JPH103466A - Central arithmetic processing unit and image generator using the processing unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase the drawing speed without increasing the hardware capacity by preparing the arithmetic parts which exclusively perform the matrix and a product-sum operations respectively. SOLUTION: An instruction cache memory 112, a data cache memory 113, a geometric arithmetic part 114 which primarily performs the matrix and function operations such as the coordinate transformation, etc., and a physical arithmetic part 115 which constructs a DDA(digital differential analyzer) are connected to a CPU main body 111 via an internal bus 110. In such a constitution, a main CPU can perform its processing to some extent without using a main bus 10. Thus, the bus 10 can be easily opened. The parts 114 and 115 consist of the arithmetic processing parts 114P and 115P and the register packets 114R and 115R respectively and can perform each arithmetic processing in every packet. A DDA array constructing the part 115P consists of plural adders and registers.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is suitable for use in an apparatus which needs to generate an image at high speed, such as an image generating apparatus using a computer, such as a graphic computer, a special effect device, and a game machine. The present invention relates to a central processing unit (hereinafter, referred to as a CPU) and an image generating apparatus using the same.

[0002]

2. Description of the Related Art For example, when a three-dimensional image is generated (drawn) by a graphic computer, an image drawing process is performed in the following processing procedure.

That is, in computer graphics, an object to be drawn (called an object)
Is divided into small basic shapes (polygons)
The shape, position, orientation, color, pattern, and the like of the polygon are given as polygon data that determines an image. The shape, position, and orientation of a polygon are determined by the coordinates of its vertices.

In the image drawing process, first, for example, drawing data stored in a memory or a recording medium is read,
Only polygon data is extracted from the drawing data. Next, based on the information on the viewpoint position input from the predetermined input means, imagining a screen directly facing the viewpoint, converting vertex coordinates of each polygon constituting the object into coordinates on the screen, The converted data is temporarily stored in the memory.

The coordinate conversion of the vertex coordinates includes three-dimensional coordinate conversion for converting the orientation of the object with respect to the viewpoint and perspective conversion (2) corresponding to the distance from the viewpoint to the object with respect to the screen. Dimensional coordinate conversion). 3
The dimensional coordinate conversion is performed using a coordinate conversion matrix, and the perspective conversion is to perform a division operation using depth information (Z information) of a vertex.

Then, the obtained converted vertex data is
It is taken out of the memory and a polygon is drawn on the video RAM according to the coordinates.

Recently, three-dimensional graphics can be displayed on a personal computer or a video game machine, and a three-dimensional world can be represented by simulation, virtual reality, or the like. Along with this, dedicated hardware for graphics has been developed, and arithmetic processing such as coordinate transformation and perspective transformation can be performed at high speed.

[0008]

As described above, with the improvement in the ability to express a three-dimensional world, the ability to express physical changes such as when an object itself is deformed in response to its movement or movement is improved. Therefore, there is an increasing demand for more realistic representation, but conventionally, the computational power for physical representation is lower than that of the above-described geometric matrix operation. This is because, in the conventional graphics, the physical expression was not so much a problem, so that even if the CPU performed an arithmetic processing for the physical expression by software, it could cope sufficiently.

[0009] However, when a recent high-level expression is required, the number of calculations for the physical expression increases, and the balance becomes worse compared with the high-speed coordinate conversion calculation.

In view of the above, the present invention provides a CPU capable of increasing a drawing speed without increasing hardware when manufacturing a small graphic device having a high drawing function. The purpose is to:

[0011]

In order to solve the above problems, a CPU according to the present invention comprises a central processing unit main body,
Connected to the central processing unit via an internal bus,
A first dedicated operation unit dedicated to executing a matrix operation function; and a second dedicated operation unit connected to the central processing unit main body via an internal bus and exclusively executing a product-sum operation. It is characterized by the following.

