WO1999024901A1 - Computer processor and method for data streaming - Google Patents
Computer processor and method for data streaming Download PDFInfo
- Publication number
- WO1999024901A1 WO1999024901A1 PCT/US1998/023628 US9823628W WO9924901A1 WO 1999024901 A1 WO1999024901 A1 WO 1999024901A1 US 9823628 W US9823628 W US 9823628W WO 9924901 A1 WO9924901 A1 WO 9924901A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- register
- operand
- address
- registers
- processor
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 18
- 230000015654 memory Effects 0.000 claims abstract description 88
- 238000012545 processing Methods 0.000 claims description 11
- 238000004891 communication Methods 0.000 claims 1
- 230000007246 mechanism Effects 0.000 abstract description 2
- 230000004044 response Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000003491 array Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- ZLHLYESIHSHXGM-UHFFFAOYSA-N 4,6-dimethyl-1h-imidazo[1,2-a]purin-9-one Chemical compound N=1C(C)=CN(C2=O)C=1N(C)C1=C2NC=N1 ZLHLYESIHSHXGM-UHFFFAOYSA-N 0.000 description 1
- 230000001154 acute effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 239000000872 buffer Substances 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/30003—Arrangements for executing specific machine instructions
- G06F9/30007—Arrangements for executing specific machine instructions to perform operations on data operands
- G06F9/3001—Arithmetic instructions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/30098—Register arrangements
- G06F9/3012—Organisation of register space, e.g. banked or distributed register file
- G06F9/3013—Organisation of register space, e.g. banked or distributed register file according to data content, e.g. floating-point registers, address registers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/34—Addressing or accessing the instruction operand or the result ; Formation of operand address; Addressing modes
- G06F9/345—Addressing or accessing the instruction operand or the result ; Formation of operand address; Addressing modes of multiple operands or results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
- G06F9/38—Concurrent instruction execution, e.g. pipeline or look ahead
- G06F9/3885—Concurrent instruction execution, e.g. pipeline or look ahead using a plurality of independent parallel functional units
Definitions
- This invention pertains in general to microprocessors and in particular to a processor for performing media operations. More particularly, this invention pertains to a processor having a modified Harvard memory architecture and performing streaming ALU operations.
- DSP digital signal processing
- graphics processing graphics processing
- audio processing audio processing
- video processing video processing
- a typical algorithm will process a large number of elements while using each element only once.
- a typical DSP filter loop will process a large number of different input samples and apply a different operand from a large coefficient array to each input sample.
- a typical cycle of operation in such a processing environment requires 1) determining the addresses of the sample and coefficient; 2) retrieving the sample and coefficient from memory; 3) operating on the sample and coefficient; and 4) storing the result of the operation.
- RISC reduced instruction set computer
- a RISC architecture uses a low-complexity design adapted to handle a small set of simple instructions in order to obtain high-speed and high-performance.
- a RISC architecture uses fixed length instructions with very few instruction formats and fixed positions for certain operand fields, like register indices, within the instruction format. This architecture allows for low-complexity instruction decoders and control logic and this lower complexity can be leveraged into increased performance from other parts of the processor.
- ALU arithmetic logic unit
- SUBSTITUTE SHEET (RULE 2B) Unlike RISC processors, many complex instruction set computer (CISC) processors have instructions allowing simultaneous memory accesses from two different memory locations. Using these instructions, a programmer can retrieve both an input sample and a coefficient in a single instruction cycle. In addition, such processors allow simultaneous ALU operation. Thus, the number of instructions necessary to perform a DSP filter loop is reduced.
- CISC complex instruction set computer
- CISC processors generally have complex instruction decoding schemes in which several operands are implicit for the instruction and the load/store mechanism is coupled to the ALU operation encoding. Accordingly, few operations allow parallel load-stores. Even those operations, moreover, are limited to combinations to and from a small register file and support only a limited subset of parallel ALU operations. In addition, such processors require complex instruction decoders and, therefore, have lower clock speeds.
