WO2000031658A1 - Processeur et dispositif de traitement d'image - Google Patents
Processeur et dispositif de traitement d'image Download PDFInfo
- Publication number
- WO2000031658A1 WO2000031658A1 PCT/JP1999/006591 JP9906591W WO0031658A1 WO 2000031658 A1 WO2000031658 A1 WO 2000031658A1 JP 9906591 W JP9906591 W JP 9906591W WO 0031658 A1 WO0031658 A1 WO 0031658A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- circuit
- data
- arithmetic
- dedicated
- pipeline
- Prior art date
Links
- 238000012545 processing Methods 0.000 title claims abstract description 72
- 230000006870 function Effects 0.000 claims description 29
- 230000004913 activation Effects 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 description 21
- 238000010586 diagram Methods 0.000 description 19
- 238000013139 quantization Methods 0.000 description 17
- 230000000694 effects Effects 0.000 description 11
- 238000000034 method Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000012546 transfer Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 229910052710 silicon Inorganic materials 0.000 description 1
- 239000010703 silicon Substances 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
- G06F9/38—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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
-
- 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
- G06F9/30036—Instructions to perform operations on packed data, e.g. vector, tile or matrix operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/20—Processor architectures; Processor configuration, e.g. pipelining
Definitions
- the present invention relates to an arithmetic device that executes multimedia signal processing at high speed and an image processing device using the same.
- a conventional program-controlled processor could achieve higher performance by implementing vector instructions.
- the conventional arithmetic unit shown in FIG. 14 includes a program control circuit ⁇ 401 that decodes a vector instruction and outputs a first start signal and a second start signal, and the first start signal
- a first address generator that outputs a first address according to the first address, a first memory that outputs first data based on the first address,
- a pipeline arithmetic circuit 144 for performing a pipeline operation based on the data of 1; a second address generator 144 for outputting a second address in response to the second start signal;
- a second data memory 1406 for storing the operation result by the pipeline operation circuit 144 based on the second address.
- the arithmetic unit when the vector instruction is decoded by the program control circuit 1401, the arithmetic unit outputs a first start signal from the program control circuit 1401, With this first start signal, the Nth I address generator 1402 starts generating N addresses.
- the first data memory 1443 to which the N addresses are inputted supplies N data to the pipeline operation circuit 144.
- the pipeline calculation circuit 1444 receives the N data (ft feed and executes the pipeline calculation process.
- the program control circuit 1441 outputs the second start signal in accordance with the output timing of the first processing data from the pipeline processing circuit 144, In response to the second start signal, the second address generators 144 to 5 output ⁇ addresses to the second data memory 144.
- the second data memory 1406 sequentially stores the operation results output from the pipeline operation circuit 144.
- the first end signal and the second end signal are program-controlled, respectively. Output to circuit 1401, which ends the vector instruction.
- the general-purpose pipeline arithmetic circuit may have insufficient computational performance.
- a specific high-load operation is processed by a dedicated pipeline operation circuit (for example, a DCT (Discrete Cosine Transform) operation circuit), and the other processing is performed by a general-purpose operation lEj path.
- the computational performance is improved by the data structure to ensure real-time performance.
- the timing design in the program control circuit is unique to the dedicated pipeline arithmetic circuit. In other words, the application pipeline There was a problem that it became unique to one silicon.
- the present invention has been made in view of such a problem, and is configured so that a general-purpose arithmetic circuit and a special-purpose arithmetic circuit are separated so that a change for each application of a dedicated arithmetic circuit lEj does not affect a general-purpose arithmetic circuit. This makes it possible to realize an arithmetic device that can be applied to various applications and an image processing device using the same.
- An arithmetic device (Claim 1) includes a general-purpose arithmetic circuit and a dedicated arithmetic circuit, wherein the general-purpose arithmetic circuit implements a plurality of vector instructions, An arithmetic device that executes a pipeline operation based on the vector instruction together with an arithmetic circuit, wherein the general-purpose arithmetic circuit includes: a dedicated pipeline arithmetic circuit selection signal for notifying the arithmetic operation of the dedicated arithmetic circuit; A plurality of operation results in the operation circuit, and a general-purpose operation circuit output data enable signal for notifying an output timing of the plurality of operation results, to the dedicated operation circuit.
- a data selection circuit for arbitrarily selecting one dedicated operation result from the results, outputting the arbitrarily selected dedicated operation result to the general-purpose operation circuit as the plurality of dedicated operation results, and the plurality of dedicated pipelines Receiving the dedicated pipeline stage number notification signal output from each of the arithmetic circuits, the dedicated pipeline arithmetic circuit selection signal of the general-purpose arithmetic circuit, and the general-purpose arithmetic circuit output data enable signal; And a control circuit for outputting an output data enable signal to the general-purpose arithmetic circuit. That.
- the arithmetic device according to the present invention (Claim 2) has an effect of being able to realize an arithmetic device applicable to various applications.
- the arithmetic device according to the present invention has a general-purpose arithmetic circuit and a dedicated arithmetic circuit.
- the general-purpose arithmetic circuit is an arithmetic device that implements a plurality of vector instructions, and executes a pipeline operation based on the vector instruction together with the special-purpose arithmetic circuit.
- Start signal, second start signal, Program control that outputs the first arithmetic circuit selection signal, the second arithmetic circuit selection signal, the dedicated pipeline arithmetic circuit selection signal, and the general-purpose arithmetic circuit output data enable signal, and receives the dedicated arithmetic circuit output data enable signal
- a first address generator that continuously outputs a first address of M ⁇ based on the first start signal from the program control circuit; and a first address.
- a first data memory that outputs M pieces of first data based on the first address from the generator; and a first operation circuit selection signal according to the first operation path selection signal from the program control circuit.
