WO1996017292A1

WO1996017292A1 - Microprocessor with compare operation of composite operands

Info

Publication number: WO1996017292A1
Application number: PCT/US1995/015719
Authority: WO
Inventors: Alexander Peleg; Yaakov Yaari; Millind Mittal; Larry M. Mennemeier; Benny Eitan
Original assignee: Intel Corporation
Priority date: 1994-12-02
Filing date: 1995-12-01
Publication date: 1996-06-06
Also published as: IL116210A0; KR980700603A; EP0795154A4; AU4507396A; EP0795154A1

Abstract

A processor includes a decoder (202) coupled to receive a control signal (207). The control signal has a first source address (602), a second source address (603), a destination address (605), and an operation field (601). The first source address corresponds to a first location, and the second source address corresponds to a second location. The destination address corresponds to a third location. The operation field indicates that a type of packed data compare operation is to be performed. The processor includes a circuit coupled to the decoder for comparing a first packed data being stored at the first location with a second packed data being stored at the second location and for communicating a corresponding result packed data to the third location.

Description

MICROPROCESSOR WITH COMPARE OPERATION OF COMPOSITE OPERANDS

BACKGROUND OF THE INVENTION

FIELD OF INVENTION The present invention includes an apparatus and method of performing arithmetic operations using a single control signal to manipulate multiple data elements. The present invention allows execution of compare operations on packed data types.

2. DESCRIPTION OF RELATED ART

Today, most personal computer systems operate with one instruction to produce one result. Performance increases are achieved by increasing execution speed of instructions and the processor instruction complexity, and by performing multiple instructions in parallel; known as Complex Instruction Set Computer (CISC). Such processors as the Intel 80386™ microprocessor, available from Intel Corp. of Santa Clara, California, belong to the CISC category of processor.

Previous computer system architecture has been optimized to take advantage of the CISC concept. Such systems typically have data buses thirty-two bits wide. However, applications targeted at computer supported cooperation (CSC - the integration of teleconferencing with mixed media data manipulation), 2D/3D graphics, image processing, video compression/decompression, recognition algorithms and audio manipulation increase the need for improved performance. But, increasing the execution speed and complexity of instructions is only one solution.

One common aspect of these applications is that they often manipulate large amounts of data where only a few bits are important. That is, data whose relevant bits are represented in much fewer bits than the size of the data bus. For example, processors execute many operations on eight bit and sixteen bit data (e.g., pixel color components in a video image) but have much wider data busses and registers. Thus, a processor having a thirty-two bit data bus and registers, and executing one of these algorithms, can waste up to seventy-five percent of its data processing, carrying and storage capacity because only the first eight bits of data are important.

As such, what is desired is a processor that increases performance by more efficiently using the difference between the number of bits required to represent the data to be manipulated and the actual data carrying and storage capacity of the processor.

SUMMARY OF THE INVENTION A processor having an improved data comparison operation is described.

The processor includes a decoder being coupled to receive a control signal. The control signal has a first source address, a second source address, a destination address, and an operation field. The first source address corresponds to a first location. The second source address corresponds to a second location. The destination address corresponds to a third location. The operation field indicates that a type of packed data compare operation is to be performed. The processor further includes a circuit being coupled to the decoder. The circuit is for comparing a first packed data being stored at the first location with a second packed data being stored at the second location. The circuit is further for communicating a corresponding result packed data to the third location. Although a great deal of detail has been included in the description and figures, the invention is defined by the scope of the claims. Only limitations found in those claims apply to the invention.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention is illustrated by way of example, and not limitation, in the figures. Like references indicate similar elements.

Figure 1 illustrates an embodiment of the computer system using the methods and apparatus of the present invention. Figure 2 illustrates an embodiment of the processor of the present invention.

Figure 3 is a flow diagram illustrating the general steps used by the processor to manipulate data in the register file.

Figure 4a illustrates memory data types.

Figure 4b, Figure 4c and Figure 4d illustrate in-register integer data representations.

Figure 5a illustrates packed data-types.

Figure 5b, Figure 5c and Figure 5d illustrate in-register packed data representations.

Figure 6a illustrates a control signal format used in the computer system to indicate the use of packed data.

Figure 6b illustrates a second control signal format that can be used in the computer system to indicate the use of packed data or integer data.

Figure 7 illustrates one embodiment of a method followed by a processor when performing a compare operation on packed data.

Figure 8 illustrates a circuit capable of manipulating individual bytes of packed data.