- processors supporting superscalar instruction scheduling dynamically extract instruction-level parallelism from the instruction stream and then group loads and stores with ALU operations. In this manner, the instructions can utilize parallel functional units in the processor.
- processors are highly complex in terms of design and size.
- VLIW very long instruction word
- the VLIW format explicitly encodes instruction-level parallelism into a very long instruction word.
- the VLIW typically has fields for frequently performed operations, such as ALU operations and memory accesses.
- VLIW format essentially demands that parallelism in the instruction stream be determined when the program is compiled. This demand results in an extremely complex programming model and, accordingly, a difficult program compilation. Thus, the gains made in processing efficiency by using the VLIW format are offset by the compile-time difficulties.
- the processor preferably has a modified Harvard memory architecture with first and second addressable memories, an address register file having first and second groups of address register sets, first and second address generation units respectively coupled to the first and second groups of address register sets, first and second stream registers, a general purpose register (GPR) file, and an arithmetic logic unit
- the address register file preferably includes eight register sets divided into two groups of four register sets each. Each register set contains a base register, a step register, and a modulo register.
- the first and second groups of address register sets respectively address the first and second addressable memories. When a memory access is made to an address register set, contents in that address register set are sent to the respective addressable memory as the address.
- the first and second address generation units respectively update the contents of selected address register sets in the first and second groups according to one of six addressing modes.
- the first and second addressable memories output data in response to receiving an address from an address register set. These data are respectively stored in the first and second stream registers.
- the GPR register file contains 32 GPRs.
- the ALU accepts two operands as inputs, the first of which is either the first stream register or a GPR and the second of which is either the second stream register or a GPR.
- the ALU output may be stored in the first or second addressable memory or a GPR.
- a stream instruction performs three actions in parallel: 1 ) it performs a specified ALU operation on the operands in the stream registers or the GPRs and stores the result in a GPR or memory; 2) it updates the stream registers by using addresses from the address register file to load new values; and 3) it updates the addresses in the address re "gtei*ster file.
- a stream instruction encodes an operand selection field and three operand fields.
- the operand selection field selects the operands for the ALU operation and specifies whether to store the result in the first or second addressable memories or a GPR.
- the first and second operand fields specify operands for use by an ALU operation performed by a subsequent stream instruction.
- the first operand can specify either a register set in the first group in the address register file and an addressing mode or a GPR.
- the second operand can specify either a register set in the second group and an addressing mode or a GPR. Five bits are needed to choose either a register set and addressing mode or a particular GPR. Accordingly, the bits for addressing a GPR are positionally the same as the bits for selecting a register set and an addressing mode.
- FIGURE 1 is a high-level block diagram illustrating a prior art Harvard memory architecture
- FIGURE 2 is a high-level block diagram illustrating the functional components of a processor supporting data streaming
- FIGURE 3 is a high-level block diagram illustrating an address register file
- FIGURE 4 is a chart illustrating an instruction word format used by the processor of FIG. 2;
- FIGURE 5 is a timing chart illustrating the steps performed by the processor of FIG. 2 when executing a stream instruction.
- FIG. 1 illustrates a high-level block diagram of a prior art Harvard Memory Architecture 100.
- the processor core logic 110 is coupled to two independent memory banks 1 12,114 via two independent sets of buses X,Y.
- the processor core 110 is coupled to an X memory 112 via an X bus having an address bus 116 and a data bus 118.
- the processor core 110 is coupled to a Y memory 1 14 via a Y bus having an address bus 120 and a data bus 122.
- a Harvard architecture holds data in one memory bank and instructions in the other. Accordingly, an architecture where data are held in two memory banks and instructions are held in a third is called a "modified Harvard architecture.”
