CN108021393B

CN108021393B - Calculation method and related product

Info

Publication number: CN108021393B
Application number: CN201711362567.3A
Authority: CN
Inventors: 胡帅; 刘恩赫; 张尧; 孟小甫
Original assignee: Anhui Cambricon Information Technology Co Ltd
Current assignee: Anhui Cambricon Information Technology Co Ltd
Priority date: 2017-12-15
Filing date: 2017-12-15
Publication date: 2020-10-27
Anticipated expiration: 2037-12-15
Also published as: CN108021393A

Abstract

The present disclosure provides an information processing method, which is applied in a computing device, and the computing device comprises: the device comprises a storage medium, a register unit and a matrix calculation unit; the method comprises the following steps: the computing device controls the matrix computing unit to obtain a first operation instruction, wherein the first operation instruction comprises a vector reading instruction required by executing the instruction; the computing device controls the arithmetic unit to send a reading command to the storage medium according to the vector reading instruction; and the computing device controls the operation unit to read the vector corresponding to the vector reading instruction by adopting a batch reading mode and execute the first operation instruction on the vector. The technical scheme provided by the application has the advantages of high calculation speed and high efficiency.

Description

Calculation method and related product

Technical Field

The present application relates to the field of data processing technologies, and in particular, to a computing method and a related product.

Background

Data processing is steps or stages which are needed to be carried out by most algorithms, after a computer is introduced into the field of data processing, more and more data processing is realized by the computer, and in the existing algorithms, the speed is low and the efficiency is low when computing equipment carries out matrix data computation.

Summary of the invention

The embodiment of the application provides a computing method and a related product, which can improve the processing speed of a computing device and improve the efficiency.

In a first aspect, a computing method is provided, which is applied in a computing device, where the computing device includes a storage medium, a register unit, and a matrix operation unit, and the method includes:

the computing device controls the matrix operation unit to obtain a first operation instruction, wherein the first operation instruction is used for realizing operation between vectors and matrixes, the first operation instruction comprises a vector reading instruction required by executing the instruction, the required vector is at least one vector, and the at least one vector is the same in length or different in length;

the computing device controls the matrix operation unit to send a reading command to the storage medium according to the vector reading instruction;

and the computing device controls the matrix operation unit to read the vector corresponding to the vector reading instruction from the storage medium in a batch reading mode and execute the first operation instruction on the vector.

In some possible embodiments, the executing the first operation instruction on the vector comprises:

and the computing device controls the matrix operation unit to adopt a multi-stage pipeline-level computing mode to execute the first operation instruction on the vector.

In some possible embodiments, each of the multiple pipeline stages includes at least one operator,

the computing device controls the matrix operation unit to adopt a multi-level pipeline computing mode, and the executing of the first operation instruction on the vector comprises the following steps:

the computing device controls the matrix operation unit to calculate the vector by using a first selection operator in a first-stage pipeline stage according to the selection of the multi-path selector to obtain a first result, inputs the first result into a second selection operator in a second-stage pipeline stage to perform calculation to obtain a second result, and so on until the ith-1 result is input into an ith selection operator in an ith-stage pipeline stage to perform calculation to obtain an ith result;

inputting the ith result into the storage medium for storage;

wherein the number i of the multiple pipeline stages is determined according to a computation topology of the first operation instruction, and i is a positive integer.

In some possible embodiments, each of the multiple pipeline stages is configured with a corresponding multiplexer, and the multiplexer is provided with an empty option for indicating that no calculation operation is performed on a k-th pipeline stage connected to the multiplexer and subsequent (k + 1) -th pipeline stages, where k is a positive integer less than or equal to i.

In some possible embodiments, the number of operators included in each of the multiple pipeline stages and the number of operators are custom set by a user side or the computing device side.

In some possible embodiments, each of the multiple pipeline stages includes a preset fixed operator, the fixed operators in each pipeline stage are different,

the computing device controls the matrix operation unit to calculate the vector by using a fixed arithmetic unit in a first-level pipeline level to obtain a first result, inputs the first result into a first fixed arithmetic unit in a second-level pipeline level to perform calculation to obtain a second result, and so on until the (i-1) th result is input into a fixed arithmetic unit in an i-level pipeline level to perform calculation to obtain an ith result;

inputting the ith result into the storage medium for storage;

In some possible embodiments, the operators in each of the multiple pipeline stages comprise any one or a combination of more of: a matrix addition operator, a matrix vector multiplication operator, a nonlinear operator, and a matrix comparison operator.

In some possible embodiments, the first operation instruction comprises any one of: a two-dimensional vector rotation instruction SVRO, a three-dimensional vector rotation instruction TVRO, a vector translation instruction VTRAN, a vector scaling instruction VZOOM, a vector clipping instruction vsear.

In some possible embodiments, the first operation instruction is a two-dimensional vector rotation instruction SVRO or a three-dimensional vector rotation instruction TVRO,

the computing device controls the matrix operation unit to perform vector 1 complementing operation on the vector by using a nonlinear operator in a first-level pipeline stage according to the selection of a multi-path selector to obtain a first result, inputs the first result into a nonlinear operator in a second-level pipeline stage to perform matrix element shifting calculation according to the acquired rotation center and rotation angle to obtain a second result, and inputs the second result into a matrix vector multiplication operator in a third-level pipeline stage to perform matrix multiplication vector calculation to obtain a third result; and inputting the third result to the storage medium for storage.

In some possible embodiments, the first operation instruction is any one of the following instructions: a vector translation instruction VTRAN, a vector scaling instruction VZOOM, a vector clipping instruction VSHEAR,

the computing device controls the matrix operation unit to perform vector 1 complementing operation on the vector by using a nonlinear operator in a first-level pipeline stage according to the selection of the multiplexer to obtain a first result, and inputs the first result into a nonlinear operator in a second-level pipeline stage to correspondingly execute any one of the following operations to obtain a second result: executing translation matrix construction operation according to the obtained translation factor, executing scaling matrix construction operation according to the obtained scaling factor, executing shearing matrix construction operation according to the obtained shearing factor, inputting the second result into a matrix vector multiplication arithmetic unit in a third-stage pipeline stage, and executing matrix multiplication vector calculation to obtain a third result; and inputting the third result to the storage medium for storage.

In some possible embodiments, the instruction format of the first operation instruction includes an opcode and at least one operation field, the opcode is used to indicate a function of the operation instruction, the operation unit can perform different matrix operations by identifying the opcode, and the operation field is used to indicate data information of the operation instruction, where the data information may be an immediate number or a register number, for example, when a vector is to be obtained, a vector start address and a vector length may be obtained in a corresponding register according to the register number, and a vector stored at a corresponding address is obtained in the storage medium according to the vector start address and the vector length. Optionally, any one or combination of more of the following information may be obtained in the respective registers: the instruction requires the number of rows, columns, data type, identification, memory address (head address), and length of the dimension of the vector, which refers to the length of the vector row and/or the length of the vector column.

In some possible embodiments, the vector read indication comprises: a memory address of a vector required by the instruction or an identification of a vector required by the instruction.

In some possible embodiments, when the vector read indicates an identification of a vector required by the instruction,

the control of the matrix operation unit by the computing device to send a read command to the storage medium according to the vector read instruction comprises:

the computing device controls the matrix operation unit to read the storage address corresponding to the identifier from the register unit in a unit reading mode according to the identifier;

and the computing device controls the matrix operation unit to send a reading command for reading the storage address to the storage medium and obtains the vector by adopting a batch reading mode.

In some possible embodiments, the computing device further comprises: a cache unit, the method further comprising:

the computing device caches operation instructions to be executed in the cache unit.