- a first pipeline operation circuit that executes a pipeline operation on the first data from the data memory and sequentially outputs M pieces of first operation results; and A second pipeline operation circuit that executes a pipeline operation on the second operation result from the dedicated operation circuit in accordance with the second operation circuit selection signal and sequentially outputs M third operation results
- a second address generator that continuously outputs M second addresses based on the second activation signal from the program control circuit; and A second data memory that stores M third operation results from the second pipeline operation circuit based on the second address of A pipeline stage number notification signal is output, a pipeline operation is performed on the first operation result from the first pipeline operation circuit in the general-purpose operation circuit, and a plurality of M dedicated pipeline operations are performed.
- the dedicated arithmetic circuit output data enable signal is transmitted to the general purpose arithmetic circuit.
- the general-purpose arithmetic circuit and the special-purpose arithmetic circuit are separated from each other, and the general-purpose arithmetic circuit is designed so that a change in each application of the special-purpose arithmetic circuit does not affect the pipeline operation in the general-purpose arithmetic circuit.
- the dedicated arithmetic circuit output data enable which is information specific to the dedicated arithmetic circuit required for timing control in the program control circuit, is notified from the dedicated arithmetic circuit to the general-purpose arithmetic circuit, and the program control circuit in the general-purpose arithmetic circuit notifies the notification information.
- the output timing of the pipeline arithmetic circuit is controlled based on the dedicated arithmetic circuit output data enable signal. That is, the program control circuit in the general-purpose arithmetic circuit asserts a first activation signal after decoding the vector instruction, and asserts the first activation signal, and thereafter, determines the number of pipeline stages of the first pipeline arithmetic circuit. The output timing of the first first calculation result from the first piezoelectric line calculation circuit is detected based on At the same time, the program control circuit asserts a general-purpose arithmetic circuit output data enable, negates the first activation signal M cycles after assertion of the first activation signal, and negates the first activation signal.
- the output timing of the M-th first calculation result from the first pipeline calculation circuit is detected based on the number of pipeline stages of the first pipeline performance circuit, and General-purpose operation Negative iHi road output data enable signal.
- the control circuit in the dedicated operation circuit after the general-purpose operation circuit output data enable signal is asserted, based on the n-th pipeline stage number notification signal selected in accordance with the dedicated pipeline operation circuit selection signal, and Detects the output timing of the first n-th dedicated operation result from the pipeline performance circuit.
- the control circuit asserts the dedicated arithmetic circuit output data enable signal, and after the general purpose arithmetic circuit output data enable signal is negated, the n-th selected by the dedicated pipeline arithmetic circuit select signal.
- the dedicated arithmetic circuit When the output timing of the M-th dedicated arithmetic operation result from the n-th dedicated pipeline arithmetic circuit is detected based on the number of pipeline stages communication g of the pipeline stage, the dedicated arithmetic circuit output data Negative enable signal. And The self-program control circuit, after assertion of the dedicated operation circuit output data enable signal, sets the third pipeline operation circuit from the second pipeline operation circuit based on the number of pipeline stages of the second pipeline operation circuit.
- the arithmetic unit of the present invention can mount any dedicated pipeline arithmetic IHJ path suitable for each application without changing the program control circuit, and as a result, can be applied to various applications. There is an effect that it is possible to realize an arithmetic device that can perform the calculation.
- the arithmetic device (claim 3) is the arithmetic device according to claim 2, wherein the first pipeline arithmetic circuit in the general-purpose arithmetic circuit is configured to receive a signal from the program control circuit.
- a first register that receives the first data from the first data memory and outputs a second data based on the first arithmetic circuit selection signal;
- a multiplier that outputs a result of the multiplication as fourth data
- a third register that receives fourth data from the multiplier as input, and outputs fifth data, and stores sixth data that is stored in advance.
- the fourth register to output and the previous An arithmetic operation unit that receives input of the fifth data from the third register and the sixth data from the fourth register, and outputs an arithmetic result thereof as seventh data;
- a fifth register that receives the seventh data from the arithmetic unit and outputs a first arithmetic result that is an output of the first pipeline arithmetic circuit, and a second register in the general-purpose arithmetic circuit.
- the pipeline operation circuit receives a second operation result from the dedicated operation circuit as an input based on a second operation circuit selection signal from the program control circuit, and outputs the output from the second pipeline operation circuit.
- a sixth dedicated register for outputting a third calculation result serving as a power supply wherein the specific dedicated pipeline calculation circuit in the dedicated calculation circuit reads the first calculation result from the first pipeline calculation circuit.
- an ID is output that performs a one-dimensional inverse discrete cosine transform and outputs a dedicated operation result that is the output of this dedicated pipeline operation circuit (Inversion Di screte and os inc i ransform It is characterized by the following.
- the inverse quantization operation is performed by the first pipeline operation circuit in the general-purpose operation circuit, and the inverse DCT operation is performed by the dedicated pipeline operation circuit in the dedicated operation circuit.
- This has the effect that the inverse quantization and the inverse DCT operation can be continuously performed in a pipeline operation.
- the arithmetic device (claim 4) is the arithmetic device according to claim 2, wherein the first pipeline arithmetic circuit in the general-purpose arithmetic circuit is the program Based on a first operation circuit selection signal from the control circuit, the first data from the first data memory is input, and the first operation result to be output from the first pipeline operation circuit is obtained.
- a first register as an output, wherein the second pipeline arithmetic circuit in the general-purpose arithmetic circuit includes a second pipeline arithmetic circuit based on a second arithmetic circuit selection signal from the program control circuit.
- a second register that receives the operation result of step 2 as an input and outputs second data, a third register that outputs third data stored in advance, and the second register from the second register.
- 2 data and the third An arithmetic operation unit that receives input of the third data from the register and outputs these arithmetic results as fourth data, and an input of the fourth data from the arithmetic operation unit, the fifth data
- a fifth register for outputting sixth data stored in advance, the fifth data from the fourth register and the sixth data from the fifth register.