An image generating apparatus according to the present invention includes a central processing unit main unit, a first dedicated operation unit connected to the central processing unit main unit via an internal bus, and exclusively executing a matrix operation function. A CPU connected to the main body of the central processing unit via an internal bus, and having a second dedicated operation unit for exclusively executing a product-sum operation, and a two-dimensional or three-dimensional representation of moving image data. When generating
A coordinate conversion operation is performed by the first dedicated operation unit, and a physical expression operation such as physical change or deformation is performed by the second dedicated operation unit.

In the image generating apparatus provided with the CPU having the above-described configuration, the CPU has a dedicated operation unit for coordinate conversion operation and a dedicated operation unit for physical expression operation. In addition, calculations for physical expression can be performed in parallel at high speed, and the calculation capability is improved. And
Since both dedicated operation units are in the same CPU, it is easy to use the relevant coordinate transformation operation result for the physical expression operation, and it is easier to use the dedicated hardware outside the CPU than in the case where dedicated hardware is provided outside the CPU. The interference of the main bus to which the CPU is connected can be reduced.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a CPU according to the present invention and an image generating apparatus using the same will be described below with reference to the drawings for a video game machine.

FIG. 2 shows an example of the configuration of an image generating apparatus according to an embodiment of the present invention. This example is an example of a game machine having a three-dimensional graphics function and a moving image reproducing function. .

FIG. 3 shows the appearance of the game machine of this embodiment. The game machine of this embodiment comprises a game machine main body 1 and a control pad 2 constituting an operation input section for a user. The control pad 2 is connected to the game machine main body 1 by connecting a connector plug 4 attached to a tip of a cable 3 connected to the control pad 2 to a connector jack 5A of the game machine main body 1. . In this example, two connector jacks 5A and 5B are provided on the game machine main body 1 so that two control pads 2 can be connected to the game machine main body 1 for a so-called battle game or the like. ing.

The game machine of this embodiment has a CD-ROM as an auxiliary storage means in which a game program and image data are written.
The game can be enjoyed by loading the ROM disk 6 into the game machine body 1.

Next, the configuration of the image generating apparatus of this example will be described with reference to FIG. A game machine as an image generating apparatus of this example includes a main bus 10 and a sub bus 2
0 is provided with two system buses. Data exchange between the main bus 10 and the sub-bus 20 is controlled by the bus controller 30.

The main bus 10 has a main CP.
U11, a main memory 12, an image decompression unit 13, a preprocessing unit 14, a drawing processing unit 15, and a main DMA controller 16 (hereinafter referred to as a main DMAC) are connected. A processing memory 17 is connected to the drawing processing unit 15, and the drawing processing unit 15 includes a so-called frame memory for display data and a D / A conversion circuit. The signal is output to the video output terminal 18. Although not shown, the video output terminal 18 is connected to, for example, a CRT display as a display device.

The sub bus 20 includes a sub CPU 21, a sub memory 22, a boot ROM 23, a sub DMA controller 24, an audio processor 25, an input unit 26, a CD-ROM decoder 27, and an extension The communication interface unit 28 is connected. Boot ROM
23 stores a program for starting up as a game machine. Further, the audio processing memory 25M is connected to the audio processing processor 25. The audio processing processor 25 has a D / A
A conversion circuit is provided to output an analog audio signal to the audio output terminal 29.

Then, the CD-ROM decoder 27
It is connected to the D-ROM driver 40 and has a CD-RO
The application program (eg, game program) and data recorded on the CD-ROM disk 6 loaded in the M driver 40 are decoded. CD-R
The OM disk 6 has, for example, a discrete cosine transform (DC
Image data of moving images and still images compressed by T) and image data of texture images for modifying polygons are also recorded. The application program on the CD-ROM disk 6 includes a polygon drawing command.

The input section 26 includes the control pad 2 as the operation input means described above, a video signal input terminal, and an audio signal input terminal.

The main CPU 11 manages and controls each unit on the main bus 10 side. Also, this main CP
U11 performs a part of the processing when the object is drawn as a group of many polygons. The main CPU 11 creates, on the main memory 12, a drawing command example for generating a drawing image for one screen, as will be described later. Main CP
Data exchange between the U11 and the main bus 10 is performed in packet units by converting data into a packet format, thereby enabling burst transfer. The internal configuration of the main CPU 11 will be described later in more detail.