In some possible embodiments, before the computing device controls the matrix operation unit to obtain the first operation instruction, the method further comprises:

the computing device determines whether the first operation instruction is associated with a second operation instruction before the first operation instruction, if so, the first operation instruction is cached in the cache unit, and after the second operation instruction is executed, the first operation instruction is extracted from the cache unit and transmitted to the operation unit;

the determining whether the first operation instruction and a second operation instruction before the first operation instruction have an association relationship includes:

extracting a first storage address interval of a required vector in the first operation instruction according to the first operation instruction, extracting a second storage address interval of the required vector in the second operation instruction according to the second operation instruction, determining that the first operation instruction and the second operation instruction have an association relationship if the first storage address interval and the second storage address interval have an overlapped area, and determining that the first operation instruction and the second operation instruction do not have an association relationship if the first storage address interval and the second storage address interval do not have an overlapped area.

In a second aspect, a computing device is provided, comprising functional units for performing the method of the first aspect described above.

In a third aspect, a computer-readable storage medium is provided, which stores a computer program for electronic data exchange, wherein the computer program causes a computer to perform the method provided in the first aspect.

In a fourth aspect, there is provided a computer program product comprising a non-transitory computer readable storage medium having a computer program stored thereon, the computer program being operable to cause a computer to perform the method provided by the first aspect.

In a fifth aspect, there is provided a chip comprising a computing device as provided in the second aspect above.

In a sixth aspect, a chip packaging structure is provided, which includes the chip provided in the fifth aspect.

In a seventh aspect, a board is provided, where the board includes the chip packaging structure provided in the sixth aspect.

In an eighth aspect, an electronic device is provided, which includes the board card provided in the seventh aspect.

In some embodiments, the electronic device comprises a data processing apparatus, a robot, a computer, a printer, a scanner, a tablet, a smart terminal, a cell phone, a tachograph, a navigator, a sensor, a camera, a server, a cloud server, a camera, a camcorder, a projector, a watch, a headset, a mobile storage, a wearable device, a vehicle, a household appliance, and/or a medical device.

In some embodiments, the vehicle comprises an aircraft, a ship, and/or a vehicle; the household appliances comprise a television, an air conditioner, a microwave oven, a refrigerator, an electric cooker, a humidifier, a washing machine, an electric lamp, a gas stove and a range hood; the medical equipment comprises a nuclear magnetic resonance apparatus, a B-ultrasonic apparatus and/or an electrocardiograph.

The embodiment of the application has the following beneficial effects:

it can be seen that, through the embodiments of the present application, the computing apparatus is provided with the register unit and the storage medium, which are respectively used for storing scalar data and vector data, and the present application allocates a unit reading mode and a batch reading mode to the two memories, and allocates a data reading mode matching the characteristics of the vector data to the characteristics of the vector data, so that the bandwidth can be well utilized, and the influence of the bottleneck of the bandwidth on the computation speed is avoided.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a schematic structural diagram of a computing device according to an embodiment of the present application.

Fig. 2 is a schematic structural diagram of an arithmetic unit according to an embodiment of the present application.

Fig. 3 is a schematic flowchart of a calculation method according to an embodiment of the present invention.

Fig. 4A and 4B are schematic diagrams of architectures of two pipeline stages according to an embodiment of the present disclosure.

Fig. 5 is a schematic structural diagram of a pipeline stage according to an embodiment of the present application.

Fig. 6A and fig. 6B are schematic diagrams of formats of two instruction sets provided by an embodiment of the present application.

Fig. 7 is a schematic structural diagram of another computing device according to an embodiment of the present application.

Fig. 8 is a flowchart illustrating a computing device executing a two-dimensional/three-dimensional vector rotation instruction according to an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The terms "first," "second," "third," and "fourth," etc. in the description and claims of this application and in the accompanying drawings are used for distinguishing between different objects and not for describing a particular order. Furthermore, the terms "include" and "have," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.

The matrix referred to in the present application may be specifically an m × N matrix, where m and N are integers greater than or equal to 1, and when m or N is 1, it may be represented as a 1 × N matrix or an m × 1 matrix, and may also be referred to as a vector; when m and n are both 1, it can be regarded as a special matrix of 1 x 1. The matrix can be any one of the three types of matrices, which is not described in detail below.

The embodiment of the application provides a computing method which can be applied to a computing device. Fig. 1 is a schematic structural diagram of a possible computing device according to an embodiment of the present invention. The computing device shown in fig. 1 includes:

the storage medium 201 is used for storing a matrix (which may also be a vector). The storage medium can be a high-speed temporary storage memory which can support matrix data (vector data) with different lengths; the application temporarily stores necessary calculation data on a scratch pad Memory (Scratcchpad Memory), so that the arithmetic device can more flexibly and effectively support data with different lengths in the matrix operation process. The storage medium may also be an off-chip database, a database or other medium capable of storage, etc.

Register unit 202 to store scalar data, wherein the scalar data includes, but is not limited to: the scalar quantity of the matrix data or vector data (also referred to herein as matrix/vector) at the memory address of the storage medium 201 and the vector and matrix operations. In one embodiment, the register unit may be a scalar register file that provides scalar registers needed during operations, the scalar registers storing not only matrix addresses, but also scalar data. It should be understood that the matrix address (i.e., the memory address of the matrix, such as the first address) is also a scalar. When matrix and vector operations are involved, the arithmetic unit needs to obtain not only the matrix address from the register unit, but also the corresponding scalar from the register unit, such as the row number and column number of the matrix, the type of matrix data (also may be referred to as data type), the length of matrix dimension (specifically, the length of matrix row, length of matrix column, etc.).

The arithmetic unit 203 (also referred to as a matrix arithmetic unit 203 in this application) is configured to obtain and execute a first arithmetic instruction. As shown in fig. 2, the arithmetic unit includes a plurality of arithmetic units, which include but are not limited to: a matrix addition operator 2031, a matrix multiplication operator 2032, a size comparison operator 2033 (which may be a matrix comparison operator), a nonlinear operator 2034, and a matrix vector multiplication operator 2035.

The method, as shown in fig. 3, includes the following steps:

step S301, the arithmetic unit 203 obtains a first arithmetic instruction, where the first arithmetic instruction is used to implement the operation of a vector and a matrix, and the first arithmetic instruction includes: the vector read indication required to execute the instruction.

In step S301, the vector read instruction required for executing the instruction may be various, for example, in an optional technical solution of the present application, the vector read instruction required for executing the instruction may be a storage address of a required vector. For another example, in another alternative embodiment of the present application, the vector read instruction required for executing the instruction may be an identifier of the required vector, and the identifier may be represented in various forms, for example, a name of the vector, an identifier of the vector, and a register number or a storage address of the vector in a register unit.

The following describes, by way of a practical example, a vector read instruction required for executing the first operation instruction, where the vector operation formula is assumed to be f (x) a + B, where A, B are vectors. Then in addition to carrying the vector operation formula, the first operation instruction may also carry the memory address of the vector required by the vector operation formula, specifically, for example, the memory address of a is 0000-0FFF, and the memory address of B is 1000-1 FFF. As another example, the identities of a and B may be carried, for example, the identity of a is 0101 and the identity of B is 1010.

In step S302, the arithmetic unit 203 sends a read command to the storage medium 201 according to the vector read instruction.

The implementation method of the step S302 may specifically be:

if the vector reading instruction can be the memory address of the required vector, the arithmetic unit 203 sends the reading command for reading the memory address to the storage medium 201 and obtains the corresponding vector by using a batch reading method.

If the vector reading instruction can be an identifier of a required vector, the arithmetic unit 203 reads a storage address corresponding to the identifier from the register unit in a unit reading manner according to the identifier, and then the arithmetic unit 203 sends a reading command for reading the storage address to the storage medium 201 and obtains a corresponding vector in a batch reading manner.

The single reading mode may be specifically that data of a unit is read each time, that is, 1bit data. The reason why the unit reading mode, i.e., the 1-bit reading mode, is provided at this time is that, for scalar data, the occupied capacity is very small, and if the batch data reading mode is adopted, the read data amount is easily larger than the required data capacity, which may cause waste of bandwidth, so that the unit reading mode is adopted for reading scalar data to reduce the waste of bandwidth.