- a multiplier that receives the data of the second and outputs the result of the multiplication as seventh data, and an output of the second pipeline arithmetic circuit that receives the seventh data from the multiplier as an input.
- a sixth register for outputting a third operation result
- the specific dedicated pipeline arithmetic circuit receives the first operation result from the first pipeline arithmetic circuit of the general-purpose arithmetic circuit, performs a one-dimensional discrete cosine transform, and performs a one-dimensional discrete cosine transform. It is characterized by including a DCT (Discrete Cosine Transform) calculator for outputting the second dedicated arithmetic output as an output.
- DCT Discrete Cosine Transform
- the quantization operation is performed by the second pipeline operation circuit in the general-purpose operation circuit, and the DCT operation is performed by the dedicated pipeline operation circuit in the special-purpose operation circuit. This has the effect that the DCT operation and the quantization operation can be continuously pipelined.
- the arithmetic unit according to the present invention is the arithmetic device according to Claim 3 or 4, wherein the arithmetic unit is configured to connect a second input and a second input.
- An adder receiving the first input and the second input, and outputting a subtraction result of the second input from the first input; and an adder receiving the first input and the second input;
- the addition result of the adder, the calculation result of the subtractor, or ⁇ is used as an input, and data selected from them is output. If the first input is a positive number, An output selector for selectively outputting the addition result of the adder, selectively outputting 0 when the first input is 0, and selectively outputting the calculation result of the subtractor otherwise. It is characterized by that.
- the inverse quantization operation is performed by the first pipeline operation IsJ path in the general-purpose operation circuit, and the inverse DCT operation is performed by the dedicated pipeline operation circuit in the dedicated operation circuit. If the inverse quantization and the inverse DCT operation can be continuously performed in a pipeline operation, the effect is obtained.
- An image processing device (Claim 6) is an image processing device equipped with a plurality of the arithmetic devices according to Claim 2, wherein the first dedicated pipeline arithmetic circuit includes: DCT operation that takes the first operation result as input, performs a one-dimensional discrete cosine transform on this input, and outputs the first dedicated operation result Circuit, and a second dedicated pipeline operation circuit.IDCT that receives the first operation result as input, performs a one-dimensional inverse discrete cosine transform on the input of, and outputs the second exclusive operation result.
- DCT operation that takes the first operation result as input, performs a one-dimensional discrete cosine transform on this input, and outputs the first dedicated operation result Circuit
- IDCT that receives the first operation result as input, performs a one-dimensional inverse discrete cosine transform on the input of, and outputs the second exclusive operation result.
- a first arithmetic unit having an arithmetic circuit; a first dedicated pipeline arithmetic circuit; a half-pel that receives the first arithmetic result as an input, performs a half-pel operation on the input, and outputs a first exclusive arithmetic result;
- An arithmetic circuit and a second dedicated pipeline arithmetic circuit a boost noise that receives the first arithmetic result as an input, applies a boost noise removal filter to this input, and outputs a second exclusive arithmetic result
- a second arithmetic unit having a filter arithmetic circuit, a host interface for transferring data to and from a host microcomputer, and inputting image data from an image AD converter for pre-scaling.
- Transfer data It is characterized by having a shared memory having a function.
- the quantization operation is performed by the second pipeline operation circuit in the general-purpose operation circuit, and the DCT operation is performed by the dedicated pipeline operation circuit in the dedicated operation circuit.
- This has an effect that the DCT operation and the quantization operation can be continuously performed by the pipeline operation.
- a plurality of the above-mentioned arithmetic units including a general-purpose arithmetic circuit and a special-purpose arithmetic circuit are mounted.
- the first special-purpose arithmetic circuit includes a DC-three arithmetic circuit and an IDCT arithmetic circuit.
- the image processing device of the present invention Since the image processing device of the present invention has an arithmetic circuit, the image processing device of the present invention functions as an encoder device when only the encoder operation is performed, and functions as a decoder device when only the decoder operation is performed. When executed in a time-sharing manner, there is an effect that an image processing device that operates as a codec device can be realized.
- FIG. 1 is a block diagram showing a configuration of an arithmetic unit according to Embodiment 1.
- FIG. 2 is a block diagram showing a schematic configuration of an arithmetic unit g according to the second embodiment.
- FIG. 3 is a diagram showing an inverse quantization operation expression in the operation device according to the second embodiment.
- FIG. 4 is a block diagram showing a configuration of an arithmetic operation unit in the operation device according to the second embodiment.
- FIG. 5 is a diagram illustrating output data control of an output selector in an arithmetic operation unit in the operation device according to the second embodiment.
- FIG. 6 is a diagram illustrating an input / output relationship of an arithmetic operation unit in the operation device according to the second embodiment.
- FIG. 7 is a diagram illustrating a flow of vector data in a pipeline arithmetic unit in the arithmetic device according to the second embodiment.
- FIG. 8 is a block diagram showing a schematic configuration of an arithmetic unit according to Embodiment 3.
- FIG. 9 is a diagram showing a quantization operation expression in the operation device of the third embodiment.
- FIG. 10 is a diagram showing a flow of vector data in a pipeline operation device in the operation device of the third embodiment.
- FIG. 11 is a block diagram showing a configuration of an image processing device according to a fourth embodiment.
- FIG. 12 is a diagram showing each process at the time of encoding in the image processing apparatus according to the fourth embodiment.
- FIG. 6 is a diagram showing the processing allotment in the hardware.
- the thirteenth view 1511 is a diagram illustrating the sharing of processing in each block at the time of decoding in the image processing device according to the fourth embodiment.
- FIG. 14 is a block diagram showing the configuration of a conventional arithmetic unit.
- FIGS. 1 to 8 Note that the embodiment shown here is merely an example, and is not necessarily limited to this embodiment.