The main memory 12 has a memory area for compressed image data and a memory area for decompressed decoded image data for moving image and still image data. The main memory 12 has a memory area for graphics data such as a drawing instruction sequence (this is called a packet buffer). This packet buffer is used to set a drawing command sequence by the main CPU 11 and
It is used for transferring the drawing instruction sequence to the drawing processing unit.

The image decompression unit 13 decompresses compressed image data reproduced from the CD-ROM disk 6, and includes hardware for a Huffman code decoder, an inverse quantization circuit, and an inverse discrete cosine transform circuit. Prepare. The Huffman code decoder may be processed by the main CPU 11 as software.

The drawing processor 15 executes a drawing command transferred from the main memory 12 and writes the result to the frame memory. The image data read from the frame memory is supplied to a video output terminal 1 via a D / A converter.
8 and displayed on the screen of the image monitor device.

The pre-processing unit 14 has a configuration of a processor having a CPU, and can share a part of the processing of the main CPU 11. For example, the pre-processing unit 14 may also perform a process of converting polygon data into two-dimensional coordinate data for display.

The basic processing of this game machine will be described below.

[Importing Data from CD-ROM Disk 6] When the game machine in the example of FIG. 2 is powered on and the CD-ROM disk 6 is loaded into the game machine body 1,
A program for executing a so-called initialization process for executing a game in the boot ROM 23 is stored in the sub CPU 21.
Is executed by Then, the data recorded on the CD-ROM disk 6 is captured as follows.

That is, from the CD-ROM disk 6,
The compressed image data, the drawing command, and the program executed by the main CPU 11 are stored in a CD-ROM driver 40, a CD-ROM driver.
Read via the ROM decoder 27 and the sub DMAC
24, the data is temporarily loaded into the sub memory 22.

The data fetched into the sub memory 22 is stored in the sub DMAC and the bus controller 3.
0, and further transferred to the main memory 12 by the main DMAC 16. The sub CPU 21 is configured to directly access the frame of the drawing processing unit 15, and the sub CPU 21 can also change the display image content separately from the control of the drawing processing unit 15. It has been.

[Expansion and Transfer of Compressed Image Data] Among the input data of the main memory 12, in this example, the compressed image data is transferred to the main memory 12 again by the main CPU 11 after the main CPU 11 performs the Huffman code decoding process. Is written to. Then, the main DMAC 16
Transfers the image data after the decoding processing of the Huffman code from the main memory 12 to the image decompression unit 13. The image decompression unit 13 performs dequantization processing and inverse DCT processing to perform decompression decoding processing of image data. The decompressed image data is transmitted from the main DMAC 16 to the main memory 1.
Transfer to 2.

The main CPU 11 transfers the decompressed data to the frame memory of the drawing processing unit 15 when a fixed amount of unit data called macroblocks of the decompressed image data is stored in the main memory 12. At this time, if the decompressed image data is transferred to the image memory area of the frame memory, it will be displayed on the image monitor as a background moving image. Also, the data may be transferred to the texture area of the frame memory. The image data of the texture area is used as a texture image for modifying a polygon.

[Processing and Transfer of Drawing Instruction Sequence] The polygons forming the surface of the object are drawn in order from the polygon located deeper in the depth direction in accordance with Z data which is three-dimensional depth information. An image can be stereoscopically displayed on the two-dimensional image display surface. Main CPU1
1 creates, on the main memory 12, a drawing command sequence for causing the drawing processing unit 15 to perform drawing in order from the polygon located at a deep position in the depth direction.

The main CPU 11 calculates the movement of the object and the viewpoint based on the user's operation input from the control pad of the input unit 26, and creates a polygon drawing command sequence on the main memory 12.

When this drawing command sequence is completed, the main DM
The AC 16 transfers the drawing command from the main memory 12 to the drawing processing unit 15 through the preprocessing unit 14 for each drawing command.

