JP2002082799A - Device and system for processing digital signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the processing speed of digital signals and also to improve availability of a memory. SOLUTION: This system comprises an external memory 1 and a DSP 2. The memory 1 stores the basic instructions and data. The DSP 2 is provided with an ALU 12 which generates the address of the memory 1, a bus control unit 13 which identifies the basic instructions and data read out of the memory 1 by means of the address generated by the ALU 12, the arithmetic means 17-19 which calculate the data, and an internal memory 23 which stores the fast instructions to control the operations of these ALU 12, unit 13 and means 17-19. The memory 23 is started by an external interrupt signal and reads continuously the fast instructions.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital signal processor such as a digital signal processor (DSP), and more particularly to a system in which the DSP processing has a function of processing a simple CPU instruction such as a branch instruction. The present invention relates to a digital signal processing device and a signal processing system suitable for application.

[0002]

2. Description of the Related Art Conventionally, a digital signal processing apparatus called a Harvard architecture has been known. This architecture is suitable for high-speed arithmetic processing such as multiplication, and includes two memories for storing coefficient data and multiplier data which are input to a multiply-adder and subjected to multiply-addition processing, and a DSP instruction storage. There is a total of three memories, one memory provided for each. This device has the advantage that the speed of digital signal processing can be increased because the coefficient data and the multiplication data can be simultaneously set in the multiplier / adder.

[0003]

However, in the above-mentioned conventional digital signal processing device, it is necessary to separately set address spaces for the three memories, so that the number of address lines becomes enormous and the chip area becomes large. There is a disadvantage that. Further, in this device, since it is impossible to mutually enter the memory spaces for storing the DSP instruction, the coefficient data, and the multiplier data, for example, when the memory space for the multiplier data is insufficient, the memory space for the coefficient data becomes empty. However, this cannot be used, and there is a problem that the memory use efficiency is low.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and an object of the present invention is to provide a digital signal processing device and a signal processing system capable of improving both the speed of digital signal processing and the efficiency of memory use. And

[0005]

A digital signal processing device according to the present invention is a digital signal processing device connected via an address bus and a data bus to an external memory storing at least basic instructions and data. External address generating means for generating an address for the external memory; and a basic instruction read from the external memory by providing the address generated by the external address generating means to the external memory via the address bus. Identification means for receiving the data and the data via the data bus, and identifying the basic instruction and the data, respectively, an arithmetic means for arithmetically processing the data identified by the identification means, and a basic element whose execution is controlled by an interrupt signal. A high-speed command consisting of a control signal with a word length longer than the command Data bus width stores is characterized in that said comprising an internal memory larger than the data bus width of the external memory, an address counter for successively reading speed instructions stored in the internal memory.

A signal processing system according to the present invention includes an external memory storing at least basic instructions and data, and a digital signal processing device connected to the external memory via an address bus and a data bus. An external address generating means for generating an address for the external memory;
By providing the address generated by the external address generation means to the external memory via the address bus, the basic instruction and data read from the external memory are accepted via the data bus, and the basic instruction and data are read. Identification means for respectively identifying, and arithmetic means for arithmetically processing the data identified by the identification means,
An internal memory in which a high-speed instruction composed of a control signal whose execution is controlled by an interrupt signal and whose word length is longer than the basic instruction is stored and the data bus width is larger than the data bus width of the external memory; And an address counter for continuously reading out the stored high-speed instructions.

According to the present invention, an internal memory for storing a high-speed instruction having a word length longer than that of a basic instruction is provided in the digital signal processing device, and the address counter performs continuous reading of the high-speed instruction. , And data such as multiplier data are commonly stored in an external memory. Then, the basic instruction is stored together with the data in the external memory, and the continuous reading operation of the high-speed instruction is started by the interrupt signal. Therefore, when the continuous read operation of the high-speed instruction is started by the interrupt signal, thereafter, the coefficient data and the multiplier data are sequentially read from the external memory by the high-speed instruction continuously read from the internal memory, and the necessary arithmetic operation is executed. Is done.

According to the present invention, separately from the access to the external memory, the high-speed instruction is continuously read from the internal memory and the data is accessed between the external memory. The processing such as reading and decoding of a high-speed command and incrementing of a program counter, which are necessary when reading the data, are all unnecessary, and the processing speed can be improved accordingly.

Further, according to the present invention, since the multiplier data and the coefficient data are all stored in the same external memory and are accessed via a common address bus and data bus, the number of buses is omitted and the chip area is reduced. Can be reduced, and the address space is shared by both data, so that the memory use efficiency is improved. Even if the parallel access of both data becomes impossible by sharing the bus, the access processing time of the high-speed instruction is shortened, so that the processing speed can be prevented from lowering.

