JP3661944B2 - Arithmetic engine, microprocessor, and arithmetic control method thereof - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a calculation engine and a microprocessor on which the calculation engine is mounted, and more particularly to a calculation engine suitable for use in a multimedia processor that processes various data streams such as image processing and encryption processing, and a microprocessor on which the calculation engine is mounted.
[0002]
[Prior art]
In recent years, with the progress of processor technology, multimedia processors for processing various stream data such as image data, encrypted data, and audio data have been developed.
[0003]
Conventionally, when a plurality of stream data having different formats are used, when the real-time property is important, as many arithmetic units as the number of input stream data are provided, and the data is processed in real time by processing simultaneously. It was output. The calculation result is stored in the main storage device and is output as necessary. If real time is not important, all the stream data is processed and the result is stored in the main memory, then the next stream data is processed and stored in the main memory. Processing was performed in order.
[0004]
As a technique for handling an image stream, there is a “multi-screen display device” disclosed in Japanese Patent Application Laid-Open No. 2000-13712. This publication discloses a technique for inputting a plurality of image streams and selectively displaying them. Yes. According to this technique, data obtained by taking out only I pictures for each of a plurality of input streams is collectively stored, and expanded data can be displayed on the same screen as necessary.
[0005]
[Problems to be solved by the invention]
When processing a plurality of stream data of different formats, the operation cannot be interrupted in the middle, and it is necessary to process at least the break of the stream. For this reason, an arithmetic unit is provided for each stream so that processing can be performed exclusively. However, it is necessary to provide arithmetic units only for the types of stream data, which causes an increase in product cost.
[0006]
For example, in the above prior art, it is necessary to provide as many detection means and storage means as the number of stream data.
[0007]
Also, when processing one stream data to a break and processing stream data in another format, a storage device capable of storing the processed stream data in order to prevent a time lag from occurring It was necessary to provide.
[0008]
The present invention has been made to solve the above-described problems, and a microprocessor equipped with a calculation engine can process a plurality of stream data in real time, and has a simple internal mechanism. The object is to provide a microprocessor equipped with an arithmetic engine.
[0009]
[Means for Solving the Problems]
An arithmetic engine according to the present invention is an arithmetic engine that inputs stream data, and includes an arithmetic unit that outputs an arithmetic result and an arithmetic setting register that can set the arithmetic method of the arithmetic unit. The stream data input to the calculation engine is composed of a header part describing a calculation method or additional data, and a data part to be calculated. The calculation engine divides the input stream data into a header part and a data part, and the calculation method and additional part of the header part are input to the calculation setting register, and a part of the additional data and data of the header part are input. The data of the part is input to the calculator. Then, at the end of the calculation, the contents of the calculation setting register are output as the data of the footer unit, the contents of the calculator are output as the calculation result, and output stream data is created and output based on the footer unit and the calculation result.
[0010]
Then, the output footer information is used as the header information of the subsequent input stream data. In this way, even during the calculation, it can be output as output stream data and used for the next calculation. Therefore, various types can be obtained with a simple mechanism without preparing a special storage device. Can be processed efficiently.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments according to the present invention will be described below with reference to FIGS.
[0012]
Embodiment 1
A first embodiment according to the present invention will be described below with reference to FIGS.
[0013]
First, a computer system comprising an arithmetic engine according to the present invention, a microprocessor on which it is mounted, and a main memory will be described with reference to FIGS.
FIG. 1 is a schematic diagram showing the relationship between the configuration of a calculation engine according to the present invention and stream data.
FIG. 2 is a block diagram of a computer system comprising a microprocessor equipped with a calculation engine according to the present invention and a main memory.
[0014]
As shown in FIG. 2, the microprocessor on which the arithmetic engine of the present invention is mounted includes an arithmetic engine 200 and a core 205. The arithmetic engine 200 is a dedicated processor for performing special arithmetic operations, and the core 205 loads and stores data from the main storage device 204 by interpreting and executing instructions.
[0015]
In the embodiment of the present invention, the input stream 100 placed in the main storage device 204 is transferred to the arithmetic engine 200 via the microprocessor core 205. The transferred result is written back as an output stream 105 from the microprocessor core to the main memory 204.
[0016]
Next, the configuration of the arithmetic engine according to the present invention and the operation when processing stream data will be described with reference to FIG.
[0017]
An input stream 100 input to the calculation engine of the present invention includes a header 101 describing a calculation method and additional data, and input data 102 to be calculated.
[0018]
This input stream 100 is input to the arithmetic engine 200, a part of the content of the header 101 is transferred to the arithmetic setting register 201, a part is transferred to the arithmetic unit 202, and the input data 102 is transferred to the arithmetic unit 202. Is done.
[0019]
Then, by interpreting the contents of the transferred header 101, the processing method of the input data 102 of the calculator 202 is determined, and the calculation result 104 is output. The input data 102 has a variable length, and is continuously input to the calculator 202 until the break of the input data 102 is output, and the calculation result 104 is output.
[0020]
The output stream 105 output from the calculation engine after the calculation is completed has a configuration in which a footer 104 is attached after the calculation result 103.