The drawing processing section 15 sequentially executes the transmitted data and stores the result in a drawing area of the frame memory. At the time of rendering the polygon, the data is sent to the gradient calculation unit of the rendering processing unit 15, and the gradient is calculated. The gradient calculation is a calculation for calculating the inclination of the plane of the mapping data when filling the inside of the polygon with the mapping data by polygon drawing. In the case of texture, polygons are filled with texture image data,
In the case of Gouraud shading, polygons are filled with luminance values.

Furthermore, the texture of a moving image is possible.
That is, in the case of the moving image texture, the compressed moving image data from the CD-ROM disk 6 is once read into the main memory 12 as described above. Then, the compressed image data is sent to the image decompression unit 13. The image data is expanded by the image expansion unit 13. At this time, as described above, a part of the decompression processing is performed by the main CP.
U11 bears.

The decompressed moving image data is sent to the texture area on the frame memory of the drawing processing unit 15. Since the texture area is provided in the frame memory of the drawing processing unit 15, the texture pattern itself can be rewritten for each frame.
As described above, when a moving image is sent to the texture area, the texture is dynamically rewritten and changed every frame. If texture mapping to polygons is performed using the moving image in the texture area, the texture of the moving image is realized.

[Configuration Example of Main CPU] FIG. 1 shows an example of the internal configuration of the main CPU 11. The main CPU 11 provides an internal bus 1 to the CPU main body 111.
10, an instruction cache memory 112, a data cache memory 113, a geometric operation unit 114 that mainly performs a matrix operation or a function operation such as coordinate conversion,
Digital Differential Analyzer (DDA)
(Analyzer) is connected.

As described above, since the main CPU 11 has the instruction cache memory 112, the geometric operation unit 114, and the physical operation unit 115 inside, the processing is performed to some extent without using the main bus 10. Therefore, the main bus 10 can be easily opened.

Then, the internal bus 110 and the main bus 1
Data input / output between the bus interface 1 and the bus interface 1
16, and may also be performed via a programmable packet engine (hereinafter, referred to as PKE) 117 that can change the procedure of packetizing and unpacketizing data.

The geometric operation unit 114 and the physical operation unit 115 are operated by the operation processing units 114P and 114P, respectively.
5P and the register packets 114R and 115R, and the arithmetic processing can be executed in packet units.

The DDA array (hereinafter, referred to as DDA array 115P) constituting the arithmetic processing section 115P of the physical arithmetic section 115 is composed of a plurality of adders and registers. FIG. 4 shows an operation unit 1 comprising this DDA array.
15P shows an example.

That is, a plurality of DDA arrays 115P for performing parallel processing, in this example, four pairs of registers 51 and 52, 53 and 54, 55 and 56, 5
7 and 58, and a corresponding number, in this example, four adders 61, 62, 63 and 64. The contents of one register 51, 53, 55, 57 of each of the pair of registers are sequentially rewritten by the input data. Each other register 52, 54, 56, 58 of the pair of registers
Stores the result of the previous addition operation.

The adders 61, 62, 63, 64
Adds the data of the pair of registers 51 and 52, 53 and 54, 55 and 56, 57 and 58 to each other, and outputs the added output data to the other register 5
It is configured to overwrite 2, 54, 56, 58.

Thus, one of the registers 51, 53, 5
When the amounts of change are sequentially set in 5, 57, and the calculation is repeatedly executed, the other registers 52, 54, 56, 58 obtain the integration result as the solution of the differential equation.

Therefore, when the amount of change of an object or the like calculated by the CPU main body 111 or the geometric operation unit 114 is sent to the physical operation unit 115, the target physical amount can be obtained at a high speed. .

For example, based on a user input or a game environment of a game program read from a CD-ROM, the CPU main body 111 or the geometric operation unit 114 calculates the acceleration F from the data of the applied force F and the mass M (weight) of the object. Is calculated, data on the change position of the object is obtained based on the acceleration data.

FIG. 5 shows a DDA array 115P in this case.
It is a figure for explaining the example of utilization of. In this case, in FIG. 5, 1 / M is 1 / weight (mass) array data, F is force array data, A is acceleration array data, V is velocity array data, and P is position array data. , On a packet-by-packet basis and at a frame rate.