In step S303, the arithmetic unit 203 reads the vector corresponding to the instruction in a batch reading manner, and executes the first operation instruction on the vector.

The batch reading mode in step S303 may specifically be that each reading is performed on data of multiple bits, for example, the number of bits of the data read each time is 16 bits, 32 bits, or 64 bits, that is, the data read each time is data of a fixed number of bits regardless of the required data amount, and this batch reading mode is very suitable for reading large data.

The technical scheme's that this application provided calculating device is provided with register unit and storage medium, it stores scalar data and vector data respectively, and this application has distributed unit reading mode and batch reading mode for two kinds of memory, through the data reading mode to the characteristics distribution matching of vector data, the utilization bandwidth that can be fine, avoid because the bottleneck of bandwidth is to the influence of volume computational rate, in addition, to the register unit, because its storage be scalar data, the reading mode of scalar data has been set up, the utilization ratio of bandwidth has been improved, so the technical scheme that this application provides can be fine utilizes the bandwidth, avoid the influence of bandwidth to computational rate, so it has the advantage that computational rate is fast, high efficiency.

Optionally, the executing the first operation instruction on the vector may specifically be:

the operation unit 203 may adopt a multi-stage pipeline stage calculation manner, and in the embodiment of the present application, the first operation instruction may be executed on the vector by adopting an i-stage pipeline stage calculation manner. The method specifically comprises the following embodiments.

In a first embodiment, the computing device may further design at least one multiplexer MMUX to implement multi-level pipeline computation. Specifically, fig. 4A and 4B respectively show the implementation architectures of two multi-flow water stages. As shown in fig. 4A, the computing device may design a multiplexer MMUX for each pipeline stage; fig. 4B shows that one multiplexer MMUX is provided for a plurality of pipeline stages. The multiplexer is connected with the pipeline stage which is correspondingly required to be controlled/selected and is used for selecting the arithmetic unit in the pipeline stage to realize related calculation. It should be understood that the operator selected by the multiplexer from the pipeline stage is determined according to the computing network topology corresponding to the first operation instruction, which will be described in detail below.

In a specific implementation, the operation unit 203 may calculate the vector by using a first selection operator selected by a first (stage) pipeline stage according to the selection of the multiplexer to obtain a first result, then input the first result to a second pipeline stage, perform the calculation by using a second selection operator selected by the second pipeline stage to obtain a second result, and so on, input the i-1 th result to the ith pipeline stage, and perform the calculation by using the selected ith selection operator to obtain the ith result. Here, the ith result is an output result (specifically, an output vector). Further, the arithmetic unit 203 may store the output result to the storage medium 201.

The number i of the multiple pipeline stages is specifically determined according to a calculation topology of the first operation instruction, and i is a positive integer. Typically, i ═ 3. A respective operator may be provided in each pipeline stage, including, but not limited to, any one or combination of: matrix addition operators, matrix vector multiplication operators, nonlinear operators, matrix comparison operators, and other matrix operators. That is, the number of the operators and the number of the operators included in each pipeline stage may be set by a user side or the computing device side in a self-defined manner, and is not limited.

Taking i as 3, three-stage pipeline as an example, the arithmetic unit may select, through three multiplexers, an arithmetic unit (also referred to as an arithmetic unit) to be used in each of the first to third pipeline stages; meanwhile, performing a first pipeline stage calculation on the vector to obtain a first result, (optionally) inputting the first result to a second pipeline stage to perform a second pipeline stage calculation to obtain a second result, (optionally) inputting the second result to a third pipeline stage to perform a third pipeline stage calculation to obtain a third result, and (optionally) storing the third result in the storage medium 201. Fig. 5 shows a flow chart of the operation of a pipeline stage.

The first effluent stage includes, but is not limited to: matrix multiplication operators, etc.

The second pipeline stage includes but is not limited to: matrix addition operators, magnitude comparison operators, and the like.

Such third effluent stages include, but are not limited to: non-linear operators, matrix vector multiplication operators, matrix scalar multiplication operators, and the like.

For vector calculation, for example, when a general-purpose processor is used for calculation, the calculation steps may specifically be that the processor performs calculation on a vector to obtain a first result, then stores the first result in a memory, reads the first result from the memory, performs a second calculation to obtain a second result, then stores the second result in the memory, reads the second result from the memory, performs a third calculation to obtain a third result, and then stores the third result in the memory. It can be seen from the above calculation steps that when the general-purpose processor performs vector calculation, it does not perform calculation at the split water level, and then the calculated data needs to be stored after each calculation, and needs to be read again when performing the next calculation, so this scheme needs to repeatedly store and read data many times.

In another embodiment of the present application, the flow components may be freely combined or one stage of flow stages may be adopted. For example, the second pipeline stage may be merged with the third pipeline stage, or both the first and second pipelines and the third pipeline may be merged, or each pipeline stage may be responsible for different operations. For example, the first stage pipeline is responsible for comparison operations, partial multiplication operations, the second stage pipeline is responsible for combinations of nonlinear operations and matrix vector multiplication, etc. That is, the i pipeline stages designed in the present application support parallel connection, serial connection, and combination of any multiple pipeline stages to form different permutation and combination, which is not limited in the present application.

It should be noted that each multiplexer may also be provided with a null option, that is, the pipeline stage connected to the multiplexer and the subsequent pipeline stages do not participate in the operation. That is, the null option in this application is used to indicate that the pipeline stage k connected to the multiplexer and subsequent pipeline stages k +1 to i do not perform the calculation operation, where k is a positive integer less than or equal to i.

Taking i as 3, taking three-stage pipeline as an example, if the selector connected with the third pipeline stage selects an empty option, the third pipeline stage does not participate in the operation, and the pipeline stage currently executing the operation is less than three; for example, if a certain operation instruction includes two stages of instructions, the multiplexer corresponding to the third pipeline stage selects the null option.

By adopting the computing device (namely, a multiplexer is designed to select the arithmetic unit/arithmetic part needed to be used in each level of pipeline level), the following beneficial effects are achieved: besides improving the bandwidth, the method has the characteristics of clear logic, single input interface and output interface and good operability, and the output result of the non-operation component jumps from a rear pipeline stage to a front pipeline stage.

In a second embodiment, the computing device may design a corresponding fixed pipeline level implementation architecture for each type of operation instruction. The architecture for implementing three stages of pipeline stages corresponding to an operation instruction is shown in fig. 5. That is, for a certain arithmetic instruction, the arithmetic unit included in each pipeline stage is fixedly set in advance on the user side or the computing device side, and may be referred to as a fixed arithmetic unit in the present application. In addition, the fixed operators in each pipeline stage may be the same or different, and are usually different. For example, the first stage pipeline is a matrix addition operator, the second stage pipeline is a matrix multiplication operator, and the third stage pipeline is a nonlinear operator; for another example, the first stage pipeline is a matrix addition operator, the second stage pipeline is a matrix addition operator, the third stage pipeline is a matrix multiplication operator, and so on. I.e. different operational instructions, relate to different pipeline stage devices (implementation architectures). It should be understood that, according to the implementation requirements of different operation instructions, the number i of pipeline stages involved may be different, and may be increased or decreased correspondingly, which is not limited in the present application.

In a specific implementation, the arithmetic unit 203 sequentially uses the fixed arithmetic units in the first (second) pipeline stage to calculate the vector to obtain a first result, inputs the first result into the second pipeline stage to perform calculation using the fixed arithmetic units therein to obtain a second result, and so on until the (i-1) th result is input into the ith pipeline stage to perform calculation using the fixed arithmetic units therein to obtain the ith result. Here, the ith result is an output result (specifically, an output vector). Further, the arithmetic unit 203 may store the output result to the storage medium 201. For the number i of the multiple pipeline stages and the fixed arithmetic units designed in each pipeline stage, reference may be made to the related explanations in the foregoing embodiments, and details are not repeated here.