- An advantage of the modified Harvard architecture is that two memory accesses can be performed in a single instruction cycle. This advantage is realized when the processor having the architecture performs media operations. In such operations, the processor typically operates on two arrays of data, one array of coefficients and one array of samples. If each array is stored in a different memory, then the processor can access both arrays of data in a single instruction cycle.
- FIG. 2 is high-level a block diagram illustrating certain functional components of a processor 200 constructed according to an embodiment of the present invention. Shown are an instruction memory 210, an instruction decoder 214, X and Y address register files 216, and a general purpose register (GPR) file 218. Also shown are X and Y address generation units (AGUs), a load/store unit 222, an X memory 224X, and a Y memory 224Y. In addition, FIG. 2 shows X and Y stream registers 226, X and Y multiplexers (MUXs) 228, an arithmetic logic unit (ALU) 230, and a GPR MUX 232.
- AGUs X and Y stream registers 226, X and Y multiplexers (MUXs) 228, an arithmetic logic unit (ALU) 230, and a GPR MUX 232.
- the instruction memory 210 is coupled to the instruction decoder 214 and holds instructions for the program being executed by the processor 200.
- the instruction memory 210 receives a program counter from the instruction decoder 214 that identifies an instruction in the memory 210. The identified instruction is transmitted from the instruction memory 210 to the instruction decoder 214.
- the instruction decoder 214 is coupled to the instruction memory 210, the X and Y address register files 216, and the GPR file 218. In addition, the instruction decoder 214 is coupled via control lines (shown as dashed lines in FIG. 2) to the X and Y address register files 216, the load/store unit 222, the X and Y MUXs 228, the ALU 230, and the GPR MUX 232.
- the instruction decoder decodes instructions received from the instruction memory 210 and controls the operation of the processor 200 in response thereto. For purposes of this discussion, assume that the received instruction is a streaming ALU instruction. As will be described in more detail below, the streaming ALU instruction specifies three operands: two sources and one destination.
- the source operands are selected from among the X and Y stream registers 226 and the GPR file 218.
- the destination operand is selected from either the X or Y memory 224 or the GPR file 218.
- the instruction decoder 214 While decoding the instruction, the instruction decoder 214 sends a first index from the first operand to the X address register file 216X, a second index from the second operand to the Y address register file 216Y, and all three operands to the GPR file 218.
- the instruction decoder 214 sends control signals to the AGUs 220 indicating address modes of the instruction and to the GPR file 218 indicating whether the GPR MUX 232 will write to the GPR file 218.
- the X and Y address register files 216 respectively receive the X and Y indices from the instruction decoder 214.
- the X and Y address register files 216 are each coupled to respective AGUs 220 and to the load/store unit 222.
- FIG. 3 illustrates a more detailed view of the address register file 216.
- the address register file 216 has eight address register sets A0-A7 divided into two groups of four register sets each A0-A3, A4-A7. The first group of four register sets forms the X register file 216X while the second group of four register sets forms the Y register file 216Y.
- Each address register set of which register set A0 is exemplary, has one 32-bit base register B0, one 16-bit step register SO, and one 16-bit modulo register M0. Since there are only four register sets in each group A0-A3, A4-A7, only a two-bit index is needed to select a particular register set in a group.
- the X and Y indices received from the instruction decoder 214 respectively select a register in the X and Y address register files 216.
- the contents of the selected X and Y address registers 216 are sent to the respective AGUs 220 and the contents of the selected X and Y base registers 216 are sent to the load /store unit 222.
- the X and Y AGUs 220 respectively receive the contents of the X and Y address registers and the address mode control signals from the instruction decoder 214 and have an output coupled to the inputs of the X and Y address register files 216.
- the AGUs 220 post-modify the selected base registers B0 according to the addressing mode and the step SO and modulo M0 registers.
- Modes supported by the AGUs 220 include: 1) auto post decrement; 2) auto post increment; 3) step post decrement; 4) step post increment; 5) base (no modification); and 6) bit reversed.