DESCRIPTION OF THE PREFERRED EMBODIMENT

OVERVIEW OF ONE EMBODIMENT OF THE PRESENT INVENTION A processor having compare operations that operate on multiple data elements is described. In the following description, numerous specific details are set forth such as circuits, etc., in order to provide a thorough understanding of the present invention. In other instances, well-known structures and techniques have not been shown in detail in order not to unnecessarily obscure the present invention.

DEFINITIONS To provide a foundation for understanding the description of the embodiments of the present invention, the following definitions are provided. Bit X through Bit Y: defines a subfield of binary number. For example, bit six through bit zero of the byte 001110102 (shown in base two) represent the subfield 1110102- The '2' following a binary number indicates base 2. Therefore, 10002 equals 810. while F16 equals 15i0-

Rx: is a register. A register is any device capable of storing and providing data. Further functionality of a register is described below. A register is not necessarily part of the processor's package.

DEST is a data address, SRC1 is a data address, SRC2 is a data address.

Result: is the data to be stored in the register addressed by DEST. Sourcel : is the data stored in the register addressed by

SRC1. Source2: is the data stored in the register addressed by

SRC2.

COMPUTER SYSTFM

Referring to Figure 1 , a computer system upon which an embodiment of the present invention can be implemented is shown as computer system 100. Computer system 100 comprises a bus 101 , or other communications hardware and software, for communicating information, and a processor 109 coupled with bus 101 for processing information. Computer system 100 further comprises a random access memory (RAM) or other dynamic storage device (referred to as main memory 104), coupled to bus 101 for storing information and instructions to be executed by processor 109. Main memory 104 also may be used for storing temporary variables or other intermediate information during execution of instructions by processor 109. Computer system 100 also comprises a read only memory (ROM) 106, and/or other static storage device, coupled to bus 101 for storing static information and instructions for processor 109. Data storage device 107 is coupled to bus 101 for storing information and instructions.

Furthermore, a data storage device 107, such as a magnetic disk or optical disk, and its corresponding disk drive, can be coupled to computer system 100. Computer system 100 can also be coupled via bus 101 to a display device 121 for displaying information to a computer user. Display device 121 can include a frame buffer, specialized graphics rendering devices, a cathode ray tube (CRT), and/or a flat panel display. An alphanumeric input device 122, including alphanumeric and other keys, is typically coupled to bus 101 for communicating information and command selections to processor 109. Another type of user input device is cursor control 123, such as a mouse, a trackball, a pen, a touch screen, or cursor direction keys for communicating direction information and command selections to processor 109, and for controlling cursor movement on display device 121. This input device typically has two degrees of freedom in two axes, a first axis (e.g., x) and a second axis (e.g., y), which allows the device to specify positions in a plane. However, this invention should not be limited to input devices with only two degrees of freedom.

Another device which may be coupled to bus 101 is a hard copy device 124 which may be used for printing instructions, data, or other information on a medium such as paper, film, or similar types of media. Additionally, computer system 100 can be coupled to a device for sound recording, and/or playback 125, such as an audio digitizer coupled to a microphone for recording information. Further, the device may include a speaker which is coupled to a digital to analog (D/A) converter for playing back the digitized sounds.

Also, computer system 100 can be a terminal in a computer network (e.g., a LAN). Computer system 100 would then be a computer subsystem of a computer system including a number of networked devices. Computer system 100 optionally includes video digitizing device 126. Video digitizing device 126 can be used to capture video images that can be transmitted to others on the computer network.

Computer system 100 is useful for supporting computer supported cooperation (CSC - the integration of teleconferencing with mixed media data manipulation), 2D/3D graphics, image processing, video compression/decompression, recognition algorithms and audio manipulation.

PROCESSOR

Figure 2 illustrates a detailed diagram of processor 109. Processor 109 can be implemented on one or more substrates using any of a number of process technologies, such as, BiCMOS, CMOS, and NMOS.

Processor 109 comprises a decoder 202 for decoding control signals and data used by processor 109. Data can then be stored in register file 204 via internal bus 205. As a matter of clarity, the registers of an embodiment should not be limited in meaning to a particular type of circuit. Rather, a register of an embodiment need only be capable of storing and providing data, and performing the functions described herein.

Depending on the type of data, the data may be stored in integer registers 201 , registers 209, status registers 208, or instruction pointer register 211. Other registers can be included in the register file 204, for example, floating point registers. In one embodiment, integer registers 201 store thirty-two bit integer data. In one embodiment, registers 209 contains eight registers, Ro 212a through R7 212h. Each register in registers 209 is sixty-four bits in length. Ri 212a, R2 212b and R3 212c are examples of individual registers in registers 209. Thirty-two bits of a register in registers 209 can be moved into an integer register in integer registers 201. Similarly, an value in an integer register can be moved into thirty-two bits of a register in registers 209.