Take i-3, three-stage flowing water as an example. Referring to fig. 5, a fixed implementation architecture of a pipeline stage corresponding to an operation instruction is shown. Specifically, the arithmetic unit performs multiplication of the first pipeline stage on the vector to obtain a first result, inputs the first result to the second pipeline stage to perform addition of the second pipeline stage to obtain a second result, inputs the second result to the third pipeline stage to perform nonlinear calculation of the third pipeline stage to obtain a third result, and stores the third result (i.e., the output result) in the storage medium 201.

It should be noted that, the arithmetic unit in each pipeline stage in the computing apparatus is set by self-definition in advance, and once it is determined that the arithmetic unit cannot be changed; i.e. the i-stage pipeline stage can be designed as any permutation and combination of the arithmetic units, and once the i-stage pipeline stage is driven, the i-stage pipeline stage is not changed, different arithmetic instructions can be designed into different i-stage pipeline stage devices. Wherein the computing device may adaptively increase/decrease the number of pipeline stages as required by a particular instruction. Finally, pipeline devices designed for different instructions may be combined together to form the computing device.

By adopting the computing device (namely the arithmetic unit/arithmetic part in each level of the pipeline is designed and fixed), the following beneficial effects are achieved: besides improving the bandwidth, the method has the characteristics of high specificity, no redundant logic judgment, further improved operation performance and high operation speed.

Optionally, the computing device may further include: the cache unit 204 is configured to cache the first operation instruction. When an instruction is executed, if the instruction is the earliest instruction in the uncommitted instructions in the instruction cache unit, the instruction is back-committed, and once the instruction is committed, the change of the device state caused by the operation of the instruction cannot be cancelled. In one embodiment, the instruction cache unit may be a reorder cache.

Optionally, before step S301, the method may further include:

and determining whether the first operation instruction is associated with a second operation instruction before the first operation instruction, if so, extracting the first operation instruction from the cache unit and transmitting the first operation instruction to the operation unit 203 after the second operation instruction is completely executed. If the first operation instruction is not related to the instruction before the first operation instruction, the first operation instruction is directly transmitted to the operation unit.

The specific implementation method for determining whether the first operation instruction and the second operation instruction before the first operation instruction have an association relationship may be:

extracting a first storage address interval of a required vector in the first operation instruction according to the first operation instruction, extracting a second storage address interval of the required vector in the second operation instruction according to the second operation instruction, and determining that the first operation instruction and the second operation instruction have an association relation if the first storage address interval and the second storage address interval have an overlapped area. And if the first storage address interval and the second storage address interval are not overlapped, determining that the first operation instruction and the second operation instruction do not have an association relation.

If an overlapping area appears in the storage region section, which indicates that the first operation instruction and the second operation instruction access the same vector, and the space for storing the vectors is relatively large, for example, the same storage region is used as a condition for judging whether the vectors are in the association relationship, it may be the case that the storage region accessed by the second operation instruction includes the storage region accessed by the first operation instruction, for example, the second operation instruction accesses the a vector storage region, the B vector storage region and the C vector storage region, and if the A, B storage region is adjacent or the A, C storage region is adjacent, the storage region accessed by the second operation instruction is the A, B storage region and the C storage region, or the A, C storage region and the B storage region. In this case, if the storage area of the a vector and the storage area of the D vector are accessed by the first operation instruction, the storage area of the vector accessed by the first operation instruction cannot be the same as the storage area of the vector of the second operation instruction paradigm, and if the same judgment condition is adopted, it is determined that the first operation instruction and the second operation instruction are not associated, but practice proves that the first operation instruction and the second operation instruction belong to an association relationship at this time, so the present application judges whether the storage area is the association relationship condition by whether there is an overlapping area, and can avoid the misjudgment of the above situation.

The following describes, by way of an actual example, which cases belong to the associative relationship and which cases belong to the non-associative relationship. It is assumed here that vectors required by the first operation instruction are an a vector and a D vector, where a vector has a storage area of [ 0001, 0FFF ], and D vector has a storage area of [ a000, AFFF ], and for the second operation instruction, the vectors are an a vector, a B vector, and a C vector, and their corresponding storage areas are [ 0001, 0FFF ], [ 1000, 1FFF ], [ B000, BFFF ], respectively, and for the first operation instruction, the corresponding storage areas are: (0001, 0 FFF), (a 000, AFFF), for the second operation instruction, the corresponding storage area is: [ 0001, 1FFF ], [ B000, BFFF ], so that the storage area of the second operation instruction has an overlapping area [ 0001, 0FFF ] with the storage area of the first operation instruction, so that the first operation instruction has an association relationship with the second operation instruction.

It is assumed here that vectors required by the first operation instruction are an E vector and a D vector, where a vector has a storage area of [ C000, CFFF ], and D vector has a storage area of [ a000, AFFF ], and for the second operation instruction, the vectors are an a vector, a B vector, and a C vector, and their corresponding storage areas are [ 0001, 0FFF ], [ 1000, 1FFF ], [ B000, BFFF ], respectively, and for the first operation instruction, the corresponding storage areas are: for the second operation instruction, the corresponding storage area is: because [ 0001, 1FFF ] and [ B000, BFFF ], the storage area of the second operation instruction does not have an overlapping area with the storage area of the first operation instruction, and the first operation instruction and the second operation instruction have no relationship.

In this application, as shown in fig. 6A, the operation instruction includes an operation code and at least one operation field, where the operation code is used to indicate a function of the operation instruction, and as shown in fig. 6A, the operation unit can perform different vector operations by identifying the operation code, and the operation field is used to indicate data information of the operation instruction, where the data information may be an immediate or a register number, for example, when a vector is to be obtained, a vector start address and a vector length can be obtained in a corresponding register according to the register number, and then a vector stored in a corresponding address is obtained in a storage medium according to the vector start address and the vector length.

That is, the first operation instruction may include: the operation domains and at least one opcode, for example, vector operation instructions, are shown in table 1, where register 0, register 1, register file 2, register 3, and register 4 may be the operation domains. Wherein, each register 0, register 1, register 2, register 3, register 4 is used to identify the number of the register, which may be one or more registers. It should be understood that the number of registers in the opcode is not limited, and each register is used to store data information associated with an operation instruction.

Fig. 6B is a schematic diagram of a format of an instruction set of another instruction (which may be a first operation instruction and may also be referred to as an operation instruction) provided in the present application, where, as shown in fig. 6B, the instruction includes at least two opcodes and at least one operation field, where the at least two opcodes include a first opcode and a second opcode (shown as opcode 1 and opcode 2, respectively). The opcode 1 is used to indicate a type of an instruction (i.e., a certain class of instructions), and may specifically be an IO instruction, a logic instruction, or an operation instruction, etc., and the opcode 2 is used to indicate a function of an instruction (i.e., an interpretation of a specific instruction under the class of instructions), such as a matrix operation instruction in the operation instruction (e.g., a matrix multiplication vector instruction MMUL, a matrix inversion instruction MINV, etc.), a vector operation instruction (e.g., a vector derivation instruction VDIER, etc.), etc., which are not limited in this application.

It should be understood that the format of the instructions may be custom set either on the user side or on the computing device side. The opcode of the instruction may be designed to be a fixed length, such as 8-bit, 16-bit, and so on. The instruction format as shown in fig. 6A has the following advantageous features: the operation code occupies less bits and the design of the decoding system is simple. The instruction format as shown in fig. 6B has the following advantageous features: the variable length and the higher average decoding efficiency are achieved, and when the number of specific instructions is small and the calling frequency is high under a certain class of instructions, the length of the second operation code (i.e. operation code 2) is designed to be short, so that the decoding efficiency can be improved. In addition, the readability and the expandability of the instruction can be enhanced, and the encoding structure of the instruction can be optimized.