According to the present invention, data is stored in an external memory,
Another major feature is that high-speed instructions are stored in separate memories such as an internal memory. Normally, when increasing the signal processing capability (processing amount per unit time) of the DSP, D
It is necessary to increase the bit length of the SP instruction. This is because more gates can be controlled. On the other hand, data does not require so many bits. Therefore, if a DSP instruction with a long bit length and data are stored in the same external memory, the memory space must be adjusted to the DSP instruction with a long data length. Area, and the memory usage efficiency is reduced. In this regard, according to the present invention, since the high-speed instruction is stored in the internal memory, the memory space of the external memory is not affected even if the bit length is long. Further, by improving the machine clock of the internal memory, the signal processing capability can be further improved.

The basic instruction may be a simple instruction at least capable of instructing the start and end of a continuous read operation of a high-speed instruction from the internal memory, so that the bit length does not need to be too long. Therefore, the data can be stored in the external memory together with the data without any trouble. If a simple instruction such as a branch instruction or a jump instruction is assigned to the basic instruction, it is possible to provide a high value-added apparatus and system in which a CPU function is added to a DSP function.

In this case, a basic instruction conversion means for converting a basic instruction given from the external memory into an executable instruction, and an executable instruction from the basic instruction conversion means in the basic instruction mode operated by the basic instruction are selected. In a high-speed instruction mode operating with a high-speed instruction, a selection means for selecting a high-speed instruction read from an internal memory may be further provided.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a signal processing system according to one embodiment of the present invention.

This system is constructed by connecting an external memory 1 and a DSP 2 as a digital signal processing device via an address bus 3 and a data bus 4. The external memory 1 includes a RAM or a ROM, and stores, for example, a 16-bit basic instruction, coefficient data, and multiplier data, as shown in FIG. Of course,
Alternatively, for example, a high-speed instruction described later may be divided and stored. The DSP 2 performs arithmetic processing on the coefficient data and the multiplier data stored in the external memory 1 in order to realize predetermined signal processing.

The DSP 2 is configured as follows.
The register group 11 is a group of registers for storing an address for the external memory 1 and data read from the external memory 1, and includes a working register (WR1, WR2), a base pointer (BP), an index X (IX), and an index Y (IY). ), A stack pointer (SP), a program counter (PC), and the like. BP, IX, IY, S
P, PC, etc. are registers for storing data for generating an address for the external memory 1, and these are stored in an arithmetic logic unit (ALU) 12 by an M-BUS or an N-BUS.
Is supplied via The ALU 12 increments these pointers or adds a predetermined value to cause an address jump. The operation result is stored in each register via the ALU-BUS. The pointer value supplied to the M-BUS is supplied to the address bus 3 as an address of the external memory 1. These pointers and ALU12
Constitute the external address generation means.

Basic commands and data input from the external memory 1 via the data bus 4 are supplied to a bus control unit 13 as identification means. The bus control unit 13 identifies the input data, and stores the basic instruction in the fetch register 14 when the basic instruction is input. When data is input, it is supplied to the C-BUS. The data supplied to the C-BUS is stored in the register group 11, the coefficient register 15, the multiplier register 16, and the like according to the control signal.
The data stored in the registers 15 and 16 is
7, is processed by a calculating means including an adder 18 and an accumulator 19, and is stored in another register via the C-BUS.

On the other hand, the basic instruction fetched by fetch register 14 is transferred to queue register 20 in the same cycle. The basic instruction (16 bits) stored in the queue register 20 is coded by the basic instruction coding circuit 21 in order to match the number of internal control signals, and the high-speed instruction (32-bit control signal)
Is supplied to one input terminal of the multiplexer 22. The other input terminal of the multiplexer 22 has a DSP
A 32-bit high-speed instruction from the internal memory 23 is provided. The multiplexer 22 controls the basic instruction coding circuit 21 in a basic instruction mode (CPU mode) described later.
In the high-speed instruction mode (DSP mode), a high-speed instruction from the DSP internal memory 23 is selected. For that purpose, the instruction for mode switching is included in the basic instruction set,
The queue register 20 decodes this and outputs the mode switching signal C
The PU / DSP controls the switching of the multiplexer 22 and outputs a start command to the DSP internal memory 23.

The DSP internal memory 23 stores a 32-bit high-speed instruction, for example, as shown in FIG. This high-speed instruction includes the multiplier 17, the ALU 12, and the register group 1
1, 15, 16 and the like can be controlled by one instruction, and include, for example, a high-efficiency multiplication instruction capable of executing multiplication processing by one instruction. The address counter 24 stores the queue register 2
From 0, a continuous address is generated from the start address given together with the start instruction, and the high-speed instruction is continuously read from the internal memory main body 31. DSP internal memory 23
Is composed of a RAM or a ROM, but if it is composed of a RAM, a high-speed instruction may be externally loaded when the system is started.