Status registers 208 indicate the status of processor 109. Instruction pointer register 211 stores the address of the next instruction to be executed. Integer registers 201 , registers 209, status registers 208, and instruction pointer register 211 all connect to internal bus 205. Any additional registers would also connect to the internal bus 205.

In another embodiment, some of these registers can be used for two different types of data. For example, registers 209 and integer registers 201 can be combined where each register can store either integer data or packed data. In another embodiment, registers 209 can be used as floating point registers. In this embodiment, packed data can be stored in registers 209 or floating point data. In one embodiment, the combined registers are sixty-four bits in length and integers are represented as sixty-four bits. In this embodiment, in storing packed data and integer data, the registers do not need to differentiate between the two data types.

Functional unit 203 performs the operations carried out by processor 109. Such operations may include shifts, addition, subtraction and multiplication, etc. Functional unit 203 connects to internal bus 205. Cache 206 is an optional element of processor 109 and can be used to cache data and/or control signals from, for example, main memory 104. Cache 206 is connected to decoder 202, and is connected to receive control signal 207.

Figure 3 illustrates the general operation of processor 109. That is, Figure 3 illustrates the steps followed by processor 109 while performing an operation on packed data, performing an operation on unpacked data, or performing some other operation. For example, such operations include a load operation to load a register in register file 204 with data from cache 206, main memory 104, read only memory (ROM) 106, or data storage device 107. In one embodiment of the present invention, processor 109 supports most of the instructions supported by the Intel 80486™, available from Intel Corporation of Santa Clara, California. In another embodiment of the present invention, processor 109 supports all the operations supported by the Intel 80486™, available from Intel Corporation of Santa Clara, California. In another embodiment of the present invention, processor 109 supports all the operations supported by the Pentium™ processor, the Intel 80486™ processor, the 80386™ processor, the Intel 80286™ processor, and the Intel 8086™ processor, all available from Intel Corporation of Santa Clara, California. In another embodiment of the present invention, processor 109 supports all the operations supported in the IA™ - Intel Architecture, as defined by Intel Corporation of Santa Clara, California (see Microprocessors. Intel Data Books volume 1 and volume 2, 1992 and 1993, available from Intel of Santa Clara, California). Generally, processor 109 can support the present instruction set for the Pentium™ processor, but can also be modified to incorporate future instructions, as well as those described herein. What is important is that general processor 109 can support previously used operations in addition to the operations described herein.

At step 301 , the decoder 202 receives a control signal 207 from either the cache 206 or bus 101. Decoder 202 decodes the control signal to determine the operations to be performed.

Decoder 202 accesses the register file 204, or a location in memory, at step 302. Registers in the register file 204, or memory locations in the memory, are accessed depending on the register address specified in the control signal 207. For example, for an operation on packed data, control signal 207 can include SRC1 , SRC2 and DEST register addresses. SRC1 is the address of the first source register. SRC2 is the address of the second source register. In some cases, the SRC2 address is optional as not all operations require two source addresses. If the SRC2 address is not required for an operation, then only the SRC1 address is used. DEST is the address of the destination register where the result data is stored. In one embodiment, SRC1 or SRC2 is also used as DEST. SRC1 , SRC2 and DEST are described more fully in relation to Figure 6a and Figure 6b. The data stored in the corresponding registers is referred to as Sourcel , Source2, and Result respectively. Each of these data is sixty- four bits in length.

In another embodiment of the present invention, any one, or all, of SRC1 , SRC2 and DEST, can define a memory location in the addressable memory space of processor 109. For example, SRC1 may identify a memory location in main memory 104 while SRC2 identifies a first register in integer registers 201 , and DEST identifies a second register in registers 209. For simplicity of the description herein, references are made to the accesses to the register file 204, however, these accesses could be made to memory instead.

In another embodiment of the present invention, the operation code only includes two addresses, SRC1 and SRC2. In this embodiment, the result of the operation is stored in the SRC1 or SRC2 register. That is SRC1 (or SRC2) is used as the DEST. This type of addressing is compatible with previous CISC instructions having only two addresses. This reduces the complexity in the decoder 202. Note, in this embodiment, if the data contained in the SRC1 register is not to be destroyed, then that data must first be copied into another register before the execution of the operation. The copying would require an additional instruction. To simplify the description herein, the three address addressing scheme will be described (i.e. SRC1 , SRC2, and DEST). However, it should be remembered that the control signal, in one embodiment, may only include SRC1 and SRC2, and that SRC1 (or SRC2) identifies the destination register.