- FIG. 1 is a block diagram showing a configuration of an arithmetic unit according to Embodiment 1 of the present invention.
- the arithmetic device is a program-controlled processor equipped with vector instructions, and is separated into a general-purpose arithmetic circuit 101 and a dedicated arithmetic circuit 102, as shown in FIG.
- the general-purpose arithmetic circuit 101 has a program control circuit 103, a first address generator ⁇ 04, a first data memory 105, a first pipeline arithmetic circuit 106, a second Address generator 1 13, second data memory 1 14, and second pipeline arithmetic circuit 1 12, and a dedicated arithmetic circuit ⁇ 0 2 includes a control circuit 1 15,
- the first dedicated pipeline arithmetic circuit 107, the second dedicated pipeline arithmetic circuit 108, and the third dedicated pipeline arithmetic circuit 109 specializing in high-load arithmetic for each application And a N-th dedicated pipeline arithmetic circuit 1 ⁇ 0.
- the program control circuit 103 includes a program memory, an instruction decoder, and a sequencer, and has an instruction analysis function for a vector instruction and an instruction execution control function in addition to a normal scalar instruction.
- the first operation circuit selection signal, the second operation circuit selection signal, and the dedicated pipeline operation circuit selection signal are output, and at the same time, the first activation signal is asserted.
- the general-purpose arithmetic circuit output circuit output data enable is asserted and the dedicated arithmetic circuit is output.
- the second start signal After asserting the data enable, asserting the second start signal after the number of cycles corresponding to the number of pipeline stages of the second pipeline operation circuit, and setting the length of the vector data set in advance. Assuming that there are M, after the first first start signal is asserted, the first start signal is negated after M cycles, and then the number of cycles corresponding to the number of pipeline stages of the first pipeline operation circuit 106 After that, the output data enable of the general-purpose arithmetic circuit is negated, the output data enable of the special-purpose arithmetic circuit is negated, and the second pipeline operation circuit 112 is cycled by the number of pipeline stages. Negate the start signal.
- the first address generator 104 outputs a predetermined address as the first address while the first activation signal is being asserted.
- the first data memory 105 outputs the first data according to the first address.
- the first pipeline operation circuit 106 includes basic operation units such as a multiplier, an arithmetic and logic operation unit, and a barrel shifter.
- the first operation circuit selection signal output after instruction analysis by the program control circuit 103. And performs a bipline operation on the first data to output a first operation result.
- the second address generator 113 outputs a predetermined address as a second address while the second activation signal is asserted.
- the second pipeline operation circuit 112 is a basic operation device such as a multiplier, an arithmetic and logic operation unit, a barrel shifter, etc., and a second operation circuit which is output after instruction analysis by the program control circuit 103. according to the selection signal, is constituted by combining, subjected to Paiburain operation on second data, u second data memory 1 1 4 for outputting a third operation result in accordance with the second address, third Stores the operation result of.
- the control circuit # 15 selects the first pipeline stage number notification signal, the second pipeline stage number notification signal, and the second pipeline stage number notification signal selected in accordance with the dedicated pipeline arithmetic circuit selection signal.
- the dedicated arithmetic circuit output The data enable is asserted, and after the negation of the general-purpose arithmetic circuit output data enable, the dedicated arithmetic circuit output data enable is negated after the number of cycles corresponding to the number of stages indicated by the n-th pipeline stage number notification signal.
- the first pipeline stage number notification signal, the second pipeline stage number notification signal, and the i-th pipeline stage number notification signal of the number of pipeline stages of each dedicated pipeline operation circuit 107 to 11 1 The first, Nth pipeline stage number notification signals are output.
- the data selection circuit 1 1 1 outputs one of the first exclusive ffl operation result, the second exclusive operation result, the third exclusive operation result, the first, and the Nth exclusive operation result. Select and output D.
- the first activation signal is asserted, and the first address generator 104 starts to issue M consecutive first addresses.
- M consecutive first data are read from the first data memory and input to the first pipeline operation path 1 ⁇ 6.
- the first pipeline operation circuit 106 sequentially performs an operation on the first data, and sequentially outputs the result as a first operation result.
- the output timing of the first S-first data is detected from the first pipeline arithmetic circuit 106, and the general-purpose arithmetic circuit output enable is asserted.
- the first dedicated pipeline operation circuit 107, the second dedicated pipeline operation circuit 1 ⁇ 8, the third dedicated pipeline operation circuit 109, the first and Nth dedicated pipeline operation circuits 110 Perform the operation on the first operation result sequentially, and convert the result as the first exclusive operation result, the second exclusive operation result, the third exclusive operation result, , And output them sequentially. From these exclusive operation results, the vector instruction is executed by the program control circuit 103. Is selected by the data selection circuit 111 according to the dedicated pipeline operation circuit selection signal output when the data is analyzed, and is output as a second operation result.
- assertion timing of the general-purpose arithmetic circuit output data enable, the first pipeline stage number notification signal selected by the dedicated pipeline selection signal output from the program control circuit, and the second pipeline stage number notification The output timing of the second operation result of the second operation is detected from one of the N-th pipeline stage number notification signal and the N-th pipeline stage number notification signal.
- Assert the operation circuit output data enable The second pipeline operation circuit 112 sequentially executes the operation on the second operation result, and sequentially outputs the result as the third operation result.
- Dedicated operation [From the timing of assertion of the Hi-route output data enable and the number of pipeline stages of the second pipeline operation circuit 112, the output timing of the second operation result of No. 1B is detected. Assert start signal 2.
- the second address generator 113 In response to the asserted second activation signal, the second address generator 113 starts to issue M consecutive addresses from the second address generator 113. According to the second address, M consecutive third operation results are stored in the second data memory.