In the embodiment of the present application, the instruction set includes operation instructions with different functions, which may specifically be:

a two-dimensional vector rotation instruction (SVRO) according to which the device extracts vector data of a predetermined size from a predetermined address of a memory (preferably a scratch pad memory or a scalar register file) and changes the vector data into an augmented form, extracts predetermined rotation center coordinate data and rotation angle data from the scalar register file to generate a rotation matrix, performs a matrix-by-vector multiplication operation in an operation unit, and writes back the calculation result to the predetermined address of the memory (preferably the scratch pad memory or the scalar register file). It is worth noting that the vectors may be stored in the scratch pad memory as a special form of matrix (a matrix with only one row of elements). The memory may include, but is not limited to, a scratch pad memory, as follows.

A three-dimensional vector rotation instruction (TVRO) according to which the device extracts vector data of a prescribed size from a prescribed address of a memory (preferably a scratch pad memory or a scalar register file) and changes it into an augmented form, extracts specified rotation axis data and rotation angle data from the scalar register file to generate a rotation matrix, performs multiplication of a matrix by a vector in a matrix operation unit, and writes back the calculation result to the prescribed address of the memory; it is worth noting that the vectors may be stored in the scratch pad memory as a special form of matrix (a matrix with only one row of elements).

According to the vector translation instruction (VTRAN), the device extracts vector data with a set size from a designated address of a memory and changes the vector data into an augmentation form, extracts data in each designated translation direction from a scalar register file to generate a translation matrix, performs matrix multiplication operation in a matrix operation unit, restores the original dimension from the augmentation form and writes the original dimension back to the designated address of the memory; it is worth noting that the vectors may be stored in the scratch pad memory as a special form of matrix (a matrix with only one row of elements).

Vector scaling instructions (VZOOM) according to which the device fetches vector data of a specified size from a specified address of the memory and changes it into an augmented form, fetches specified scaled magnitude data from a scalar register file, generates a shearing matrix, performs a matrix-by-vector multiplication operation in a matrix operation unit, and writes the calculation result back to the specified address of the memory; it is worth noting that the vectors may be stored in the scratch pad memory as a special form of matrix (a matrix with only one row of elements).

According to the vector clipping instruction (VSHEAR), the device takes out vector data with specified size from the specified address of the memory and changes the vector data into an augmentation form, takes out the amplitude data in each specified direction from the scalar register file to generate a clipping matrix, carries out matrix multiplication in the matrix operation unit, and writes the calculation result back to the specified address of the memory; it is worth noting that the vectors may be stored in the scratch pad memory as a special form of matrix (a matrix with only one row of elements).

It should be understood that the operation/operation instruction provided in the present application is mainly used for operation operations in an input vector form and an output vector form, and does not limit the form of intermediate data generated by an operation process. The arithmetic units designed in each stage of the pipeline stage in the present application include, but are not limited to, any one or combination of more of the following: matrix addition arithmetic unit, matrix multiplication arithmetic unit, matrix vector multiplication arithmetic unit, nonlinear arithmetic unit, matrix comparison arithmetic unit.

The following exemplifies calculation of an operation instruction (i.e., a first operation instruction) according to the present application.

Taking the first operation instruction as a two-dimensional vector rotation instruction SVRO as an example, a two-dimensional vector of a given rotation center is calculated. In a specific implementation, given a two-dimensional vector X (X, y), a center of rotation (a, b), and a rotation angle c, a rotation vector of the given vector X is calculated as follows.

Accordingly, the instruction format of the two-dimensional vector rotation instruction SVRO is specifically:

with reference to the foregoing embodiment, the arithmetic unit may obtain the two-dimensional vector rotation instruction SVRO, decode the SVRO, read the vector X, the rotation center, and the rotation angle from the register unit, and use the first-level multiplexer to select the non-linear arithmetic unit to perform the 1-complementing operation on the vector X to obtain a first result (specifically, a to-be-rotated vector formed after the 1-complementing operation); meanwhile, a second pipeline-level multiplexer is used for selecting a nonlinear operator to execute rotation matrix construction operation (optional, movement operation of matrix elements is also involved) according to the read rotation center and the read rotation angle so as to correspondingly obtain a second result (specifically, a rotation matrix); and inputting the first result and the second result into a multiplexer of a third pipeline stage to select a matrix vector multiplication operator to perform matrix multiplication vector operation (specifically to perform matrix multiplication vector operation on the vector to be rotated and the rotation matrix) to obtain a third result (namely an output result). Optionally, the third result is stored in a storage medium. Optionally, the first result may be input into the second pipeline stage, and the second result is obtained together after the rotation matrix is constructed by the second pipeline stage. Namely, the second result comprises a rotation matrix and a vector to be rotated, so that the second result is only input into a third stream water stage for processing to obtain an output result.

Taking the first operation instruction as a three-dimensional vector rotation instruction TVRO as an example, the rotation in the three-dimensional space is performed. In a specific implementation, a three-dimensional vector X (X, y, z), a rotation center (u, v, w) and a rotation angle c are given, and the rotated three-dimensional vector is calculated according to the following formula.

Correspondingly, the instruction format of the three-dimensional vector rotation instruction TVRO is specifically:

with reference to the foregoing embodiment, the arithmetic unit may obtain the three-dimensional vector rotation instruction TVRO, decode the TVRO, read the vector X, the rotation center, and the rotation angle from the register unit, and use the first-stage multiplexer to select the non-linear arithmetic unit to perform the 1-complementing operation on the vector X to obtain a first result (specifically, a to-be-rotated vector formed after the 1-complementing operation); meanwhile, a second pipeline-level multiplexer is used for selecting a nonlinear operator to execute rotation matrix construction operation (optional, movement operation of matrix elements is also involved) according to the read rotation center and the read rotation angle so as to correspondingly obtain a second result (specifically, a rotation matrix); and inputting the first result and the second result into a multiplexer of a third pipeline stage to select a matrix vector multiplication operator to perform matrix multiplication vector operation (specifically to perform matrix multiplication vector operation on the vector to be rotated and the rotation matrix) to obtain a third result (namely an output result). Optionally, the third result is stored in a storage medium. Optionally, the first result may be input into the second pipeline stage, and the second result is obtained together after the rotation matrix is constructed by the second pipeline stage. Namely, the second result comprises a rotation matrix and a vector to be rotated, so that the second result is only input into a third stream water stage for processing to obtain an output result.

Taking the first operation instruction as a vector translation instruction VTRAN as an example, a translation vector of a given vector is calculated. In particular implementations, given a vector X (X)₁,x₂,x₃,..x_n) And a translation vector Y (d)_x1,d_x2,…d_xn) (also referred to as a translation factor), the translated vector for a given vector X is calculated as follows.

Accordingly, the instruction format of the vector translation instruction VTRAN specifically is:

with reference to the foregoing embodiments, the arithmetic unit may obtain the vector translation instruction VTRAN, decode the vector translation instruction VTRAN, read the vector X and the translation vector Y from the register unit, and use the first-stage multiplexer to select the non-linear arithmetic unit to perform 1-complementing operation on the vector X to obtain a first result (specifically, a vector to be translated formed after 1-complementing operation); meanwhile, a second pipeline-level multiplexer is used for selecting a nonlinear operator to execute translation matrix construction operation (optional, and also involving the moving operation of matrix elements) according to the read translation vector Y so as to correspondingly obtain a second result (specifically, a translation matrix); and inputting the first result and the second result into a multiplexer of a third pipeline stage to select a matrix vector multiplication operator to perform matrix multiplication vector operation (specifically to perform matrix multiplication vector operation on a vector to be translated and a translation matrix) to obtain a third result (namely an output result). Optionally, the third result is stored in a storage medium. Optionally, the first result may be input into the second pipeline stage, and the second pipeline stage waits for the translation matrix to be constructed by the second pipeline stage to obtain the second result together. Namely, the second result comprises a translation matrix and a vector to be translated, so that the second result is only input into a third stream water stage for processing to obtain an output result.

Taking the first operation instruction as a vector scaling instruction VZOOM as an example, a scaling vector of a given vector is calculated. In particular implementations, given a vector X (X)₁,x₂,x₃,..x_n) And a scaling factor a, which calculates the scaled vector for a given vector X as follows.