Where the control signal requires an operation, at step 303, functional unit 203 will be enabled to perform this operation on accessed data from register file 204. Once the operation has been performed in functional unit 203, at step 304, the result is stored back into register file 204 according to requirements of control signal 207.

DATA AND STORAGE FORMATS Figure 4a illustrates some of the data formats as may be used in the computer system of Figure 1. These data formats are fixed point. Processor 109 can manipulate these data formats. Multimedia algorithms often use these data formats. A byte 401 contains eight bits of information. A word 402 contains sixteen bits of information, or two bytes. A doubleword 403 contains thirty-two bits of information, or four bytes. Thus, processor 109 executes control signals that may operate on any one of these memory data formats. ln the following description, references to bit, byte, word, and doubleword subfields are made. For example, bit six through bit zero of the byte 00111O1O2 (shown in base 2) represent the subfield 1110102.

Figure 4b through Figure 4d illustrate in-register representations used in one embodiment of the present invention. For example, unsigned byte in-register representation 410 can represent data stored in a register in integer registers 201. In one embodiment, a register, in integer registers 201 , is sixty-four bits in length. In another embodiment, a register, in integer registers 201 , is thirty-two bits in length. For the simplicity of the description, the following describes sixty-four bit integer registers, however, thirty-two bit integer registers can be used.

Unsigned byte in-register representation 410 illustrates processor 109 storing a byte 401 in integer registers 201 , the first eight bits, bit seven through bit zero, in that register are dedicated to the data byte 401. These bits are shown as {b}. To properly represent this byte, the remaining 56 bits must be zero. For an signed byte in-register representation 411 , integer registers 201 store the data in the first seven bits, bit six through bit zero, to be data. The seventh bit represents the sign bit, shown as an {s}. The remaining bit sixty-three through bit eight are the continuation of the sign for the byte.

Unsigned word in-register representation 412 is stored in one register of integer registers 201. Bit fifteen through bit zero contain an unsigned word 402. These bits are shown as {w}. To properly represent this word, the remaining bit sixty-three through bit sixteen must be zero. A signed word 402 is stored in bit fourteen through bit zero as shown in the signed word in-register representation 413. The remaining bit sixty-three through bit fifteen is the sign field.

A doubleword 403 can be stored as an unsigned doubleword in- register representation 414 or a signed doubleword in-register representation 415. Bit thirty-one through bit zero of an unsigned doubleword in-register representation 414 are the data. These bits are shown as {d}. To properly represent this unsigned doubleword, the remaining bit sixty-three through bit thirty-two must be zero. Integer registers 201 stores a signed doubleword in-register representation 415 in its bit thirty through bit zero; the remaining bit sixty-three through bit thirty-one are the sign field.

As indicated by the above Figure 4b through Figure 4d, storage of some data types in a sixty-four bit wide register is an inefficient method of storage. For example, for storage of an unsigned byte in-register representation 410 bit sixty-three through bit eight must be zero, while only bit seven through bit zero may contain non-zero bits. Thus, a processor storing a byte in a sixty-four bit register uses only 12.5% of the register's capacity. Similarly, only the first few bits of operations performed by functional unit 203 will be important.

Figure 5a illustrates the data formats for packed data. Three packed data formats are illustrated; packed byte 501 , packed word 502, and packed doubleword 503. Packed byte, in one embodiment of the present invention, is sixty-four bits long containing eight data elements. Each data element is one byte long. Generally, a data element is an individual piece of data that is stored in a single register (or memory location) with other data elements of the same length. In one embodiment of the present invention, the number of data elements stored in a register is sixty-four bits divided by the length in bits of a data element.

Packed word 502 is sixty-four bits long and contains four word 402 data elements. Each word 402 data element contains sixteen bits of information.

Packed doubleword 503 is sixty-four bits long and contains two doubleword 403 data elements. Each doubleword 403 data element contains thirty-two bits of information.

Figure 5b through Figure 5d illustrate the in-register packed data storage representation. Unsigned packed byte in-register representation 510 illustrates the storage of packed byte 501 in one of the registers Ro 212a through Rn 212af. Information for each byte data element is stored in bit seven through bit zero for byte zero, bit fifteen through bit eight for byte one, bit twenty-three through bit sixteen for byte two, bit thirty-one through bit twenty-four for byte three, bit thirty-nine through bit thirty-two for byte four, bit forty-seven through bit forty for byte five, bit fifty-five through bit forty-eight for byte six and bit sixty-three through bit fifty-six for byte seven. Thus, all available bits are used in the register. This storage arrangement increases the storage efficiency of the processor. As well, with eight data elements accessed, one operation can now be performed on eight data elements simultaneously. Signed packed byte in-register representation 511 is similarly stored in a register in registers 209. Note that only the eighth bit of every byte data element is the necessary sign bit; other bits may or may not be used to indicate sign.