- the base register BO is modified according to the contents of a user processor status register (not shown).
- the base register BO is either incremented or decremented according to the contents of the step register SO.
- the modulo value in the modulo register MO qualifies the address calculation to give addressing for circular buffers. Since there are six address modes, three bits are needed to select a particular mode.
- the respective X and Y address register files 216 also send the contents of the selected base registers to the load/store unit 222.
- the load/store unit 222 is coupled to the X and Y address register files 216, the X and Y stream registers 226, the ALU 230, the GPR MUX 232, and the GPR file 218 and receives control signals from the instruction decoder 214.
- the load/store unit 222 is bi-directionally coupled to the X and Y memories 224.
- the couplings 223 between the load/store unit 222 and the memories 224 each comprise separate address and data busses like those shown in FIG. 1 and, thus, form a modified Harvard architecture.
- the load/store unit 222 loads data from and stores data to addressable memory locations within the processor 200, including the X and Y memories 224. For stream accesses, the load/store unit 222 loads data from the X and Y memories 224 to the X and Y stream registers 228, respectively. For normal accesses, the load/store unit 222 loads data from the memories to the GPR file 218 via the GPR MUX 232. If specified by a streaming instruction, the load/store unit 222 stores input data received from the ALU 230 in the X or Y memory 224. The load/store unit 222 also stores input data received from the GPR file 218 in memory in response to an explicit store instruction.
- the X register file 216X can reference only addresses in X memory 216X while the Y register file 216Y can reference only addresses in Y memory 216Y.
- the load/store unit 222 does not need to determine to which memory 224 an address received from the address register file 216 applies.
- the address received by the load/store unit 222 explicitly identifies the memory to which the address applies.
- the X and Y 224 memories are coupled to the load/store unit 222 and are addressable memories. In response to receiving a memory address and control signals from the load/store unit 222, the memories either store an input at the referenced memory address or output the contents of the referenced memory address.
- the X and Y stream registers 226 receive inputs from the load/store unit 222, output to the X and Y MUXs 228, and are at least 32 bits wide.
- the X and Y stream registers 226 store the data fetched by the load/store unit 222 from the X and Y memories 224, respectively, when streaming ALU 230 instructions are performed.
- the GPR file 218 has inputs coupled to the instruction decoder 214 and the GPR MUX 232 and outputs coupled to the X and Y MUXs 228 and the load/store unit 222. There are 32 GPRs in the GPR file 218 and each GPR is at least 32 bits wide.
- the GPRs 218 hold general purpose data used by the processor 200. Accordingly, the contents of the GPRs within the file 218 can be loaded from and stored to memory via the load/store unit 222 in response to explicit load and store instructions.
- the instruction decoder 214 When decoding a stream instruction, the instruction decoder 214 sends the three operand fields in the stream instruction to the GPR file 218.
- the first two operands specify the sources for the streaming ALU 230 operation. Since there are 32 GPRs in the file 218, five-bit indices in each of the first two operand fields specify particular GPRs in the GPR file 218.
- the contents of the GPR specified by the first operand are sent to the MUX 228X corresponding to the X register 226X and the contents of the GPR specified by the second operand are sent to the MUX 228 Y corresponding to the Y register 226 Y.
- the third operand along with the operand selection field described below, specifies the destination of the result of the ALU 230 operation. If this destination is a GPR, then the GPR file 218 receives the result from the ALU 230 via the GPR MUX 232 and stores it in the GPR identified by the third operand.
- the X and Y MUXs 228 respectively receive the contents of the X and Y stream registers 226 and the contents of the specified first and second GPRs.
- the outputs of the MUXs 228 are passed to the ALU 230.
- Each MUX receives control signals from the instruction decoder 214 specifying which input the MUX will output to the ALU 230.
- the ALU 230 receives inputs from the MUXs 228, receives control signals from the instruction decoder 214, and has an output coupled to the load/store unit 222 and the GPR MUX 232.