[0021]
The footer 103 is created based on the header 101, and is a calculation method or a calculation result.
[0022]
Next, processing of the input stream 100 and the output stream 105 will be described with reference to FIGS. 3 and 4.
FIG. 3 is a schematic diagram showing the relationship between each part of the input stream 100 and the output stream 105.
FIG. 4 is a schematic diagram illustrating a state when a plurality of input streams 100 are input and a plurality of output streams 105 are output.
[0023]
As shown in FIG. 3, the header h ₁ and the data d ₁ are input to the arithmetic engine 200 as the input stream 100. Then, it is assumed that the calculation result r ₁ and the footer f ₁ are output as the output stream 105 of the calculation result.
[0024]
At this time, the microprocessor equipped with the arithmetic engine of the present invention uses the footer f ₁ as the header h ₂ of the next input stream 100.
[0025]
The contents of the calculation may be the same as the contents of the footer f _{1 or} may be processed and used.
[0026]
In addition, it is assumed that a plurality of input streams are input to the arithmetic engine as shown in FIG. At this time, it is assumed that the input stream A ₁ and the input stream A ₂ are related data, and the input stream A ₂ is a continuation data of the input stream A ₁ .
[0027]
The input stream B is data that is processed independently of them.
[0028]
Originally, it is most efficient to process the input stream A ₁ and then the input stream A ₂ . However, if that must be processed within a certain time an input stream B will be processed in preference to the input stream A _2.
[0029]
In such a case, the present invention reflects such information in the footer A ₁ created based on the header A ₁ of the input stream A ₁ , and creates the header A ₂ of the input stream A _2. by, taken over the intermediate result of the operation of the input stream a _1, it is possible to resume the operation.
[0030]
Therefore, interruption (order disorder) by such an input stream B is also permitted.
[0031]
Next, specific examples of the input stream and output stream calculated by the calculation engine of the present invention will be described with reference to FIG.
FIG. 5 is a schematic diagram showing an input stream and an output stream used in the cryptographic module.
[0032]
The arithmetic engine of the present invention can be used for processing of a cryptographic module, for example. The calculation method applied to the header part is a mode of what kind of encryption is used, and the additional data includes a data length, an encryption key, an initial value of encryption, and the like.
[0033]
For example, when using the CBC mode of encryption, the header includes an encryption mode indicating the CBC mode from word 0 to word 5, a data length n, an encryption key of 64 bits, and an encryption initial value of 64 bits. The encryption initial values stored in the words 4 and 5 are encrypted by taking the exclusive OR of the operation result with the next input data, particularly in the CBC mode. Since there is no calculation result, it is necessary to set an initial value in advance, and this is the initial value.
[0034]
The words 6 to (n + 5) of the input stream 100 are plain text, that is, data before encryption.
[0035]
The calculation method of the calculation engine is defined in the encryption mode portion stored in the header, and the processing amount of plain text as data is shown in the data length n. When processing of this value is completed, the encryption engine Start footer output. The information stored in the encryption mode and the data length is sent to the operation setting register 201 in the operation engine 200 shown in FIG.
[0036]
The key and the initial encryption value are data used for encrypting the transmitted plaintext. Similar to the plaintext data, these are sent to the computing unit 202 of the computation engine 200, and the plaintext data is encrypted by the computing unit. Then, the data is output as the operation result sequentially from the encrypted data, and when the processing of the data length n is completed, the encryption mode used for the encryption as the footer, the data length n, the key 64 bits, and the intermediate result 64 Output bits.
[0037]
The intermediate result is a value used in the exclusive OR of the CBC mode, and becomes the initial encryption value for the next encryption.
[0038]
Here, the data portion corresponds to the values from word 0 to word n of the output stream in which the ciphertext is stored, and the footer corresponds to the output in which the encryption mode, the data length n, the key and the intermediate result are stored. This is a value from word n to word (n + 5) of the stream. In the next cryptographic processing operation, the information of these footers can be used as a header as it is, and it is possible to obtain a ciphertext by giving plaintext as data subsequently.
[0039]
[Embodiment 2]
Hereinafter, a second embodiment according to the present invention will be described with reference to FIG.
FIG. 6 is a system configuration diagram of a system in which the arithmetic engine of the present invention is applied to processing of moving image data.
[0040]
The present embodiment is an example in which the stream processing of the calculation engine of the present invention is applied to the processing of moving image data.
[0041]
In this system, as shown in FIG. 6, the moving image data A209 and the moving image B210 in the auxiliary storage device such as a hard disk are connected to the microprocessor 203 using the network 206, and the moving image A209 and the moving image B210 can be output simultaneously. It is.
[0042]
First, the system transfers data A 207 to the microprocessor 203 using the network 206. The memory controller 212 in the microprocessor 203 saves the received data A 214 after the header A 213 that is always saved in the main storage device 204, and creates stream data A.
[0043]
The stream data A is transferred via the memory controller 212 in response to a request from the coprocessor 211, performs an operation according to the header A 213, and stores the result in the main storage device 204 via the memory controller 212.