Unsigned packed word in-register representation 512 illustrates how word three through word zero are stored in one register of registers 209. Bit fifteen through bit zero contain the data element information for word zero, bit thirty-one through bit sixteen contain the information for data element word one, bit forty-seven through bit thirty-two contain the information for data element word two and bit sixty-three through bit forty- eight contain the information for data element word three. Signed packed word in-register representation 513 is similar to the unsigned packed word in-register representation 512. Note that only the sixteenth bit of each word data element contains the necessary sign indicator.

Unsigned packed doubleword in-register representation 514 shows how registers 209 store two doubleword data elements. Doubleword zero is stored in bit thirty-one through bit zero of the register. Doubleword one is stored in bit sixty-three through bit thirty-two of the register. Signed packed doubleword in-register representation 515 is similar to unsigned packed doubleword in-register representation 514. Note that the necessary sign bit is the thirty-second bit of the doubleword data element.

As mentioned previously, registers 209 may be used for both packed data and integer data. In this embodiment of the present invention, the individual programming processor 109 may be required to track whether an addressed register, Ri 212a for example, is storing packed data or simple integer/fixed point data. In an alternative embodiment, processor 109 could track the type of data stored in individual registers of registers 209. This alternative embodiment could then generate errors if, for example, a packed addition operation were attempted on simple/fixed point integer data. CONTRQL SIGNAL FORMATS

The following describes one embodiment of control signal formats used by processor 109 to manipulate packed data. In one embodiment of the present invention, control signals are represented as thirty-two bits. Decoder 202 may receive control signal 207 from bus 101. In another embodiment, decoder 202 can also receive such control signals from cache 206.

Figure 6a illustrates a general format for a control signal operating on packed data. Operation field OP 601 , bit thirty-one through bit twenty-six, provides information about the operation to be performed by processor 109; for example, packed addition, packed subtraction, etc.. SRC1 602, bit twenty-five through twenty, provides the source register address of a register in registers 209. This source register contains the first packed data, Sourcel , to be used in the execution of the control signal. Similarly, SRC2 603, bit nineteen through bit fourteen, contains the address of a register in registers 209. This second source register contains the packed data, Source2, to be used during execution of the operation. DEST 605, bit five through bit zero, contains the address of a register in registers 209. This destination register will store the result packed data, Result, of the packed data operation.

Control bits SZ 610, bit twelve and bit thirteen, indicates the length of the data elements in the first and second packed data source registers. If SZ 610 equals 012. then the packed data is formatted as packed byte 501. If SZ 610 equals 102, then the packed data is formatted as packed word 502. SZ 610 equaling 002 or 112 is reserved, however, in another embodiment, one of these values could be used to indicate packed doubleword 503.

Control bit T 611 , bit eleven, indicates whether the operation is to be carried out with saturate mode. If T 611 equals one, then a saturating operation is performed. If T 611 equals zero, then a nonsaturating operation is performed. Saturating operations will be described later.

Control bit S 612, bit ten, indicates the use of a signed operation. If S 612 equals one, then a signed operation is performed. If S 612 equals zero, then an unsigned operation is performed. Figure 6b illustrates a second general format for a control signal operating on packed data. This format corresponds with the general integer opcode format described in the "Pentium™ Processor Family User's Manual," available from Intel Corporation, Literature Sales, P.O. Box 7641 , Mt. prospect, IL, 60056-7641. Note that OP 601 , SZ 610, T 611 , and S 612 are all combined into one large field. For some control signals, bits three through five are SRC1 602. In one embodiment, where there is a SRC1 602 address, then bits three through five also correspond to DEST 605. In an alternate embodiment, where there is a SRC2 603 address, then bits zero through two also correspond to DEST 605. For other control signals, like a packed shift immediate operation, bits three through five represent an extension to the opcode field. In one embodiment, this extension allows a programmer to include an immediate value with the control signal, such as a shift count value. In one embodiment, the immediate value follows the control signal. This is described in more detail in the "Pentium™ Processor Family User's Manual," in appendix F, pages F-1 through F-3. Bits zero through two represent SRC2 603. This general format allows register to register, memory to register, register by memory, register by register, register by immediate, register to memory addressing. Also, in one embodiment, this general format can support integer register to register, and register to integer register addressing.