- the ALU 230 performs an arithmetic operation on its inputs and produces an output. The type of operation performed is specified by the control signals received from the instruction decoder 214.
- the GPR MUX 232 has an input coupled to the ALU 230 output and an input coupled to an output of the load/store unit 222, receives control signals from the instruction decoder 214, and has an output coupled to the GPR file 218. When so controlled by the instruction decoder 214, the GPR MUX 232 outputs a received input to the GPR file 218.
- FIG. 4 illustrates instruction encoding for a streaming ALU instruction of the type decoded by the instruction decoder 214.
- FIG. 4 there are eight instruction encoding formats and each instruction encoding format is identified by a three letter code that specifies the operands for that format. The three letters respectively specify the source of the first operand, the source of the second operand, and the destination for the results of the ALU 230 operation.
- the letter "X" specifies the X stream register 226X
- the letter "Y” specifies the Y stream register 226Y
- the letter "R" specifies a GPR.
- instruction format 414 is labeled "XYR", indicating that the source operands respectively come from the X and Y stream registers 226 and the destination is a GPR.
- the instruction format 400 is preferably 32 bits in length. Bits 14-10, labeled “A,” identify the first operand, bits 9-5, labeled “B,” identify the second operand, and bits 20-16, labeled “C,” identify the third operand. Op6 bits 2-0, labeled as “D,” specify the type of encoding (operands) used by the instruction format. Op 1-5 bits 27-21, 15, and 4-3, labeled as “E,” contain operation codes (opcodes) for the instruction format. Finally, COND bits 31-28, labeled as "F,” identify the conditions under which the instruction will execute.
- Field D is a three-bit operand select field that specifies from where the first and second operands of the ALU are read and to where the result of the ALU 230 operation is stored (operand 3).
- Table 1 indicates the possible operand combinations, the corresponding value of field D (Op6), and the corresponding reference numeral in FIG. 4.
- the first operand is selected from either a GPR or the X stream register 226X.
- the second operand in contrast, is selected from either a GPR or the Y stream register 226Y.
- the third operand can be stored in X memory 224X, Y memory 224Y, or a GPR.
- Fields A, B, and C respectively specify the first, second, and third operands.
- the first A and second B operands are used to update the inputs to the ALU 230 (the stream registers 226 and the GPR file 218) for use by a subsequent ALU 230 operation.
- the values specified by the first and second operands overwrite the values previously held in the stream registers 226.
- the third operand indicates where to store the result of the current ALU 230 operation.
- each of the operand fields A,B,C is five bits wide. Since there are only four register sets in each of the X and Y address register files 216, and since the first operand A can only select register sets in the X register file 216X and the second operand B can only select register sets in the Y register file 216Y, only two bits of encoding within the first A and second B operand fields are needed to select a particular register set within the address register file 216. In addition, three bits of encoding are needed to select among the possible addressing modes. Table 2 lists the possible addressing modes and the associated address mode bits.
- operand fields A,B,C the five bits used to select a GPR are overlapped with the two-bit sub field for the address register index and the three-bit subfield for the addressing mode. Thus, only a single five-bit field A,B,C is needed for each of the three operands.
- instruction format 410 is for an RRR instruction and, accordingly, all three operand are illustrated as five-bit fields A,B,C.
- Instruction format 418 is an RYX instruction and the five-bit fields for the second B and third C operands are shown as divided into three- and two-bit subfields.
- the programmer When the programmer wants to use the X stream register 226X as the first operand, the programmer puts the two-bit index for the register set and three bits specifying the addressing mode in the first operand field A. The programmer can use the Y stream register 226Y as the second operand in the same manner. If the address mode bits are set to 000, then the value in the X or Y stream register 226 is used without initiating a memory fetch or updating the address register 216.
- the opfields E describe the computation to be performed by the processor.
- the COND field F describes conditions to be checked before the computation is performed.