In FIG. 5, portions surrounded by alternate long and short dash lines are portions where the DDA array 115P executes calculations. That is, in the case of this example, the DDA array 115P is used twice, and the position array data is obtained from the acceleration array data.

That is, in the case of the example of FIG. 5, for example, 1/1/3 of the object set by the game program
The weight array data 1 / M and force array data F generated from user input and the environment at that time are stored in the CPU main body 110.
A multiplication operation is performed by the geometric operation unit 114 and the acceleration array data A is obtained.

The acceleration array data A is the DDA
The data is transferred to and written to the registers 51, 53, 55, and 57 of the array 115P. The written acceleration array data A is added to the previous speed array data (or the initial speed array data) V (previous) stored in the registers 52, 54, 56, 58 by adders 61, 62, 63, 64. Is added. Then, the speed array data V resulting from the addition is written into the registers 52, 54, 56, 58.

Next, the speed array data V obtained in this way is saved as the previous speed array data V (previous) used for the next calculation, and this speed array data V is stored in the DDA array 115P. Registers 51, 53, 5
Write to 5, 57. And registers 52, 54, 5
The saved previous position array data (or initial position array data) P (previous) is set to 6, 58.

Then, the operation is performed by the adders 61, 62, 63 and 64, and the addition result is written in the registers 52, 54, 56 and 58 in the same manner as described above. Thus, the position array data P of the object in the frame is obtained from the registers 52, 54, 56, and 58. The obtained position array data P is saved as the previous speed array data P (previous) to be used for the next calculation, and is output as the position array data P of the frame.

In the next frame, the 1 / weight array data 1 / M of the object set by the game program and the force array data F generated from the user input and the environment at that time are again stored in the CPU main body. A multiplication operation is performed by 110 and a geometric operation unit 114 to obtain acceleration array data A, and the data is stored in the register 5 of the DDA array 115P.
1, 53, 55 and 57 are set. Then, the saved previous speed data V (previous) is added by the adders 61 to 64, and the addition result is added to the registers 52, 54, 5
Write to 6, 58. Hereinafter, by repeating the above operation in each frame, the position array data P of the object in each frame can be obtained.

Next, a case where a simple periodic motion is calculated using the DDA array 115P will be described with reference to FIG.
Also in this case, the portion surrounded by the dashed line is the DDA array 1
15P is a part for executing the operation. That is, in the case of this example, the DDA array 115P is used repeatedly to obtain the position array data from the acceleration array data.

That is, in this example, from the position array data P as output data of the DDA array 115P, the CPU main body 110 or the geometric operation unit 114 moves toward the position (-P) on the opposite side of the periodic motion. The acceleration array data A (-P) is obtained by calculation. Then, the acceleration array data A (-P) at the opposite position is transferred to the DDA array 115.
It is set in the registers 51, 53, 55, and 57 of P. The previous speed array data (or initial speed array data) is stored in the registers 52, 54, 5 of the DDA array 115P.
Set to 6, 58.

Then, the current speed array data, which is the addition result of the adders 61 to 64, is written into the registers 52, 54, 56, 58 of the DDA array 115P. The obtained speed array data V is saved as the previous speed array data V (previous) to be used in the next calculation, and the speed array data V is stored in the registers 51, 53 of the DDA array 115P.
Write to 55 and 57. And registers 52, 54,
The saved previous position array data (or initial position array data) P (previous) is set in 56 and 58.

Then, the obtained position array data P is
The internal bus 1 is sent to the PU main body 111 or the geometric operation unit 114.
10, the acceleration array data A (-P) for the reverse position is obtained as described above. Then, as described above, the acceleration array data A (−
P) is passed to the DDA array 115P as described above and set (negative feedback), and the above processing is repeated. Thereby, the periodic motion can be easily realized.

In this manner, by performing the above-described operation for each frame by using the physical operation unit 115, position array data can be obtained at high speed from acceleration array data relating to an object, and the drawing expression ability is improved.