[0044]
At this time, the data to be stored are the moving image A 215 and the footer A 216 which are the calculation results. Since this footer A216 is used as the header A213 of the next data coming from the data A207, it is held in the main memory 204 as it is. The moving image data 215 in the main storage device 204 passes through the NTSC output 213 again through the memory controller 212 and is output to the outside as the moving image A209. The data B 208 to be output simultaneously is transferred to the main storage device 204 through the memory controller 212 via the network 206.
[0045]
On the other hand, at this time, the transferred data B 218 is stored after the header B 217 describing the calculation method performed by the coprocessor 211 to form stream data B. The stream data B is transferred to the coprocessor 211 via the memory controller and processed according to the header B217, and the moving image B219 and the footer B as the calculation results are transferred to the main storage device.
[0046]
Then, only the portion of the moving image B219 is output to the outside as the moving image B210 from the main storage device 204 via the memory controller 212 and the NTSC output 213. A series of these processes are performed by time-sharing control, and the coprocessor 211 performs the process by switching between the data A 214 and the data B 218, and the processing method is as described in the header A 213 and the header B 217, respectively. Can be done.
[0047]
As the contents of the header, specifically, the enlargement ratio, display position, reproduction speed, etc. of the moving image data can be considered.
[0048]
What is important here is that the header and the footer are always resident in the main storage device 204, so that the operation speed is not reduced without accessing the auxiliary storage device.
[0049]
【The invention's effect】
According to the present invention, it is possible to provide a microprocessor that can process a plurality of stream data in real time in a microprocessor equipped with an arithmetic engine, and that includes an arithmetic engine with a simple internal mechanism. Can do.
[Brief description of the drawings]
FIG. 1 is a schematic diagram showing a relationship between a configuration of a calculation engine according to the present invention and stream data.
FIG. 2 is a block diagram of a computer system comprising a microprocessor equipped with a calculation engine according to the present invention and a main memory.
FIG. 3 is a schematic diagram showing the relationship between each part of an input stream 100 and an output stream 105;
FIG. 4 is a schematic diagram showing a state when a plurality of input streams 100 are input and a plurality of output streams 105 are output.
FIG. 5 is a schematic diagram showing an input stream and an output stream used in the cryptographic module.
FIG. 6 is a system configuration diagram of a system in which the arithmetic engine of the present invention is applied to processing of moving image data.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 100 ... Input stream 105 ... Output stream 124 ... Input data group 125 ... Output data group 200 ... Calculation engine 201 ... Calculation setting register 202 ... Calculator

Claims (5)

In the calculation engine that inputs and calculates stream data,
The stream data to be input is a header portion describing the calculation method or additional data, and a data portion to be calculated,
The stream data that is calculated and output is a calculation result part and a footer part that describes a calculation method or additional data.
An arithmetic engine characterized in that the footer part is used as information that becomes a source of a header part of stream data to be input next to the arithmetic engine. In the calculation engine that inputs and calculates stream data,
The calculation engine includes a calculation setting register and a calculator.
The stream data to be input is a header portion describing the calculation method or additional data, and a data portion to be calculated,
The stream data that is calculated and output is a calculation result part and a footer part that describes a calculation method or additional data.
The calculation setting register inputs a calculation method and additional data of the header part, outputs a footer part,
The computing unit inputs the data included in the additional data of the header part and the data part, and outputs the computation result of the computation result part,
An arithmetic engine characterized in that the footer part of the output stream data is used as information that becomes the basis of the header part of the stream data to be input to the arithmetic engine next time. In a calculation control method of a calculation engine that inputs and calculates stream data,
The stream data to be input is a header portion describing the calculation method or additional data, and a data portion to be calculated,
The stream data that is calculated and output is a calculation result part and a footer part that describes a calculation method or additional data.
An arithmetic control method for an arithmetic engine, characterized in that the footer part is used as information that is a source of a header part of stream data that is next input to the arithmetic engine. In a microprocessor that inputs and operates stream data,
The microprocessor includes an arithmetic engine,
The main storage device stores one or more data groups to be calculated by different calculation methods, and a header portion describing the calculation method or additional data of the data,
From the main storage device, read out the header part describing the calculation method or additional data as the input data stream, and the data part to be calculated,
Input to the arithmetic engine,
A calculation result based on the calculation method or additional data described in the header part is output as an output data stream as a calculation result part and a footer part describing the calculation method or additional data,
A microprocessor characterized in that they are written again into the main memory. In a computer system for processing stream data,
This computer system
An auxiliary storage device for storing the stream data;
A microprocessor;
A main storage device;
An output device for outputting a result processed by the microprocessor;
The stream data is composed of a part describing a calculation method or additional data and a data part to be calculated,
The microprocessor reads the stream data from the auxiliary storage device,
Holding the information of the part describing the calculation method or additional data in the main storage device,
According to the information of the part describing the calculation method or additional data held in the main storage device, the microprocessor performs an operation on the data part to be calculated,
As a result of the calculation, information on the calculation method or additional data is written back to the main memory,
And by outputting the calculation result to the output device,
A computer system, wherein stream data corresponding to different calculation methods are continuously read from the auxiliary storage device and processed.

Images