- FIG. 5 is a timing chart illustrating the steps performed by the processor 200 when executing a stream instruction like those illustrated in FIG. 4.
- time flows in the direction indicated by the arrow 500 on the left of the figure and the boxes to the right illustrate the order in which the indicated steps are performed by functional units of the processor 200. Since the functional units of the processor 200 operate in parallel, FIG. 5 should not be construed to indicate that time-aligned steps occur at the same time, but rather merely indicates the approximate sequence in which steps are performed.
- a stream instruction like that illustrated in FIG. 4 is received by the instruction decoder 214 TO.
- the instruction decoder 214 sends the index bits of the first and second operands to the X and Y address register files 216, respectively.
- the instruction decoder 214 sends all three of the operands to the GPR file 218.
- the instruction decoder decodes the Op6 field to determine the operands to the current ALU 230 operation and where the result should be stored.
- the following pseudocode explains the instruction decoding performed by the decoder 214 when decoding the Op6 field:
- MUX 228X selects Register 226X else
- MUX 228X selects GPR(first operand)
- MUX 228Y selects GPR(second operand).
- the address register files 216 pass the contents of the selected registers to the load/store unit 222 and cause the appropriate values to be loaded into the X and Y stream registers 226. Also, the AGUs 220 update the address registers selected by the index bits. At the same time, the ALU 230 operation indicated by the decoded instruction is performed on the previously read operands.
- GPR(third operand) ⁇ ALU 230 output.
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Advance Control (AREA)
- Executing Machine-Instructions (AREA)
Abstract
Description
Claims
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AU13108/99A AU1310899A (en) | 1997-11-07 | 1998-11-06 | Computer processor and method for data streaming |
JP2000519831A JP3866513B2 (en) | 1997-11-07 | 1998-11-06 | Computer processor and method for data streaming |
CA2284772A CA2284772C (en) | 1997-11-07 | 1998-11-06 | Computer processor and method for data streaming |
EP98956631A EP1027641B1 (en) | 1997-11-07 | 1998-11-06 | Computer processor and method for data streaming |
DE69824032T DE69824032T2 (en) | 1997-11-07 | 1998-11-06 | COMPUTER PROCESSOR AND METHOD FOR CONTINUOUS WRITEING AND READING OF DATA |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US08/966,904 | 1997-11-07 | ||
US08/966,904 US5958038A (en) | 1997-11-07 | 1997-11-07 | Computer processor with two addressable memories and two stream registers and method of data streaming of ALU operation |
Publications (1)
Publication Number | Publication Date |
---|---|
WO1999024901A1 true WO1999024901A1 (en) | 1999-05-20 |
Family
ID=25512033
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/US1998/023628 WO1999024901A1 (en) | 1997-11-07 | 1998-11-06 | Computer processor and method for data streaming |
Country Status (7)
Country | Link |
---|---|
US (1) | US5958038A (en) |
EP (1) | EP1027641B1 (en) |
JP (1) | JP3866513B2 (en) |
AU (1) | AU1310899A (en) |
CA (1) | CA2284772C (en) |
DE (1) | DE69824032T2 (en) |
WO (1) | WO1999024901A1 (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20230176863A1 (en) * | 2021-12-03 | 2023-06-08 | Taiwan Semiconductor Manufacturing Company, Ltd. | Memory interface |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2773234B1 (en) * | 1997-12-31 | 2003-07-25 | Sgs Thomson Microelectronics | DUAL ACCESS MEMORY FOR DIGITAL SIGNAL PROCESSOR |
US6212604B1 (en) * | 1998-12-03 | 2001-04-03 | Sun Microsystems, Inc. | Shared instruction cache for multiple processors |
US7254231B1 (en) * | 1999-10-14 | 2007-08-07 | Ati International Srl | Encryption/decryption instruction set enhancement |
US7039906B1 (en) | 2000-09-29 | 2006-05-02 | International Business Machines Corporation | Compiler for enabling multiple signed independent data elements per register |