In this case, since the CPU main body 111 only needs to calculate the acceleration array data, the burden is small and it contributes to high-speed processing. That is, when the above-described product-sum operation is executed by the CPU main body 111, the processing procedure is such that the physical calculation result is temporarily stored in a register and read out and the operation is executed. Becomes very large. However, in the case of this embodiment, when the CPU main body 111 sends a calculation result such as acceleration to the physical operation unit 115, the product-sum operation is automatically executed. Therefore, the load on the CPU main body 111 is greatly reduced.

The geometric operation unit 114 and the CPU body 11
Since the physical operation unit 115 can execute the physical operation using the data such as the acceleration calculated in step 1, it is possible to associate the coordinate operation with the coordinate conversion operation and other complicated motion operations.
Since the physical operation can be performed by the physical operation unit 115,
More natural and complicated movements can be realized by high-speed arithmetic processing.

Further, the coordinate transformation and the physical operation having a relation can be calculated simultaneously and in parallel, and the overhead of loading and reloading data to registers can be reduced. Then, as in the case of the coordinate conversion operation, by performing the physics operation using a dedicated operation unit, it is possible to obtain an image generation apparatus that realizes the physical phenomena in the three-dimensional world at high speed and with good balance.

In particular, in the case of a game machine, a complicated and visually natural image in which an image of an object is physically deformed with a collision is obtained by using the geometric operation unit 114 and the physical operation unit 115. Thus, processing in which movement and coordinate calculation are closely related can be easily realized, and the drawing function can be improved.

In the case where the physical calculation is executed in the CPU main body, various programming styles may come out and unification may not be achieved. And physical operation unit 1
15 is provided as dedicated hardware separately from the CPU main body, so that there is an advantage that the programming style is defined and unification can be achieved.

The above example is based on the CPU according to the present invention.
Is applied to a game machine.
It goes without saying that the PU can be used for various applications not limited to the image generation device.

[0068]

As described above, according to the present invention, a simple hardware consisting of a DDA array is provided in a CPU to execute a physical operation exclusively.
In conjunction with a geometric operation unit that performs coordinate conversion operation etc. exclusively,
It is possible to realize a CPU and an image generating device having a high drawing function.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of an internal configuration of a CPU according to the present invention.

FIG. 2 is a block diagram illustrating a configuration example of a game machine as an embodiment of an image generation device according to the present invention.

FIG. 3 is a diagram showing an example of an appearance of the game machine in the example of FIG. 2;

FIG. 4 is a diagram showing a configuration example of a main part of a physical operation unit inside a CPU according to the present invention;

5 is a diagram for explaining a physical calculation example using the physical operation unit in FIG. 4;

FIG. 6 is a diagram for explaining a physical calculation example using the physical operation unit of FIG. 4;

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 10 ... Main bus, 11 ... Main CPU, 110 ... Internal bus, 111 ... CPU main body, 112 ... Instruction cache memory, 113 ... Data cache memory, 114 ... Geometric operation part, 115 ... Physical operation part, 115P ... Physical operation processing part (DDA array) 51-58 ... register, 61-6
4 ... Adder

Claims (3)

[Claims] 1. A central processing unit main body, connected to the central processing unit main unit via an internal bus,
A first dedicated operation unit that exclusively executes a matrix operation function, connected to the central processing unit main unit via an internal bus,
A central processing unit comprising: a second dedicated operation unit that exclusively performs a product-sum operation. 2. A method according to claim 1, further comprising the steps of:
A central processing unit, which is connected to the main body of the central processing unit via an internal bus, and a dedicated processing unit for executing the product-sum operation exclusively. When generating image data of a moving image in a three-dimensional expression or a three-dimensional expression, a coordinate conversion operation is performed by the first dedicated operation unit, and a physical expression operation such as physical change or deformation is performed by the second dedicated operation unit. An image generation apparatus characterized in that the image generation is performed. 3. An apparatus for reading out data for drawing an image stored in a storage medium, wherein the read-out means generates an image by using the read-out data. 4. The image generation device according to claim 3, wherein the data of the image generation device is stored together with the data for drawing the image, thereby forming a game machine.