- the program control circuit 115 detects the timing at which the M-th first data is read from the first data memory 105, and negates the first activation signal. Next, based on the negation timing of the first start signal and the number of pipeline stages of the first general-purpose arithmetic circuit 106, the first M-th arithmetic result is output from the first pipeline arithmetic circuit 106.
- the general-purpose arithmetic circuit output data enable is negated by detecting the timing of the operation.
- the control circuit 115 includes a first pipeline stage number notification signal selected by the dedicated pipeline selection signal output from the general-purpose arithmetic circuit output data enable and the program control circuit, and a second pipeline. Dedicated operation by detecting the output timing of the M-th second operation result from one of the stage number notification signal, the third pipeline stage number notification signal, ' ⁇ , and the N-th pipeline stage number notification signal Neglect the circuit output data enable.
- the program control circuit determines the M-th second pipeline arithmetic circuit based on the negation timing of the dedicated arithmetic circuit output data enable and the number of pipeline stages of the second pipeline arithmetic circuit.
- the general-purpose arithmetic circuit 10 ⁇ and the special-purpose arithmetic circuit 10 2 The timing control of the program control circuit 103 in the general-purpose arithmetic circuit 101 is performed so that the change of the dedicated arithmetic circuit 102 for each application does not affect the general-purpose arithmetic circuit 101.
- Necessary dedicated arithmetic circuit 102 The dedicated arithmetic circuit output data enable, which is information unique to the dedicated arithmetic circuit 102, is notified from the dedicated arithmetic circuit 102 to the general-purpose arithmetic circuit 101, and the program control circuit in the general-purpose arithmetic circuit 1 1 103 is configured to control the timing based on the dedicated information output signal enable of the dedicated arithmetic circuit, so that it is possible to realize an arithmetic device that can be applied to various applications. I can do it There is an effect. Furthermore, considering the era of intellectual property (IP), which will come in the future, it will be more effective for applications-if it can be done, it will be effective.
- IP intellectual property
- FIG. 2 is a block diagram showing a schematic configuration of an arithmetic unit according to Embodiment 2 of the present invention.
- the arithmetic device according to the second embodiment is the same as the arithmetic device according to the first embodiment shown in FIG. 1, and in FIG. 2, 201 corresponds to the first data memory 105 according to the first embodiment. , 211 correspond to the second data memory 114 of the first embodiment, 211 corresponds to the first pipeline operation circuit 106 of the first embodiment, and 211 corresponds to The second pipeline arithmetic circuit 1 12 of the first embodiment corresponds to the second pipeline arithmetic circuit 112, and the second pipeline arithmetic circuit 210 corresponds to one of the plurality of dedicated pipeline arithmetic circuits 107 to 110 of the first embodiment. You. Other configurations of the arithmetic device according to the second embodiment other than the above have the same configuration as the first embodiment shown in FIG.
- the first pipeline operation circuit 2 12 includes a first register 202, a second register 203, a multiplier 204, a third register 205, a fourth register 206, An arithmetic unit 207 and a fifth register 208 are provided.
- the second pipeline operation circuit 21 includes a sixth register 210.
- Specialty Pipeline operation circuit 2 14 includes an I DCT (Inversion Discrete Cosine Transfer) device 209.
- the IDCT calculator 209 performs one-dimensional inverse discrete cosine transform.
- FIG. 3 shows an arithmetic expression for inverse quantization in the arithmetic device according to the second embodiment.
- FIG. 4 shows the configuration of the arithmetic operation unit 207 in FIG.
- the arithmetic operation unit 207 includes an adder 401, a subtractor 402, and an output selector 4 (J3).
- the output selector 403 of the arithmetic operation unit 207 is configured as shown in FIG.
- the operation according to the sign of the first input is shown in Fig. 5. In Fig.
- the first input corresponds to the fifth data which is the output of the third register 205, and the second input is the fourth input.
- the inverse quantization operation is performed by the first pipeline arithmetic circuit 2 12 in the general-purpose arithmetic circuit 101, and the dedicated pipeline arithmetic in the dedicated arithmetic circuit 1 CJ is performed.
- the inverse DCT operation is performed by the circuit 214, thereby providing an effect that the inverse quantization and the inverse DCT operation can be continuously performed by the pipeline operation.
- FIG. 8 is a procedure diagram showing a schematic configuration of an arithmetic unit according to Embodiment 3 of the tree invention.
- the arithmetic device according to the third embodiment is the same as the arithmetic device according to the first embodiment shown in FIG. 1, and in FIG. 8, reference numeral 81 denotes a first data memory 105 according to the first embodiment; 8 1 2 corresponds to the first pipeline operation circuit 106 of the first embodiment, 8 105 corresponds to the second pipeline operation circuit 1 12 of the first embodiment, 8 1
- One of the plurality of dedicated pipeline operation circuits 107 to I 1 ⁇ of the first embodiment corresponds to one of the plurality of dedicated pipeline operation circuits.
- U having a configuration similar to that of Embodiment 1 shown in the figure
- the first pipeline operation circuit 812 includes a first register 8002.
- the second pipeline operation circuit 813 includes a second register 804, a third register 805, an arithmetic operation section 805, a fourth register 807, and a fifth register 808. , A multiplier 809, and a sixth register 810.
- the dedicated pipeline operation circuit 814 includes a DCT (Discrete Cosine Trans for) operation unit 803. This DCT calculator 803 performs discrete cosine transform.
- DCT Discrete Cosine Trans for
- FIG. 9 shows an arithmetic expression for quantization in the arithmetic device of the third embodiment.
- the first data memory 801 stores the REC in FIG. 9 which is the data to be subjected to DCT, and the second register 802 stores an odd number of QU ANs in the second register 802. ) And (1 QUANT + 1) in the case of an even number, and the reciprocal of 2 XQUANT in the fifth register 808.
