Data processor

Details Data processor Data processor

2001-01-18

2003-04-15

Pan, Daniel H. (Department: 2183)

Electrical computers and digital processing systems: processing

Instruction decoding

Decoding instruction to accommodate variable length...

C712S245000, C712S206000, C711S212000, C710S307000

Reexamination Certificate

active

06549999

ABSTRACT:

BACKGROUND OF THE INVENTION
The present invention relates to an improved data processor implemented as CPU or microprocessor, for example, and more particularly relates to a data processor adapted for use with an instruction set suitable for downsizing a program.
As semiconductor technologies and architecture of processors have been vigorously researched and developed over the past few years, the performance of programmable data processors of various types, termed “CPU's”, have also been tremendously improved. Depending on their specific applications, CPU's may be roughly classified into the following two types: general-purpose microprocessors; and built-in microcontrollers for use in numerous kinds of control units and consumer electronics appliances. As for CPU's of the first type, that is, general-purpose microprocessors, improvement of performance is a top priority. Therefore, for the sake of improving the performance of general-purpose microprocessors, a wide variety of techniques have heretofore been employed. In contrast, with regards to built-in microcontrollers, it is true that improvement of performance is one of most significant objects to be achieved. What is more important is, however, striking an appropriate balance among performance improvement, cost effectiveness and reduction in power consumption. Among other things, the cost effectiveness plays a key role in meeting a high demand in consumer electronics applications.
There are two tips for realizing a CPU at a lower cost: reducing the size of a CPU itself (i.e., reducing the area of a CPU core); and shrinking the size of a program (or the size of a ROM). In recent years, as the performance of a CPU has been improved, the number of functions implementable by a single CPU has increased and the size of an application program has further increased correspondingly. Under the circumstances such as these, the size of a ROM for storing such a large-sized program dominates over the area of a CPU core. Accordingly, the cost effectiveness of a CPU is greatly dependent on how small the size of a program, applicable to the CPU, can be.
The prior art, developed to solve this task, will be described. In accordance with this technique, the architecture of an instruction set for a general-purpose microprocessor is extended to reduce the size of a program.
FIG. 22
illustrates examples of MIPS architecture instruction format for a data processor in the pertinent prior art. specifically, FIG.
22
(
a
) illustrates a MIPS-II/III instruction format used for executing a register-to-register instruction where the length of a basic instruction word (hereinafter, simply referred to as a “basic instruction length”) is fixed at 32 bits. In contrast, FIG.
22
(
b
) illustrates MIPS
16
instruction formats where the basic instruction length is fixed at 16 bits.
In accordance with the MIPS architecture, 32 registers are provided. Thus, an instruction set in the MIPS-II/III format includes a plurality of register-addressing fields each composed of 5 bits. Also, since three operands are specified according to the MIPS-II/III format, this instruction set includes three register-addressing fields rs, rt and rd. The operation and functions of the instruction are defined using a 6-bit OP field, a 5-bit shamt field and a 6-bit func field. Accordingly, this instruction set has a fixed length of 32 bits in total.
In contrast, two types of instruction formats are definable for a register-to-register instruction included in an instruction set according to the MIPS16 format. In one of the two types of instruction formats, two 3-bit register-addressing fields rx and ry are provided to specify two operands and the operation and function of the instruction are defined using a 5-bit OP field and a 5-bit func field. In he other instruction format, three 3-bit register-addressing fields rx, ry and rz are provided to specify three operands and the operation and function of the instruction are defined using a 5-bit OP field and a 2-bit F field.
In accordance with the MIPS16 format shown in FIG.
22
(
b
), only 3 bits are available for each register-addressing field. Accordingly, not all the 32 registers included in the original MIPS-II/III format, but some of these registers can be accessed.
Any instruction in the MIPS16 instruction format can be replaced with an associated instruction in the MIPS-II/III instruction format. Such replacement of an instruction in the MIPS16 instruction format with a counterpart in the MIPS-II/III instruction format is called an “extension” of an instruction.
FIG. 23
is a block diagram illustrating a main part of a data processor for executing instructions in the MIPS
16
and MIPS-II/III formats. Hereinafter, the operation of this data processor will be described.
An instruction fetch section
300
is a block for fetching an instruction. Specifically, the instruction fetch section
300
fetches an instruction set in the MIPS16 instruction format with a fixed length of 16 bits or in the MIPS-II/III instruction format with a fixed length of 32 bits, and then outputs the fetched instruction set to an instruction extender
310
. The type of the instruction set, i.e., whether the instruction set is in the MIPS16 or MIPS-II/III instruction format, is always specified by a mode setting signal.
The instruction extender
310
is also controlled by the mode setting signal. If the input instruction set is in the MIPS16 instruction format, then the instruction extender
310
extends the instruction set in the MIPS16 instruction format into that in the MIPS-II/III instruction format. Alternatively, if the input instruction set is in the MIPS-II/III instruction format, then the instruction extender
310
outputs the instruction set as it is without performing the extension. It is controlled by the mode setting signal whether or not the extension should be performed. Accordingly, the instruction extender
310
cannot determine the necessity from the instructions themselves. Since the mode setting signal is a programmable signal, the mode of operations can be switched at a desired time.
An instruction decoder
320
is a block for decoding the instruction in the MIPS-II/III instruction format and thereby producing a control signal. The operation of the data processor is controlled by the control signal produced by the instruction decoder
320
.
The data processor having such a configuration can execute both a program described in the MIPS16 instruction format with a basic instruction length of 16 bits and a program described in the MIPS-II/III instruction format with a basic instruction length of 32 bits. Accordingly, if the code size should be prioritized, then programming is preferably carried out using the MIPS16 instruction format with a fixed length of 16 bits. On the other hand, if the performance should be respected first to access as large a number of register files as possible, then programming may be conducted using the MIPS-II/III instruction format. Thus, a program can be developed flexibly with a good balance struck between performance and code size. Nevertheless, it depends sometimes on the specifications of a particular system and sometimes on the size of a program which type of instruction formats should be used. For example, a certain type of format is used only when the size of a program reaches that of a task.
In order for a microprocessor to perform such an application (like signal processing) as requiring a large number of registers, the number of available registers should preferably be increased by adding some registers to preexistent ones such that the application can be performed at an even higher speed. In such a case, an instruction format, allowing the user to specify a larger number of registers than a conventional instruction format, may be produced and used instead of the conventional instruction format. However, the size of a resultant program considerably increases by the use of such an alternate instruction format.
Thus, the prior art may be modified in the following manner. A new instruction f

Affiliated with

Also associated with

Rating

Data processor does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Data processor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Data processor will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3115148