Accordingly, the instruction format of the vector scaling instruction VZOOM is specifically:

with reference to the foregoing embodiment, the arithmetic unit may obtain the vector scaling instruction VZOOM, decode the vector scaling instruction VZOOM, read the vector X and the scaling factor a from the register unit, and use the first-stage multiplexer to select the non-linear arithmetic unit to perform the 1-complementing operation on the vector X to obtain a first result (specifically, a vector to be scaled formed after the 1-complementing operation); meanwhile, a multi-path selector of a second pipeline stage is used for selecting a non-linear arithmetic unit to execute scaling matrix construction operation (optional, and also relates to moving operation of matrix elements) according to the read scaling factor a so as to correspondingly obtain a second result (specifically, a scaling matrix); and inputting the first result and the second result into a multiplexer of a third pipeline stage to select a matrix vector multiplication operator to perform matrix multiplication vector operation (specifically, to perform matrix multiplication vector operation on a vector to be scaled and a scaling matrix) to obtain a third result (namely, an output result). Optionally, the third result is stored in a storage medium. Optionally, the first result may be input into the second pipeline stage, and the first result and the second result are obtained together after the scaling matrix is constructed by the second pipeline stage. Namely, the second result comprises a scaling matrix and a vector to be scaled, so that the second result is only input into a third stream level for processing to obtain an output result.

Taking the first operation instruction as a vector clipping instruction VSHEAR as an example, a clipping matrix of a given matrix is calculated. In particular implementations, given a vector X (X)₁,x₂,x₃,..x_n) And a shear vector Y (d)₁,d₂,…d_n) (also referred to as clipping factor), the clipped vector for a given vector X is calculated as follows.

Correspondingly, the instruction format of the vector clipping instruction vsear is specifically:

with reference to the foregoing embodiment, the arithmetic unit may obtain the vector clipping instruction VSHEAR, decode the vector clipping instruction VSHEAR, read the vector X and the clipping vector Y from the register unit, and use the first-stage multiplexer to select the non-linear arithmetic unit to perform 1-complementing operation on the vector X to obtain a first result (specifically, a to-be-clipped vector formed after the 1-complementing operation); meanwhile, a second pipeline-level multiplexer is used for selecting a nonlinear operator to execute shearing matrix construction operation (optional, movement operation of matrix elements is also involved) according to the read shearing vector Y so as to correspondingly obtain a second result (specifically, a shearing matrix); and inputting the first result and the second result into a multiplexer of a third pipeline stage to select a matrix vector multiplication operator to perform matrix multiplication vector operation (specifically to perform matrix multiplication vector operation on the to-be-sheared vector and the sheared matrix) to obtain a third result (namely an output result). Optionally, the third result is stored in a storage medium. Optionally, the first result may be input into the second pipeline stage, and the second result is obtained together after the second pipeline stage constructs the cut matrix. Namely, the second result comprises a shearing matrix and a vector to be sheared, so that the second result is only input into a third stream level for processing to obtain an output result.

It should be noted that the fetching and decoding of the various operation instructions will be described in detail later. It should be understood that, by adopting the structure of the computing device to realize the computation of each operation instruction (such as the two-dimensional vector rotation instruction SVRO, the three-dimensional vector rotation instruction TVRO, etc.), the following advantages can be obtained: the vector format stored at certain intervals is supported, so that the execution overhead of converting the vector format and the space occupation of storing intermediate results are avoided; and the transformation information is supported to be stored in a scalar value, so that the storage overhead of the rotation matrix is avoided.

The set length in the above operation instruction (i.e. vector operation instruction/first operation instruction) can be set by the user, and in an alternative embodiment, the user can set the set length to one value, but in practical application, the user can also set the set length to multiple values. The specific value and the number of the set length are not limited in the embodiments of the present invention. In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is further described in detail below with reference to the accompanying drawings in combination with specific embodiments.

Referring to fig. 7, fig. 7 is a block diagram of another computing device 50 according to an embodiment of the present disclosure. As shown in fig. 7, the computing device 50 includes: a storage medium 501, a register unit 502 (preferably, a scalar data storage unit, a scalar register unit), an operation unit 503 (may also be referred to as a matrix operation unit 503), and a control unit 504;

a storage medium 501 for storing a matrix (which may also be a vector);

a scalar data storage unit 502 for storing scalar data including at least: the storage address of the matrix (which may also be a vector) in the storage medium;

a control unit 504, configured to control the arithmetic unit to obtain a first arithmetic instruction, where the first arithmetic instruction is used to implement an operation between a vector and a matrix, and the first arithmetic instruction includes a vector read instruction required to execute the instruction;

an arithmetic unit 503, configured to send a read command to the storage medium according to the vector read instruction; and executing the first operation instruction on the vector according to the vector corresponding to the vector reading instruction read by adopting a batch reading mode.

Optionally, the vector reading indication includes: a memory address of a vector required by the instruction or an identification of a vector required by the instruction.

Optionally if the vector read indicates the identity of the vector required by the instruction,

a control unit 504, configured to control the arithmetic unit to read, according to the identifier, the storage address corresponding to the identifier from the register unit in a unit reading manner, control the arithmetic unit to send a read command for reading the storage address to the storage medium, and obtain the vector in a batch reading manner.

Optionally, the operation unit 503 is specifically configured to execute the first operation instruction on the vector in a multi-level pipeline computing manner.

Optionally, each pipeline stage in the multi-stage pipeline stages comprises at least one arithmetic unit,

an operation unit 503, specifically configured to calculate the vector by using a first selection operator in the first-stage pipeline stage to obtain a first result according to the selection of the multi-path selector, input the first result to a second selection operator in the second-stage pipeline stage to perform calculation to obtain a second result, and so on until the i-1 th result is input to the i-th selection operator in the i-th pipeline stage to perform calculation to obtain an i-th result; inputting the ith result into the storage medium for storage; wherein the number i of the multiple pipeline stages is determined according to a computation topology of the first operation instruction, and i is a positive integer.

Optionally, each pipeline stage in the multiple pipeline stages is configured with a corresponding multiplexer, and the multiplexer is provided with an empty option, where the empty option is used to indicate that the k-th pipeline stage connected to the multiplexer and subsequent (k + 1) -th pipeline stages do not perform a calculation operation, where k is a positive integer less than or equal to i.

Optionally, the number of the operators included in each of the multiple pipeline stages and the number of the operators are set by a user side or the computing device side in a self-defined manner.

Optionally, each pipeline stage in the multiple pipeline stages includes a preset fixed arithmetic unit, the fixed arithmetic units in each pipeline stage are different,

an operation unit 503, configured to calculate the vector by using a fixed operator in the first-stage pipeline stage to obtain a first result, input the first result to a second fixed operator in the second-stage pipeline stage to perform calculation to obtain a second result, and so on until the (i-1) th result is input to the fixed operator in the i-stage pipeline stage to perform calculation to obtain an ith result; inputting the ith result into the storage medium for storage; wherein the number i of the multiple pipeline stages is determined according to a computation topology of the first operation instruction, and i is a positive integer.

Optionally, the arithmetic unit in each pipeline stage in the multiple pipeline stages includes any one or a combination of more than one of the following items: a matrix addition operator, a matrix multiplication operator, a matrix vector multiplication operator, a nonlinear operator, and a matrix comparison operator.

Optionally, the first operation instruction is a two-dimensional vector rotation instruction SVRO or a three-dimensional vector rotation instruction TVRO,

an operation unit 503, configured to perform vector 1 complementing operation on the vector by using a non-linear operator in a first-stage pipeline stage according to selection of the multiplexer to obtain a first result, perform matrix element shifting calculation according to the acquired rotation center and rotation angle by using a non-linear operator in a second-stage pipeline stage to obtain a second result, and input the first result and the second result into a matrix vector multiplication operator in a third-stage pipeline stage to perform matrix multiplication to obtain a third result; and inputting the third result to the storage medium for storage.