- Arithmetic operation unit 806 has the same configuration as that of the second embodiment shown in FIG. 4, and its first input corresponds to the second data from second register 804. The second input corresponds to the third data from the third register 805. The input / output relationship of the arithmetic performance device 806 is as shown in FIGS. 5 and 6.
- the quantization operation is performed by the second pipeline arithmetic circuit 8 13 in the general-purpose arithmetic circuit 101, and the dedicated pipeline in the special arithmetic circuit 102 is performed.
- the DCT operation is performed by the operation circuit 814, which has an effect that the DCT operation and the quantization operation can be continuously performed to perform the pipeline operation.
- FIG. 11 is a block diagram showing a configuration of an image processing apparatus according to Embodiment 4 of the present invention.
- a first general-purpose arithmetic circuit 1 1 1 8 and a second general-purpose arithmetic circuit 1 1 1 9 This corresponds to the general-purpose operation IHJ road 101 shown in the form.
- the first dedicated arithmetic unit 1103 and the second dedicated arithmetic circuit 1104 correspond to the dedicated arithmetic circuit 102 shown in Embodiment 1 of FIG.
- the 1103 is equipped with the DCT arithmetic circuit 11 ⁇ 5 and the IDCT arithmetic circuit 1106 as one of the dedicated pipeline arithmetic circuits 107 to 110, and the second dedicated arithmetic circuit 1 104 is dedicated A boost noise filter operation circuit 1107 and a half-bell operation circuit 1108 are mounted as one of the pipeline operation circuits 107 to 110.
- the image processing apparatus includes a first shared memory 110, a second shared memory 111, a third shared memory 111, a host interface 111, and the like.
- DMA Direct memory access
- the first DSP core 110 executes the closed operation in the first DSP core 1101, executes the operation using the first dedicated ffl operation circuit 1103, and It has a function to execute data transfer with each of the two DSP cores 1102, the host interface 111, and the DMA control circuit 111.
- the first dedicated operation circuit 1103 has a function of executing a D-joint operation and a 1DCT operation under the control of the first DSP core # 101.
- the second DSP core 1102 executes the closed operation in the second DSP core 1102, executes the operation using the second dedicated arithmetic circuit ⁇ 104, and executes the second operation in accordance with the program. It has the function of initiating data transfer with each of the DSP core 111 and the DMA control circuit 111.
- the second dedicated arithmetic circuit 1104 has a function of executing a half-pel operation and a post noise filter operation under the control of the second DSF core # 102.
- the first shared memory 1109 has a function of transferring data between the first DSP core # 101 and the DMA control circuit 111.
- the second shared memory 1 1 1 0 and the 2nd DSP core 1 102 and the DMA control circuit It has a function of transferring data to and from the channels 111.
- the third shared memory 111 has a function of transferring data between the first DSP core 1101 and the second DSP core 1102.
- the host interface 1 1 1 2 has a bit stream command data input / output function, a first DSP core 1101, and a DMA control circuit 1 1 1
- 5 has a function of transferring data to and from each of 5.
- Frame memory 1 1 1 4 outputs stored data to DMA control IHJ path 1 1 1 5
- the DMA control circuit 111 is connected to the input from each of the first shared memory 110, the second shared memory 110, the host interface 111, and the video interface 111.
- the function of storing data in the frame memory 1114 and the output data from the frame memory 1113 are transferred to the first shared memory 1109, the second shared memory 1110, and the host interface. It has a function of outputting to each of the video interface 111 and the video interface 111.
- the video interface 1 1 1 3 has a function of transferring data to and from the DMA control circuit 1 1 1 5, a post-scaling function, and a post-scaling data converter for converting the post-scaled data into a DA converter. It has a function to output image data from the A / D converter, a function to input image data from the A / D converter, and a function to prescale image data input from the A / D converter.
- the AD converter 16 has a function of converting the input analog image data into a digital signal and outputting the digital image data to the video interface 111.
- the DA converter 111 has a function of converting digital image data input from the video interface 111 to analog and outputting the analog image data. Next, encoder processing and decoder processing in the image processing apparatus according to the fourth embodiment will be described.
- each block 1 1 1 1 2, 1 1 0 1, 1 of encoder processing by the image processing apparatus of the fourth embodiment Shows the processing contents in 102 and 1113.
- the encoder processing by the image processing apparatus according to the fourth embodiment is efficiently executed by performing the processing of sharing each block summarized in FIG. 12 in parallel in each block.
- the analog image data is input to the AD converter 111, and is converted to digital, and then input to the video interface 111.
- the data input to the video interface 11 13 is prescaled and converted to CIF (Common Internet File) or ii QCIF (Quadrature Common Internet File) format. After 15 is stored in a predetermined area of the frame memory 111.
- the ME (motion detection) processing is performed on the data to be encoded after the prescaler is applied, via the second shared memory 111, according to the program of the second DSP core 1102. Is applied. Since a half-pel operation is required in this process, the half-pel operation is performed by the half-pel operation circuit # 108 using the second dedicated operation circuit 1104.
- the data to be encoded is transferred to the first DSP core 111 via the third shared memory 111, and the i-th DSP core 111 According to, MC (motion compensation) processing, DCT calculation processing, Q (quantization) processing, IQ (dequantization) processing, IDCT calculation processing, VLC (variable length coding) processing were performed and coded.
- the image data is finally stored in a predetermined area of the frame memory 1114 via the DMA control circuit 1115.
- the DCT operation processing and the IDCT operation processing are executed using the DCT operation circuit 1106 and the IDCT operation circuit 1106 of the first dedicated operation circuit 1103.
- the encoded data is transmitted from the frame memory 111 to the host interface 112 via the DMA control circuit 111 in accordance with the command data received by the host interface 111. It is read out and sent out as a bitstream.
- FIG. 13 shows the contents of processing performed by the blocks 111, 120, 1102, and 111 of the decoder processing by the image processing apparatus according to the fourth embodiment.