DESCRIPTION OF SA TURA TE/UNSATURA TE As mentioned previously, T 611 indicates whether operations optionally saturate. Where the result of an operation, with saturate enabled, overflows or underflows the range of the data, the result will be clamped. Clamping means setting the result to a maximum or minimum value should a result exceed the range's maximum or minimum value. In the case of underflow, saturation clamps the result to the lowest value in the range and in the case of overflow, to the highest value. The allowable range for each data format is shown in Table 1. Data Format Minimum Value Maximum Value

Unsigned Byte 0 255

Signed Byte -128 127

Unsigned Word 0 65535

Signed Word -32768 32767

Unsigned Doubleword 0 26 -1

Signed Doubleword -263 263-1

Table 1

As mentioned above, T 611 indicates whether saturating operations are being performed. Therefore, using the unsigned byte data format, if an operation's result = 258 and saturation was enabled, then the result would be clamped to 255 before being stored into the operation's destination register. Similarly, if an operation's result = -32999 and processor 109 used signed word data format with saturation enabled, then the result would be clamped to -32768 before being stored into the operation's destination register.

COMPARE OPERATION

In one embodiment of the present invention, the performance of multimedia applications is improved by not only supporting a standard CISC instruction set (unpacked data operations), but by supporting a compare operation on packed data. The packed compare can be used to increase the speed of video compression, graphics and game technologies, voice and handwriting recognition and other digital/image signal processing algorithms.

In one embodiment of the present invention, the SRC1 602 register contains data (Sourcel) to be compared, the SRC2 603 register contains the data (Source2) to be compared against, and DEST 605 register will contain the result of the compare (Result). That is, Sourcel will have each data element independently compared by the each data element of Source2, according to an indicated relationship. In one embodiment of the present invention, the following compare relationships are supported: equal; signed greater than; signed greater than or equal; unsigned greater than; or unsigned greater than or equal. The relationship is tested in each pair of corresponding data elements. For example, Sourcel [7:0] is greater than Source2[7:0], with the result being Result[7:0]. If the result of the comparison satisfies the relationship, then, in one embodiment, the corresponding data element in Result is set to all ones. If the result of the comparison does not satisfy the relationship, then the corresponding data element in Result is set to all zeroes.

Figure 7 illustrates one embodiment of a method of performing a compare operation on packed data. This embodiment can be implemented in the processor 109 of Figure 2.

At step 701 , decoder 202 decodes control signal 207 received by processor 109. Thus, decoder 202 decodes: the operation code for the appropriate compare operation; SRC1 602, SRC2 603 and DEST 605 addresses in registers 209; saturate/unsaturate (not necessarily needed for compare operations), signed/unsigned, and length of the data elements in the packed data. As mentioned previously, SRC1 602 (or SRC2 603) can be used as DEST 605.

At step 702, via internal bus 205, decoder 202 accesses registers 209 in register file 204 given the SRC1 602 and SRC2 603 addresses. Registers 209 provides functional unit 203 with the packed data stored in the SRC1 602 register (Sourcel ), and the packed data stored in SRC2 603 register (Source2). That is, registers 209 communicate the packed data to functional unit 203 via internal bus 205.

At step 703, decoder 202 enables functional unit 203 to perform the appropriate packed compare operation. Decoder 202 further communicates, via internal bus 205, the size of data elements and the relationship for the compare operation.

At step 710, the size of the data element determines which step is to be executed next. If the size of the data elements is eight bits (packed byte 501 data), then functional unit 203 performs step 712. However, if the size of the data elements in the packed data is sixteen bits (packed word 502 data), then functional unit 203 performs step 714. In one embodiment, only eight bit and sixteen bit data element size packed compares are supported. However, in another embodiment, a thirty-two bit data element size packed compare is also supported (packed doubleword 503).

Assuming the size of the data elements is eight bits, then step 712 is executed. In step 712, the following is performed. Sourcel bits seven through zero are compared to Source2 bits seven through zero generating Result bits seven through zero. Sourcel bits fifteen through eight are compared to Source2 bits fifteen through eight generating Result bits fifteen through eight. Sourcel bits twenty-three through sixteen are compared to Source2 bits twenty-three through sixteen generating Result bits twenty-three through sixteen. Sourcel bits thirty- one through twenty-four are compared to Source2 bits thirty-one through twenty-four generating Result bits thirty-one through twenty-four. Sourcel bits thirty-nine through thirty-two are compared to Source2 bits thirty-nine through thirty-two generating Result bits thirty-nine through thirty-two. Sourcel bits forty-seven through forty are compared to Source2 bits forty- seven through forty generating Result forty-seven through forty. Sourcel bits fifty-five through forty-eight are compared to Source2 bits fifty-five through forty-eight generating Result bits fifty-five through forty-eight. Sourcel bits sixty-three through fifty-six are compared to Source2 bits generating Result bits sixty-three through fifty-six.