Optionally, the first operation instruction is any one of the following instructions: a vector translation instruction VTRAN, a vector scaling instruction VZOOM, a vector clipping instruction VSHEAR,

an operation unit 503, configured to perform a vector complement 1 operation on the vector by using a nonlinear operator in the first-stage pipeline stage according to the selection of the multiplexer, to obtain a first result, and correspondingly perform any one of the following operations by using the nonlinear operator in the second-stage pipeline stage to obtain a second result: executing translation matrix construction operation according to the obtained translation factor, executing scaling matrix construction operation according to the obtained scaling factor, executing shearing matrix construction operation according to the obtained shearing factor, inputting the first result and the second result into a matrix vector multiplication arithmetic unit in a third-stage pipeline stage, and executing matrix multiplication vector calculation to obtain a third result; and inputting the third result to the storage medium for storage.

Optionally, the computing apparatus further includes:

a cache unit 505, configured to cache an operation instruction to be executed;

the control unit 504 is configured to cache an operation instruction to be executed in the cache unit 504.

Optionally, the control unit 504 is configured to determine whether an association relationship exists between the first operation instruction and a second operation instruction before the first operation instruction, if the association relationship exists between the first operation instruction and the second operation instruction, cache the first operation instruction in the cache unit, and after the second operation instruction is executed, extract the first operation instruction from the cache unit and transmit the first operation instruction to the operation unit;

extracting a first storage address interval of a required vector in the first operation instruction according to the first operation instruction, extracting a second storage address interval of the required vector in the second operation instruction according to the second operation instruction, if the first storage address interval and the second storage address interval have an overlapped area, determining that the first operation instruction and the second operation instruction have an association relation, and if the first storage address interval and the second storage address interval do not have an overlapped area, determining that the first operation instruction and the second operation instruction do not have an association relation.

Optionally, the control unit 503 may be configured to obtain an operation instruction from the instruction cache unit, process the operation instruction, and provide the processed operation instruction to the operation unit. The control unit 503 may be divided into three modules, which are: an instruction fetching module 5031, a decoding module 5032 and an instruction queue module 5033,

the instruction fetching module 5031 is configured to obtain an operation instruction from the instruction cache unit;

a decoding module 5032, configured to decode the obtained operation instruction;

the instruction queue 5033 is configured to store the decoded operation instructions sequentially, and to cache the decoded instructions in consideration of possible dependencies among registers of different instructions, and to issue the instructions when the dependencies are satisfied.

Referring to fig. 8 and fig. 8 are flowcharts illustrating a computing device according to an embodiment of the present invention to execute an operation instruction, as shown in fig. 8, a hardware structure of the computing device refers to the structure shown in fig. 7, and a process of executing a two-dimensional/three-dimensional vector rotation instruction, taking a scratch pad as an example of a storage medium shown in fig. 7, includes:

in step S601, the computing device controls the instruction fetching module to fetch a two-dimensional/three-dimensional vector rotation instruction, and sends the two-dimensional/three-dimensional vector rotation instruction to the decoding module.

Step S602, the decoding module decodes the two-dimensional/three-dimensional vector rotation instruction and sends the two-dimensional/three-dimensional vector rotation instruction to the instruction queue.

Step S603, in the instruction queue, the two-dimensional/three-dimensional vector rotation instruction needs to obtain, from the scalar register file, data in scalar registers corresponding to six operation domains in the instruction, where the data includes an input vector address, an input vector length, an input rotation center vector address, an input rotation center scalar, an output vector address, and an output vector length.

Step S604, the control unit determines whether the two-dimensional/three-dimensional vector rotation instruction and the operation instruction before the two-dimensional/three-dimensional vector rotation instruction have an association relationship, if so, stores the two-dimensional/three-dimensional vector rotation instruction in the cache unit, and if not, transmits the two-dimensional/three-dimensional vector rotation instruction to the operation unit.

In step S605, the arithmetic unit fetches the required vector data (optionally, including scalar data) from the register (preferably, a scratch pad memory or a scalar register file) according to the data in the scalar registers corresponding to the six operation domains, and then completes the vector rotation operation in the arithmetic unit.

In step S606, after the arithmetic unit completes the operation, the result is written into the designated address of the memory (preferably, a scratch pad memory or a scalar register file), and the two-dimensional/three-dimensional vector rotation instruction in the reorder buffer is submitted.

Optionally, in step S605, when the operation unit executes the vector rotation operation, the calculation device may adopt a non-linear operator to execute a 1-complementing operation and a rotation matrix constructing operation, and then adopt a matrix vector multiplication operator to execute a matrix-by-vector calculation to obtain a corresponding rotation vector.

In a specific implementation, after the decoding module decodes the two-dimensional/three-dimensional vector rotation instruction, according to a control signal generated by decoding, the vector obtained in S603 is input to the non-linear operator selected by the first-stage pipeline-level multiplexer to perform a 1 complementing operation, so as to obtain a first result (i.e., a vector to be rotated); and simultaneously inputting the rotation center and the rotation angle obtained in S603 to a non-linear operator selected by a multiplexer in a second-stage pipeline stage to execute a rotation matrix to construct a second result (namely, a rotation matrix), and then inputting the first result and the second result to a matrix vector multiplication operator selected by the multiplexer in a third-stage pipeline stage to execute a matrix multiplication vector operation to obtain a third result. Further, the multiplexer of the fourth stage pipeline stage can know that the stage is an empty option according to the control signal. Correspondingly, the third result is directly transmitted to an output end as an output result.

The operation instruction in fig. 8 is exemplified by a two-dimensional/three-dimensional vector rotation instruction, and in practical applications, the two-dimensional/three-dimensional vector rotation instruction in the embodiment shown in fig. 8 may be replaced by a vector operation/operation instruction such as a vector translation instruction, a vector scaling instruction, a vector clipping instruction, and the like, which is not described herein again.

Embodiments of the present application also provide a computer storage medium, wherein the computer storage medium stores a computer program for electronic data exchange, and the computer program enables a computer to execute some or all of the steps of any implementation described in the above method embodiments.

Embodiments of the present application also provide a computer program product comprising a non-transitory computer readable storage medium storing a computer program operable to cause a computer to perform any, some or all of the steps of any of the implementations described in the above method embodiments.

An embodiment of the present application further provides an acceleration apparatus, including: a memory: executable instructions are stored; a processor: for executing the executable instructions in the memory unit, and when executing the instructions, operate according to the embodiments described in the above method embodiments.

Wherein the processor may be a single processing unit, but may also comprise two or more processing units. In addition, the processor may also include a general purpose processor (CPU) or a Graphics Processor (GPU); it may also be included in a Field Programmable Gate Array (FPGA) or Application Specific Integrated Circuit (ASIC) to set up and operate the neural network. The processor may also include on-chip memory (i.e., including memory in the processing device) for caching purposes.

In some embodiments, a chip is also disclosed, which includes the neural network processor for performing the above method embodiments.

In some embodiments, a chip packaging structure is disclosed, which includes the above chip.

In some embodiments, a board card is disclosed, which includes the above chip package structure.

In some embodiments, an electronic device is disclosed that includes the above board card.

The electronic device comprises a data processing device, a robot, a computer, a printer, a scanner, a tablet computer, an intelligent terminal, a mobile phone, a vehicle data recorder, a navigator, a sensor, a camera, a server, a cloud server, a camera, a video camera, a projector, a watch, an earphone, a mobile storage, a wearable device, a vehicle, a household appliance, and/or a medical device.

The vehicle comprises an airplane, a ship and/or a vehicle; the household appliances comprise a television, an air conditioner, a microwave oven, a refrigerator, an electric cooker, a humidifier, a washing machine, an electric lamp, a gas stove and a range hood; the medical equipment comprises a nuclear magnetic resonance apparatus, a B-ultrasonic apparatus and/or an electrocardiograph.