Next, the operation of the thus configured system will be described. FIG. 4A is a timing chart showing the operation in the basic instruction mode. First, the value of the program counter (PC) of the register group 11 is output to the address bus 3, and data is output from the external memory 1 to the data bus 4. When this data is identified by the bus control unit 13 as a basic instruction, the basic instruction is input to the basic instruction coding circuit 21 via the fetch register 14 and the queue register 15 and coded. As a result, a control signal required for executing the instruction controls opening and closing of the gate of each unit.

Assume that the basic instruction is a load instruction for data X (for example, multiplier data). In this case, the group of registers 1 is controlled by the coded control signal.
The content of the IX register 1 is output to the address bus 3 via the M-BUS, and at the same time, the ALU 12 is set to the increment mode, and IX + 1 is output to the ALU-BUS. At the next clock, the increment data is stored in the IX register via the ALU-BUS, and at the same time, the multiplier data read from the external memory 1 is stored in the register group 11 via the bus control unit 12 and the C-BUS.
Is stored in the working register WR1.

At the next clock, the program counter P
MB at the same time as C is incremented by ALU12
The external memory 1 is addressed via the US and address bus 3, and the next basic instruction is read. Assuming that the basic instruction is a load instruction of data Y (for example, coefficient data), the external memory 1 is determined by the contents of the IY register.
Is read out from the coefficient data.

As described above, in the basic instruction mode, addressing is performed by the program counter PC and is updated via the ALU 12, so that a conditional jump, a coefficient table jump, and the like can be easily realized by the basic instruction. Since the basic instructions do not code all control signals, the full function of the DSP cannot be exhibited. However, such access to the external memory 1 as a conditional jump or branch instruction is effective. It is.

FIG. 4B is a timing chart of the high-speed instruction mode. In order to improve the original signal processing capability of the DSP, an internal high-speed instruction is executed. The high-speed instruction is stored in the DSP internal memory 23 in advance. DS
When the P internal memory 23 is a RAM, a setup instruction is added to a basic instruction, and a high-speed instruction is loaded from the external memory 1 during system setup.

The start address of the DSP internal memory 23 is set via the queue register 20 by the basic instruction, and then the continuous reading operation of the internal high-speed instruction is started. Thereafter, the address counter 32 generates a continuous address from the start address, and the high-speed instruction continuously read out along with the address includes almost all the control signals so as to exhibit the sufficient performance of the DSP. be able to. Since the DSP internal memory 23 is a free-running type, an addressing cycle by the program counter PC is not required. For this reason, as shown in FIG.
No. 12 is dedicated to calculating the addresses of the multiplier data and the coefficient data, and can realize faster processing than in the basic instruction mode. The return from the internal instruction mode is realized by loading a basic instruction indicating a return instruction into the queue register 20 by a high-speed instruction.

It is conceivable that the same operation as the basic instruction is realized by an internal high-speed instruction. However, since the high-speed instruction tends to have a longer word length, a simple instruction such as the basic instruction is stored in the DSP internal memory. When stored in the memory 23, the memory access time increases and the cost performance decreases, and at the same time, the memory use efficiency also decreases. Since the capacity of the high-speed instruction itself, which requires the high-speed operation and is used for the digital signal processing, is not so large, by storing only the high-speed instruction in the internal memory 23, the capacity of the internal memory 23 can be suppressed. The flexibility of processing by the basic instructions is determined by the capacity of the external memory 1. However, since the external instructions 1 store the compressed basic instructions with emphasis on memory efficiency, cost performance is consequently low. improves.

The switching between the basic command mode and the high-speed command mode is applied to a search sequence of a CD or the like, automatic adjustment of gain and offset, and the like. For example, the example of FIG. 5 is an example in which a disturbance is appropriately applied to the coefficients of the digital filter in a search sequence or the like. In such a case where coefficient data is changed at a specific timing, a coefficient table jump is required. In a normal system, the coefficient table is switched by the CPU. However, according to this system, one DSP is executed by performing a table jump by a basic instruction.
Can be dealt with.

The switching between the two modes can be performed by an interrupt signal. For example, as shown in FIG. 6, the sampling period 1 /
fs, an interrupt signal IRQ is given to the CPU operation, and D
In the case of a system that executes SP processing, an interrupt signal IRQ / TR is sent to the bus control unit 13 in FIG.
Input P to load an interrupt instruction into the queue register 20. As a result, the mode is switched from the basic instruction mode to the high-speed instruction mode, and when the operation is completed, the mode returns to the basic instruction mode.