Assuming the size of the data elements is sixteen bits, then step 714 is executed. In step 714, the following is performed. Sourcel bits fifteen through zero are compared to Source2 bits fifteen through zero generating Result bits fifteen through zero. Sourcel bits thirty-one through sixteen are compared to Source2 bits thirty-one through sixteen generating Result bits thirty-one through sixteen. Sourcel bits forty-seven through thirty-two are compared to Source2 bits forty-seven through thirty-two generating Result bits forty-seven through thirty-two. Sourcel bits sixty-three through forty-eight are compared to Source2 bits sixty- three through forty-eight generating Result bits sixty-three through forty- eight. In one embodiment, the compares of step 712 are performed simultaneously. However, in another embodiment, these compares are performed serially. In another embodiment, some of these compares are performed simultaneously and some are performed serially. This discussion also applies to the compares of step 714 as well.

At step 720, the Result is stored in the DEST 605 register.

Table 2 illustrates the in-register representation of packed compare unsigned greater than operation. The first row of bits is the packed data representation of Sourcel . The second row of bits is the data representation of Source2. The third row of bits is the packed data representation of the Result. The number below each data element bit is the data element number. For example, Sourcel data element three is 100000002.

Table 2

Table 3 illustrates the in-register representation of packed compare signed greater than or equal to operation on packed byte data.

Table 3

PACKED DATA CIRCUITS

In one embodiment, the compare operation can occur on multiple data elements in the same number of clock cycles as a single compare operation on unpacked data. To achieve execution in the same number of clock cycles, parallelism is used. That is, registers are simultaneously instructed to perform the compare operation on the data elements. This is discussed in more detail below. Figure 8 illustrates one embodiment of a portion of a circuit that can perform a compare operation on packed data in the same number of clock cycles as a compare operation on unpacked data.

Figure 8 illustrates the use of a modified byte slice compare circuit, byte slice stagej 899. Each byte slice, except for the most significant data element byte slice, includes a compare unit and bit control. The most significant data element byte slice need only have a compare unit.

Compare unitj 811 and compare unitj+1 871 each allow eight bits from Sourcel to be compared to the corresponding eight bits from Source2. In one embodiment, each compare unit operates like a known eight bit compare circuit. Such a known eight bit compare circuit includes a byte slice circuit allowing the subtraction of Source2 from Sourcel . The results of the subtraction are processed to determine the results of the compare operation. In one embodiment, the results of the subtraction include an overflow information. This overflow information is tested to determine whether the result of the compare operation is true.

Each compare unit has a Sourcel input, a Source2 input, a control input, a next stage signal, a last stage signal, and a result output. Therefore, compare unitj 811 has Sourcel j 831 input, Source2j 833 input, control! 801 input, next stage; 813 signal, last stage; 812 input, and a result stored in result register) 851. Therefore, compare unitj+1 871 has Sourcel j+1 832 input, Source2j+ι 834 input, controlj+1 802 input, next stagβj+1 873 signal, last stagej+1 872 input, and a result stored in result registeη+1 852.

The Sourcel n input is typically an eight bit portion of Sourcel . The eight bits represents the smallest type of data element, one packed byte 501 data element. Source2 input is the corresponding eight bit portion of Source2. Operation control 800 transmits control signals to enable each compare unit to perform the required compare. The control signals are determined from the relationship for the compare (e.g. signed greater than) and the size of the data element (e.g. byte or word). The next stage signal is received from the bit control for that compare unit. Compare units are effectively combined by the bit control units when a larger than byte size data element is used. For example, when the word packed data is compared, the bit control unit between the first compare unit and the second compare unit will cause the two compare units to act as one sixteen bit compare unit. Similarly, the compare unit between the third and fourth compare units will cause these two compare units to act as one compare unit. This continues for the four packed word data elements.

Depending on the desired relationship and the values of Sourcel and Source2, the compare unit performs the compare by allowing result of the higher order compare unit to be propagated down to the lower order compare unit or vice versa. That is, each compare unit will provide the results of the compare using the information communicated by the bit control unitj 820. If double word packed data is used, then four compare units act together to form one thirty-two bit long compare unit for each data element. The result output of each compare unit represents the result of the compare operation on the portion of Sourcel and Source2 the compare unit is operating upon.

