Control signal RSU completion signal

Claims (3)

Compression execution means for performing lossless compression on input data and outputting compressed data;
Input means for reading data from the first storage device and repeating start and stop of input to the compression execution means;
For each input by the input means, a head address holding means for storing the address in the first storage device at the head of the data input this time;
For each input by the input means, the compressed data output by the compression execution means for the data input this time is written to the second storage device, and a completion signal indicating the success or failure of the current write is provided. Writing means for outputting, in writing, the writing means for padding predetermined data into the compressed data so that the size of the data to be written is a writing unit of the second storage device;
A control unit that resets the internal state of the compression execution unit every time the input unit stops input, or saves the internal state of the compression execution unit every time the input unit starts input;
The input means stops the input before the maximum possible size after compression of the data already input to the compression executing means by the current input reaches the writing unit of the second storage device, The input means restarts the input when the completion signal is output, and upon restart, according to the address corresponding to the previous input held in the head address holding means and the completion signal, A compression apparatus for determining data to be input to the compression execution means. A size measuring unit that measures the size of the compressed data output by the compression execution unit from the start of the input for each input by the input unit;
The compression execution means adds an end word to the end of the compressed data obtained by compressing the data input each time,
The input means stops input when the size measured by the size measuring means reaches a threshold,
The compression apparatus according to claim 1, wherein the threshold value is determined according to Equation (1).
Threshold = Writing unit −Maximum bit length of code word × (Number of data words that have been input but not yet output as compressed data) −End word bit length (1)   The compression apparatus according to claim 1, wherein the input unit includes an input queue that temporarily stores data read from the first storage device.

Images