- the decoder processing is performed by dividing each block summarized in FIG. 13 into each block. It is executed efficiently by parallel processing on the server.
- a bit stream is input according to the command data received by the host interface 111, and is stored in a predetermined area of the frame memory 111 via the DMA control circuit 115.
- the bit stream data is read out from the frame memory 111 through the DMA control circuit 111, the first shared memory 110 is read from the frame memory 111, and the first DSP core 111 VLD (variable length decoding) processing, IQ (dequantization) processing, IDCT processing, and MC (motion compensation) processing are performed in accordance with the program of
- VLD variable length decoding
- IQ dequantization
- IDCT processing IDCT processing
- MC motion compensation
- the decoded image data is stored in the DMA control circuit 111 according to the program of the second DSP core 112 Second shared from frame memory 1 1 1 4 via 1 5 Read out to the memory, subjected to a post-noise elimination filter using the second dedicated arithmetic circuit 1104, and passed from the second shared memory 1 1 1 10 via the DMA control circuit 1 1 1 5 to the frame.
- This data is stored in a predetermined area of the memory 111.
- the data subjected to the post noise filter is transferred from the frame memory 111 to the video interface 1 via the DMA control circuit 115.
- the data is input to 113, subjected to a postscaler, and output to a DA converter 1 1 1 7.
- the DA converter 1 1 1 7 converts the input digital image data into analog data and outputs it.
- Embodiment 4 functions as an encoder device when only the encoder operation is performed, and functions as a decoder device when only the decoder operation is performed.
- the decoder operation is performed in a time-division manner, there is an effect that a ghost image processing device that works as a codec device can be realized.
- the arithmetic device can incorporate any dedicated pipeline arithmetic circuit suitable for each application without changing the program control circuit. As a result, it can be applied to various applications. This is extremely useful as a device that can perform calculations. Furthermore, an image processing apparatus using the arithmetic device according to the present invention functions as an encoder device when only the encoder operation is performed, and functions as a decoder device when only the decoder operation is performed. This is extremely useful as an image processing device that operates as a codec device when the operation of the decoder and the decoder operation are subject to time division.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Mathematics (AREA)
- Algebra (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Databases & Information Systems (AREA)
- Complex Calculations (AREA)
- Image Processing (AREA)
- Advance Control (AREA)
Description
Claims
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP99972758A EP1050828A4 (en) | 1998-11-26 | 1999-11-25 | PROCESSOR AND IMAGE PROCESSING DEVICE |
JP2000584407A JP3619456B2 (ja) | 1998-11-26 | 1999-11-25 | 演算装置およびこれを用いた画像処理装置 |
US09/600,247 US6671708B1 (en) | 1998-11-26 | 1999-11-25 | Processor and image processing device |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP10/335443 | 1998-11-26 | ||
JP33544398 | 1998-11-26 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2000031658A1 true WO2000031658A1 (fr) | 2000-06-02 |
Family
ID=18288625
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP1999/006591 WO2000031658A1 (fr) | 1998-11-26 | 1999-11-25 | Processeur et dispositif de traitement d'image |
Country Status (5)
Country | Link |
---|---|
US (1) | US6671708B1 (ja) |
EP (1) | EP1050828A4 (ja) |
JP (1) | JP3619456B2 (ja) |
CN (1) | CN1109992C (ja) |
WO (1) | WO2000031658A1 (ja) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPWO2005025230A1 (ja) * | 2003-08-28 | 2006-11-16 | 株式会社日立超エル・エス・アイ・システムズ | 画像処理装置 |
US8833650B1 (en) | 2006-05-25 | 2014-09-16 | Sean I. Mcghie | Online shopping sites for redeeming loyalty points |
US8944320B1 (en) | 2006-05-25 | 2015-02-03 | Sean I. Mcghie | Conversion/transfer of non-negotiable credits to in-game funds for in-game purchases |
US9704174B1 (en) | 2006-05-25 | 2017-07-11 | Sean I. Mcghie | Conversion of loyalty program points to commerce partner points per terms of a mutual agreement |
US10062062B1 (en) | 2006-05-25 | 2018-08-28 | Jbshbm, Llc | Automated teller machine (ATM) providing money for loyalty points |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6898646B1 (en) * | 2000-05-03 | 2005-05-24 | Hewlett-Packard Development Company, L.P. | Highly concurrent DMA controller with programmable DMA channels |
US6819715B2 (en) * | 2000-07-25 | 2004-11-16 | Matsushita Electric Industrial Co., Ltd. | Image decoding method, image decoding apparatus, and data storage medium |
US7181070B2 (en) * | 2001-10-30 | 2007-02-20 | Altera Corporation | Methods and apparatus for multiple stage video decoding |
FI118654B (fi) * | 2002-11-06 | 2008-01-31 | Nokia Corp | Menetelmä ja järjestelmä laskuoperaatioiden suorittamiseksi ja laite |
FI118612B (fi) * | 2002-11-06 | 2008-01-15 | Nokia Corp | Menetelmä ja järjestelmä laskuoperaatioiden suorittamiseksi ja laite |