Bit controlj 820 is enabled from operation control 800 via packed data enablej 806. Bit controlj 820 controls next stagej 813 and last stagβj+1 872. Assume, for example, compare unitj 811 is responsible for the eight least significant bits of Sourcel and Source2, and compare unitj+1 871 is responsible for the next eight bits of Sourcel and Source2. If a compare on packed byte data is performed, bit controlj 820 will not allow the result information from compare unitj+ 871 to be communicated with the compare unitj 811 , and vice versa. However, if a compare on packed words is performed, then bit controlj 820 will allow the result (in one embodiment, an overflow) information from compare unitj 811 to be communicated to compare unitj+1 and result (in one embodiment, an overflow) information from compare unitj+1 871 to be communicated to compare unitj 811.

For example, in Table 4, a packed byte unsigned greater than compare is performed. Assume that compare unitj+1 871 operates on data element one, and compare unitj 811 operates on data element zero. Compare unitj+1 871 compares the most significant eight bits of a word and communicates the result information via last stagej₊ 872. Compare unitj 811 compares the least significant eight bits of the word and communicates the result information via next stagej 813. However operation control 800 will cause bit controlj 820 to stop the propagation of that result information, received from the last stagej+1 872 and next stagej 813, between the compare units.

Table 4

However, if a packed word unsigned greater than compare is performed, then the result of compare unitj+1 871 will be communicated to the compare unitj 811 , and vice versa. Table 5 illustrates this result. This type of communication would be allowed for packed doubleword compares as well.

Table 5

Each compare unit is optionally connected to a result register. The result register temporarily stores the result of the compare operation until the complete result, Result[63:0] 860, can be transmitted to the DEST 605 register.

For a complete sixty-four bit packed compare circuit, eight compare units and seven bit control units are used. Such a circuit can also be used to perform a compare on sixty-four bit unpacked data, thereby using the same circuit to perform the unpacked compare operation and the packed compare operation. Therefore, the packed compare operation can operate on multiple data elements. In prior art processors, to compare eight pairs of numbers, each represented by one byte, eight separate operations would be needed. Each operation would compare two bytes at a time, effectively wasting data lines and circuitry that are used for the bits that are higher than bit seven. Using the packed compare operation, the eight pairs can be compared by the processor in one operation and the data lines, in one embodiment, all carry relevant data. This leads to a higher performance computer system.

A microprocessor having an improved data comparison operation has been described.

Claims

THE CLAIMS What is claimed is:

1. A processor including :

a decoder being coupled to receive a control signal, said control signal having a first source address, a second source address, a destination address, and an operation field, said first source address corresponding to a first location, said second source address corresponding to a second location, said destination address corresponding to a third location, said operation field indicating that a type of packed data compare operation is to be performed; and

a circuit being coupled to said decoder, said circuit for comparing a first packed data being stored at said first location with a second packed data being stored at said second location, and for communicating a corresponding result packed data to said third location.

2. The processor of claim 1 wherein said first packed data includes a plurality of data elements, each data element of said plurality of data elements has a size, and said operation field further includes an indicator corresponding to said size.

3. The processor of claim 2 wherein said size is one of packed byte, packed word, and packed doubleword.

4. The processor of claim 2 where said first packed data is sixty-four bits.

5. The processor of claim 1 wherein said destination address is said first source address.

6. The processor of claim 1 wherein said operation field includes a sign indicator, said sign indicator for determining whether said comparing is to be performed signed or unsigned.

7. The processor of claim 1 wherein said type of packed data compare operation is one of compare for equal and compare for greater than.

8. The processor of claim 1 wherein said processor includes a register file, said register file includes a register, and wherein said second location corresponds to said register.

9. The processor of claim 8 wherein said first location corresponds to a memory location.

10. A method of comparing packed data in a processor, said processor having a decoder, a functional unit, a first register and a second register, said decoder being coupled to said functional unit, said first register and said second register, said method comprising the steps of:

said decoder decoding a control signal;

accessing a first packed data stored in said first register;

accessing a second packed data stored in said second register;

in response to said decoder decoding said control signal, said functional unit comparing said first packed data to said second packed data and generating a result packed data;

storing said result packed data in said first register.

11. The method of claim 10 wherein said control signal includes a type of compare indicator, said type of compare indicator indicating one compare operation from the group equal to, and greater than.

12. The method of claim 10 wherein said first packed data includes a plurality of data elements, each data element of said plurality of data elements being represented by a predetermined number of bits, said control signal includes a size indicator, said size indicator indicating said predetermined number of bits.

13. The method of claim 10 wherein said first register is sixty-four bits long, and said first packed data includes eight packed byte data elements.

14. The method of claim 10 wherein said comparing includes a step of subtracting said first packed data from said second packed data and testing a result of said subtracting.

15. The method of claim 10 wherein result packed data includes a plurality of result data elements, said comparing causing each data element of said plurality of data elements to be either all ones or all zeros.