US6834337B1 (en) | 2000-09-29 | 2004-12-21 | International Business Machines Corporation | System and method for enabling multiple signed independent data elements per register |
US6906978B2 (en) * | 2002-03-19 | 2005-06-14 | Intel Corporation | Flexible integrated memory |
US7533232B2 (en) * | 2003-11-19 | 2009-05-12 | Intel Corporation | Accessing data from different memory locations in the same cycle |
JP3760999B2 (en) * | 2004-06-15 | 2006-03-29 | セイコーエプソン株式会社 | Information processing apparatus, microcomputer and electronic device |
US20080177980A1 (en) * | 2007-01-24 | 2008-07-24 | Daniel Citron | Instruction set architecture with overlapping fields |
JP5708210B2 (en) * | 2010-06-17 | 2015-04-30 | 富士通株式会社 | Processor |
BR112013029505A2 (en) * | 2011-05-16 | 2020-01-07 | Romark Laboratories L.C. | PHARMACEUTICAL COMPOSITION AND USE OF A COMPOUND |
US10022189B2 (en) | 2013-12-16 | 2018-07-17 | Stryker Sustainability Solutions, Inc. | Apparatus and method for cleaning an instrument |
US10514915B2 (en) | 2015-04-24 | 2019-12-24 | Optimum Semiconductor Technologies Inc. | Computer processor with address register file |
US10713174B2 (en) * | 2016-12-20 | 2020-07-14 | Texas Instruments Incorporated | Streaming engine with early and late address and loop count registers to track architectural state |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4241397A (en) * | 1977-10-25 | 1980-12-23 | Digital Equipment Corporation | Central processor unit for executing instructions with a special operand specifier of indeterminate length |
US5450555A (en) * | 1990-06-29 | 1995-09-12 | Digital Equipment Corporation | Register logging in pipelined computer using register log queue of register content changes and base queue of register log queue pointers for respective instructions |
US5455955A (en) * | 1985-04-08 | 1995-10-03 | Hitachi, Ltd. | Data processing system with device for arranging instructions |
US5638524A (en) * | 1993-09-27 | 1997-06-10 | Hitachi America, Ltd. | Digital signal processor and method for executing DSP and RISC class instructions defining identical data processing or data transfer operations |
US5692207A (en) * | 1994-12-14 | 1997-11-25 | International Business Machines Corporation | Digital signal processing system with dual memory structures for performing simplex operations in parallel |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5805913A (en) * | 1993-11-30 | 1998-09-08 | Texas Instruments Incorporated | Arithmetic logic unit with conditional register source selection |
-
1997
- 1997-11-07 US US08/966,904 patent/US5958038A/en not_active Expired - Lifetime
-
1998
- 1998-11-06 CA CA2284772A patent/CA2284772C/en not_active Expired - Lifetime
- 1998-11-06 DE DE69824032T patent/DE69824032T2/en not_active Expired - Lifetime
- 1998-11-06 JP JP2000519831A patent/JP3866513B2/en not_active Expired - Lifetime
- 1998-11-06 EP EP98956631A patent/EP1027641B1/en not_active Expired - Lifetime
- 1998-11-06 WO PCT/US1998/023628 patent/WO1999024901A1/en active IP Right Grant
- 1998-11-06 AU AU13108/99A patent/AU1310899A/en not_active Abandoned
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4241397A (en) * | 1977-10-25 | 1980-12-23 | Digital Equipment Corporation | Central processor unit for executing instructions with a special operand specifier of indeterminate length |
US5455955A (en) * | 1985-04-08 | 1995-10-03 | Hitachi, Ltd. | Data processing system with device for arranging instructions |
US5450555A (en) * | 1990-06-29 | 1995-09-12 | Digital Equipment Corporation | Register logging in pipelined computer using register log queue of register content changes and base queue of register log queue pointers for respective instructions |
US5638524A (en) * | 1993-09-27 | 1997-06-10 | Hitachi America, Ltd. | Digital signal processor and method for executing DSP and RISC class instructions defining identical data processing or data transfer operations |