CN110865959B (zh) * | 2018-08-27 | 2021-10-15 | 武汉杰开科技有限公司 | 一种用于唤醒i2c设备的方法及电路 |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5299320A (en) * | 1990-09-03 | 1994-03-29 | Matsushita Electric Industrial Co., Ltd. | Program control type vector processor for executing a vector pipeline operation for a series of vector data which is in accordance with a vector pipeline |
WO1994023384A1 (en) * | 1993-03-31 | 1994-10-13 | Sony Corporation | Apparatus for adaptively processing video signals |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS58134357A (ja) * | 1982-02-03 | 1983-08-10 | Hitachi Ltd | ベクトルプロセッサ |
JP2584156B2 (ja) | 1990-09-03 | 1997-02-19 | 松下電器産業株式会社 | プログラム制御型プロセッサ |
EP0607988B1 (en) * | 1993-01-22 | 1999-10-13 | Matsushita Electric Industrial Co., Ltd. | Program controlled processor |
US5984512A (en) * | 1994-07-29 | 1999-11-16 | Discovision Associates | Method for storing video information |
US6490607B1 (en) * | 1998-01-28 | 2002-12-03 | Advanced Micro Devices, Inc. | Shared FP and SIMD 3D multiplier |
-
1999
- 1999-11-25 WO PCT/JP1999/006591 patent/WO2000031658A1/ja active Application Filing
- 1999-11-25 CN CN99802424A patent/CN1109992C/zh not_active Expired - Fee Related
- 1999-11-25 EP EP99972758A patent/EP1050828A4/en not_active Withdrawn
- 1999-11-25 US US09/600,247 patent/US6671708B1/en not_active Expired - Lifetime
- 1999-11-25 JP JP2000584407A patent/JP3619456B2/ja not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5299320A (en) * | 1990-09-03 | 1994-03-29 | Matsushita Electric Industrial Co., Ltd. | Program control type vector processor for executing a vector pipeline operation for a series of vector data which is in accordance with a vector pipeline |
WO1994023384A1 (en) * | 1993-03-31 | 1994-10-13 | Sony Corporation | Apparatus for adaptively processing video signals |
Non-Patent Citations (2)
Title |
---|
See also references of EP1050828A4 * |
TAKASHI HASHIMOTO ET AL.: "MPEG4 Codec DSP", INSTITUTE OF ELECTRONICS, INFORMATION AND COMMUNICATION ENGINEERS,, vol. 99, no. 145, May 1999 (1999-05-01), pages 45 - 50, XP002929675 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPWO2005025230A1 (ja) * | 2003-08-28 | 2006-11-16 | 株式会社日立超エル・エス・アイ・システムズ | 画像処理装置 |
JP4516020B2 (ja) * | 2003-08-28 | 2010-08-04 | 株式会社日立超エル・エス・アイ・システムズ | 画像処理装置 |
US8833650B1 (en) | 2006-05-25 | 2014-09-16 | Sean I. Mcghie | Online shopping sites for redeeming loyalty points |
US8944320B1 (en) | 2006-05-25 | 2015-02-03 | Sean I. Mcghie | Conversion/transfer of non-negotiable credits to in-game funds for in-game purchases |
US8950669B1 (en) | 2006-05-25 | 2015-02-10 | Sean I. Mcghie | Conversion of non-negotiable credits to entity independent funds |
US8973821B1 (en) | 2006-05-25 | 2015-03-10 | Sean I. Mcghie | Conversion/transfer of non-negotiable credits to entity independent funds |
US9704174B1 (en) | 2006-05-25 | 2017-07-11 | Sean I. Mcghie | Conversion of loyalty program points to commerce partner points per terms of a mutual agreement |
US10062062B1 (en) | 2006-05-25 | 2018-08-28 | Jbshbm, Llc | Automated teller machine (ATM) providing money for loyalty points |
Also Published As
Publication number | Publication date |
---|---|
CN1289422A (zh) | 2001-03-28 |
CN1109992C (zh) | 2003-05-28 |
EP1050828A1 (en) | 2000-11-08 |
US6671708B1 (en) | 2003-12-30 |
JP3619456B2 (ja) | 2005-02-09 |
EP1050828A4 (en) | 2008-08-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP3546437B2 (ja) | 適応形ビデオ信号演算処理装置 | |
US20030185305A1 (en) | Method of communicating between modules in a decoding system | |
US9665540B2 (en) | Video decoder with a programmable inverse transform unit | |
WO2000031658A1 (fr) | Processeur et dispositif de traitement d'image | |
CN1305313C (zh) | 以管线架构应用于离散余弦变换与反离散余弦变换的系统 | |
US6930689B1 (en) | Hardware extensions for image and video processing | |
JP2007299279A (ja) | 演算装置、プロセッサシステム、及び映像処理装置 | |
Gove | The MVP: a highly-integrated video compression chip | |
JP2001309386A (ja) | 画像処理装置 | |
Lo et al. | Improved SIMD architecture for high performance video processors | |
KR20030005199A (ko) | 스케일링 가능한 계산 복잡성 비디오 및 정지 이미지복호를 위한 근사적 역 이산 코사인 변환 | |
WO2000043868A1 (en) | Data processor and device for arithmetic operation | |
Hinrichs et al. | A 1.3-GOPS parallel DSP for high-performance image-processing applications | |
De Souza et al. | OpenCL parallelization of the HEVC de-quantization and inverse transform for heterogeneous platforms | |
CN101931804B (zh) | 无需使用存储器的旋转图像的设备及方法 | |
JP4740992B2 (ja) | オーバーラップフィルタリングとコア変換を行うための方法及び装置 | |
US9819951B2 (en) | Image processing method, devices and system | |
US20090201989A1 (en) | Systems and Methods to Optimize Entropy Decoding | |
EP1351511A2 (en) | Method of communicating between modules in a video decoding system | |
JP2005182499A (ja) | 並列演算装置 | |
Lin et al. | High performance architecture for unified forward and inverse transform of HEVC | |
Okuhata et al. | A low-power DSP Core architecture for low bitrate speech codec | |
JP4498848B2 (ja) | 画像処理装置 | |
KR100365729B1 (ko) | 이산 코사인 변환 및 역이산 코사인 변환 장치 | |
Yu et al. | Design for network camera based on DSP |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 99802424.4 Country of ref document: CN |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): CN IN JP US |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1999972758 Country of ref document: EP |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 09600247 Country of ref document: US |
|
WWP | Wipo information: published in national office |
Ref document number: 1999972758 Country of ref document: EP |