It should be noted that, for simplicity of description, the above-mentioned method embodiments are described as a series of acts or combination of acts, but those skilled in the art will recognize that the present application is not limited by the order of acts described, as some steps may occur in other orders or concurrently depending on the application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are exemplary embodiments and that the acts and modules referred to are not necessarily required in this application.

In the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus may be implemented in other manners. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one type of division of logical functions, and there may be other divisions when actually implementing, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of some interfaces, devices or units, and may be an electric or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit may be implemented in the form of hardware, or may be implemented in the form of a software program module.

The integrated units, if implemented in the form of software program modules and sold or used as stand-alone products, may be stored in a computer readable memory. Based on such understanding, the technical solution of the present application may be substantially implemented or a part of or all or part of the technical solution contributing to the prior art may be embodied in the form of a software product stored in a memory, and including several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method described in the embodiments of the present application. And the aforementioned memory comprises: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.

Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by associated hardware instructed by a program, which may be stored in a computer-readable memory, which may include: flash Memory disks, Read-Only memories (ROMs), Random Access Memories (RAMs), magnetic or optical disks, and the like.

The foregoing detailed description of the embodiments of the present application has been presented to illustrate the principles and implementations of the present application, and the above description of the embodiments is only provided to help understand the method and the core concept of the present application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims

1. A computing method applied in a computing apparatus including a storage medium, a register unit, and a matrix operation unit, the method comprising:

the computing device controls the matrix operation unit to read the vector corresponding to the vector reading instruction from the storage medium in a batch reading mode, the computing device controls the matrix operation unit to calculate the vector by using a first selection operator in a first-stage pipeline stage according to the selection of a multi-channel selector to obtain a first result, the first result is input to a second selection operator in a second-stage pipeline stage to perform calculation to obtain a second result, and the like, until the ith-1 result is input to an ith selection operator in an ith-stage pipeline stage to perform calculation to obtain an ith result;

inputting the ith result into the storage medium for storage;

wherein the ith result is an output vector, the number i of multi-stage pipeline stages is determined according to a calculation topology of the first operation instruction, and i is a positive integer;

alternatively, the first and second electrodes may be,

the computing device controls the matrix operation unit to calculate the vector by using a fixed arithmetic unit in a first-level pipeline level to obtain a first result, inputs the first result into a fixed arithmetic unit in a second-level pipeline level to perform calculation to obtain a second result, and so on until the (i-1) th result is input into a fixed arithmetic unit in an i-level pipeline level to perform calculation to obtain an ith result;

inputting the ith result into the storage medium for storage;

the number i of the multi-stage pipeline stages is determined according to a calculation topological structure of the first operation instruction, and i is a positive integer;

each pipeline stage in the multi-stage pipeline stages comprises a preset fixed arithmetic unit, and the fixed arithmetic units in each pipeline stage are different.

2. The method according to claim 1, wherein each of the multiple pipeline stages is configured with a corresponding multiplexer, and the multiplexer is provided with an empty option for indicating that no computation operation is performed on a k-th pipeline stage connected to the multiplexer and subsequent (k + 1) -th pipeline stages, where k is a positive integer less than or equal to i;

the number of the arithmetic units and the arithmetic units included in each pipeline stage in the multi-stage pipeline stages are set by a user side or the computing device side in a self-defined mode; alternatively, the operator in each of the multiple pipeline stages comprises any one or a combination of more than one of: a matrix addition operator, a matrix multiplication operator, a matrix vector multiplication operator, a nonlinear operator, and a matrix comparison operator.

3. The method of claim 1 or 2, wherein the first operation instruction comprises any one of: a two-dimensional vector rotation instruction SVRO, a three-dimensional vector rotation instruction TVRO, a vector translation instruction VTRAN, a vector scaling instruction VZOOM and a vector shearing instruction VSHEAR;

the instruction format of the first operation instruction comprises at least one operation code and at least one operation field, wherein the at least one operation code is used for indicating the function of the first operation instruction, the at least one operation field is used for indicating data information of the first operation instruction, and the data information comprises an immediate or a register number and is used for storing the matrix reading indication and the length of the matrix; wherein the at least one opcode includes a first opcode to indicate a type of the first arithmetic instruction and a second opcode to indicate a function of the first arithmetic instruction.

4. The method of claim 1, wherein the first operation instruction is a two-dimensional vector rotation instruction SVRO or a three-dimensional vector rotation instruction TVRO,

the computing device controls the matrix operation unit to perform vector 1 complementing operation on the vector by using a nonlinear operator in a first-stage pipeline stage according to the selection of the multi-path selector to obtain a first result, simultaneously performs rotation matrix construction operation by using a nonlinear operator in a second-stage pipeline stage according to the obtained rotation center and the obtained rotation angle to obtain a second result, and inputs the first result and the second result into a matrix vector multiplication operator in a third-stage pipeline stage to perform matrix multiplication vector calculation to obtain a third result; and inputting the third result to the storage medium for storage.

5. The method of claim 1, wherein the first operation instruction is any one of: a vector translation instruction VTRAN, a vector scaling instruction VZOOM, a vector clipping instruction VSHEAR,

the computing device controls the matrix operation unit to perform vector 1 complementing operation on the vector by using a nonlinear operator in a first level pipeline stage according to the selection of the multiplexer to obtain a first result, and simultaneously, the nonlinear operator in a second level pipeline stage correspondingly executes any one of the following operations to obtain a second result: executing translation matrix construction operation according to the obtained translation factor, executing scaling matrix construction operation according to the obtained scaling factor, executing shearing matrix construction operation according to the obtained shearing factor, inputting the first result and the second result into a matrix vector multiplication arithmetic unit in a third-stage pipeline stage, and executing matrix multiplication vector calculation to obtain a third result; and inputting the third result to the storage medium for storage.

6. A computing device, comprising a storage medium, a register unit, a matrix operation unit, and a controller unit;

the storage medium is used for storing vectors;

the register unit is configured to store scalar data, where the scalar data at least includes: a storage address of the vector within the storage medium;

the controller unit is configured to control the matrix operation unit to obtain a first operation instruction, where the first operation instruction is used to implement an operation between a vector and a matrix, the first operation instruction includes a vector read instruction required to execute the instruction, the required vector is at least one vector, and the at least one vector is a vector with the same length or a vector with different lengths;

the matrix operation unit is used for sending a reading command to the storage medium according to the vector reading instruction; reading the vector corresponding to the vector reading instruction by adopting a batch reading mode, controlling the matrix operation unit to calculate the vector by utilizing a first selection operator in a first-level pipeline stage according to the selection of a multi-path selector by the calculation device to obtain a first result, inputting the first result into a second selection operator in a second-level pipeline stage to perform calculation to obtain a second result, and repeating the steps until the ith-1 result is input into the ith selection operator in the ith-level pipeline stage to perform calculation to obtain the ith result; inputting the ith result into the storage medium for storage; wherein the ith result is an output vector, the number i of multi-stage pipeline stages is determined according to a calculation topology of the first operation instruction, and i is a positive integer; or the computing device controls the matrix operation unit to calculate the vector by using a fixed arithmetic unit in a first-level pipeline level to obtain a first result, inputs the first result into a fixed arithmetic unit in a second-level pipeline level to perform calculation to obtain a second result, and so on until the (i-1) th result is input into a fixed arithmetic unit in an i-level pipeline level to perform calculation to obtain an ith result; inputting the ith result into the storage medium for storage; the number i of the multi-stage pipeline stages is determined according to a calculation topological structure of the first operation instruction, and i is a positive integer; each pipeline stage in the multi-stage pipeline stages comprises a preset fixed arithmetic unit, and the fixed arithmetic units in each pipeline stage are different.

7. A chip, characterized in that it comprises a computing device as claimed in claim 6 above.

8. A computer-readable storage medium, characterized in that the computer storage medium stores a computer program comprising program instructions that, when executed by a processor, cause the processor to perform the method according to any of claims 1-5.