US5692207A (en) * | 1994-12-14 | 1997-11-25 | International Business Machines Corporation | Digital signal processing system with dual memory structures for performing simplex operations in parallel |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20230176863A1 (en) * | 2021-12-03 | 2023-06-08 | Taiwan Semiconductor Manufacturing Company, Ltd. | Memory interface |
Also Published As
Publication number | Publication date |
---|---|
DE69824032D1 (en) | 2004-06-24 |
JP2001523022A (en) | 2001-11-20 |
CA2284772A1 (en) | 1999-05-20 |
EP1027641A4 (en) | 2002-09-04 |
AU1310899A (en) | 1999-05-31 |
EP1027641B1 (en) | 2004-05-19 |
DE69824032T2 (en) | 2005-05-12 |
JP3866513B2 (en) | 2007-01-10 |
US5958038A (en) | 1999-09-28 |
EP1027641A1 (en) | 2000-08-16 |
CA2284772C (en) | 2011-11-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP0679991B1 (en) | Data processor for variable width operands | |
US6052766A (en) | Pointer register indirectly addressing a second register in the processor core of a digital processor | |
US7493474B1 (en) | Methods and apparatus for transforming, loading, and executing super-set instructions | |
US6467036B1 (en) | Methods and apparatus for dynamic very long instruction word sub-instruction selection for execution time parallelism in an indirect very long instruction word processor | |
EP1127316B1 (en) | Methods and apparatus for dynamically merging an array controller with an array processing element | |
US6101592A (en) | Methods and apparatus for scalable instruction set architecture with dynamic compact instructions | |
US5958038A (en) | Computer processor with two addressable memories and two stream registers and method of data streaming of ALU operation | |
JP2816248B2 (en) | Data processor | |
US20030154358A1 (en) | Apparatus and method for dispatching very long instruction word having variable length | |
US20020078320A1 (en) | Methods and apparatus for instruction addressing in indirect VLIW processors | |
JP2001515635A (en) | VLIW processor commands of different widths | |
EP1010065A1 (en) | Coprocessor data access control | |
US7546442B1 (en) | Fixed length memory to memory arithmetic and architecture for direct memory access using fixed length instructions | |
EP1680735B1 (en) | Apparatus and method that accomodate multiple instruction sets and multiple decode modes | |
US20130145127A1 (en) | Zero value prefixes for operands of differing bit-widths | |
US20040255095A1 (en) | Data access program instruction encoding | |
EP1499956B1 (en) | Method and apparatus for swapping the contents of address registers | |
US7134000B2 (en) | Methods and apparatus for instruction alignment including current instruction pointer logic responsive to instruction length information | |
US6704857B2 (en) | Methods and apparatus for loading a very long instruction word memory |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AL AM AT AU AZ BA BB BG BR BY CA CH CN CU CZ DE DK EE ES FI GB GD GE GH GM HR HU ID IL IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SL TJ TM TR TT UA UG UZ VN YU ZW |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GH GM KE LS MW SD SZ UG ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ CF CG CI CM GA GN GW ML MR NE SN TD TG |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 1998956631 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2284772 Country of ref document: CA Ref country code: CA Ref document number: 2284772 Kind code of ref document: A Format of ref document f/p: F |
|
ENP | Entry into the national phase |
Ref country code: JP Ref document number: 2000 519831 Kind code of ref document: A Format of ref document f/p: F |
|
REG | Reference to national code |
Ref country code: DE Ref legal event code: 8642 |
|
NENP | Non-entry into the national phase |
Ref country code: KR |
|
WWP | Wipo information: published in national office |
Ref document number: 1998956631 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 1998956631 Country of ref document: EP |