As described above, according to this system, basic instructions and data requiring flexibility are stored in the external memory 1, and DSP instructions requiring high speed are stored in the internal memory 32 as high-speed instructions. Thereby, both high-speed processing and memory use efficiency can be improved.

[0029]

As described above, according to the present invention,
Since the multiplier data and coefficient data are all stored in the same external memory and are accessed via a common address bus and data bus, the number of buses can be omitted to reduce the chip area, and both data can be used. Since the address space is shared, the efficiency of memory use is improved. Even if parallel access of both data becomes impossible by sharing the bus, high-speed instructions are read continuously from the internal memory, and the processing time of high-speed instructions is reduced, resulting in lower processing speed. Can be prevented. Since the high-speed instruction is stored in the internal memory, no matter how long the bit length is, it does not affect the memory space of the external memory. Further, if a simple instruction such as a branch instruction or a jump instruction is assigned to the basic instruction, it is possible to provide a high value-added device and system in which the CPU function is added to the DSP function.

[Brief description of the drawings]

FIG. 1 is a block diagram of a signal processing system according to an embodiment of the present invention.

FIG. 2 is a diagram showing contents of an external memory in the system.

FIG. 3 is a diagram showing contents of an internal memory in the system.

FIG. 4 is a diagram illustrating an application example of the system.

FIG. 5 is a diagram illustrating an application example of the system.

[Explanation of symbols]

1. External memory, 2. Digital signal processor (DS)
P), 3 ... address bus, 4 ... data bus, 11 ... register group, 12 ... arithmetic logic unit (ALU), 13 ... bus control unit, 14 ... fetch register, 1
5 ... coefficient register, 16 ... multiplier register, 17 ... multiplier, 18 ... adder, 19 ... accumulator, 20 ... queue register, 21 ... basic instruction coding circuit, 22 ...
Multiplexer, 23 ... DSP internal memory.

────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] July 16, 2001 (2001.7.1)
6)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] Brief description of drawings

[Correction method] Change

[Correction contents]

[Brief description of the drawings]

FIG. 1 is a block diagram of a signal processing system according to an embodiment of the present invention.

FIG. 2 is a diagram showing contents of an external memory in the system.

FIG. 3 is a diagram showing contents of an internal memory in the system.

FIG. 4 is a timing chart for explaining the operation of the system.

FIG. 5 is a diagram illustrating an application example of the system.

FIG. 6 is a diagram illustrating an application example of the system.

[Description of Signs] 1 ... External memory, 2 ... Digital signal processor (DS)
P), 3 ... address bus, 4 ... data bus, 11 ... register group, 12 ... arithmetic logic unit (ALU), 13 ... bus control unit, 14 ... fetch register, 1
5 ... coefficient register, 16 ... multiplier register, 17 ... multiplier, 18 ... adder, 19 ... accumulator, 20 ... queue register, 21 ... basic instruction coding circuit, 22 ...
Multiplexer, 23 ... DSP internal memory.

Claims (4)

[Claims] 1. A digital signal processing device connected via an address bus and a data bus to an external memory storing at least a basic instruction and data, wherein an external address generating means generates an address for the external memory. Providing the address generated by the external address generating means to the external memory via the address bus, receiving the basic instruction and data read from the external memory via the data bus, Identification means for identifying data and data; arithmetic means for performing arithmetic processing on the data identified by the identification means; and a high-speed instruction comprising a control signal having a word length longer than a basic instruction whose execution is controlled by an interrupt signal. The data is stored internally and the data bus width is An internal memory greater than the bus width, the digital signal processing apparatus characterized by comprising an address counter for reading the high-speed instructions stored in the internal memory continuously. 2. The high-speed instruction stored in the internal memory includes a high-efficiency multiplication instruction for executing a multiplication process at a higher speed than when controlled by the basic instruction. Digital signal processor. 3. A basic instruction conversion means for converting a basic instruction given from the external memory into an executable instruction, and an executable instruction from the basic instruction conversion means in a basic instruction mode operated by the basic instruction. And select
3. The digital signal processing device according to claim 1, further comprising a selection unit that selects a high-speed instruction read from the internal memory in a high-speed instruction mode operating with the high-speed instruction. 4. An external memory in which at least basic instructions and data are stored, and a digital signal processing device connected to the external memory via an address bus and a data bus, wherein the digital signal processing device External address generating means for generating an address for the external memory; basic instructions and data read from the external memory by providing the address generated by the external address generating means to the external memory via the address bus Through the data bus, and identifying means for respectively identifying the basic instruction and data; arithmetic means for performing arithmetic processing on the data identified by the identifying means; A high-speed command consisting of a control signal with a long word length is internally recorded. Signal processing comprising: an internal memory having a data bus width larger than the data bus width of the external memory; and an address counter for continuously reading high-speed instructions stored in the internal memory. system.