CN116384505A - Data processing method and device, storage medium and electronic equipment - Google Patents
Data processing method and device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN116384505A CN116384505A CN202310164810.XA CN202310164810A CN116384505A CN 116384505 A CN116384505 A CN 116384505A CN 202310164810 A CN202310164810 A CN 202310164810A CN 116384505 A CN116384505 A CN 116384505A
- Authority
- CN
- China
- Prior art keywords
- target
- execution
- target model
- data
- determining
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003860 storage Methods 0.000 title claims abstract description 22
- 238000003672 processing method Methods 0.000 title abstract description 8
- 238000011156 evaluation Methods 0.000 claims abstract description 73
- 238000000034 method Methods 0.000 claims abstract description 66
- 238000012545 processing Methods 0.000 claims abstract description 41
- 230000008569 process Effects 0.000 claims description 23
- 238000004590 computer program Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000010586 diagram Methods 0.000 description 12
- 230000006870 function Effects 0.000 description 9
- 230000006872 improvement Effects 0.000 description 8
- 238000013136 deep learning model Methods 0.000 description 6
- 238000010606 normalization Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 239000002131 composite material Substances 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 229920001296 polysiloxane Polymers 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000010979 ruby Substances 0.000 description 1
- 229910001750 ruby Inorganic materials 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Stored Programmes (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The specification discloses a data processing method, a data processing device, a storage medium and electronic equipment, and firstly, a target model and equipment parameters of the target equipment are obtained. And secondly, generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the equipment parameters. Then, for each candidate execution scheme, the operation required when the target model is executed through the candidate execution scheme, so as to determine the execution evaluation parameter monitored when the target device obtains the output result of the target model through executing the operation, and the execution evaluation parameter is used as the execution evaluation parameter of the candidate execution scheme. And finally, determining a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme, and adjusting the operation required by the target equipment for executing the target model so as to execute data processing through the target model according to the adjusted operation. The method can ensure that the target equipment can effectively execute data processing through the target model.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a data processing method, an apparatus, a storage medium, and an electronic device.
Background
With the advent of deep learning, deep learning has found wide application in image classification, natural language processing, autopilot, augmented reality, privacy data protection, and other AI fields.
At present, the deep learning model generally needs to rely on a server to process data, and how to ensure that the server can effectively execute the operation tasks required by the deep learning model is a problem to be solved urgently.
Disclosure of Invention
The present disclosure provides a data processing method, apparatus, storage medium, and electronic device, so as to implement an operation task required by a server to effectively execute a deep learning model.
The technical scheme adopted in the specification is as follows:
the present specification provides a method of data processing, comprising:
acquiring a target model and equipment parameters of target equipment, wherein the target equipment is used for executing operation required by the target model;
generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the equipment parameters;
For each candidate execution scheme, executing operation required by running the target model through the candidate execution scheme to determine an execution evaluation parameter monitored when the target equipment obtains an output result of the target model through executing the operation, wherein the execution evaluation parameter is used as an execution evaluation parameter corresponding to the candidate execution scheme;
and determining a target execution scheme according to execution evaluation parameters corresponding to each candidate execution scheme, adjusting operation required by the target equipment to execute the target model according to the target execution scheme, and executing data processing according to the adjusted operation through the target model.
Optionally, generating, according to the device parameter, each candidate execution scheme adopted when the target device executes the operation required by the target model, specifically including:
determining the dimension number of input data of the target model, and determining the maximum dimension number of data which can be processed by a single processor in the target device;
determining the number of processors required by the target equipment when the target equipment executes the operation required by the target model according to the dimension number of the input data of the target model and the maximum dimension number;
And generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the number of processors required by the target equipment when executing the operation required by the target model.
Optionally, determining the number of processors required by the target device when executing the operation required by the target model according to the dimension number of the input data of the target model and the maximum dimension number specifically includes:
determining a dimension ratio corresponding to the target model according to the input data and the output data of the target model, wherein the dimension ratio is used for representing the ratio between the dimension number of the input data and the dimension number of the output data of the target model;
if the dimension ratio exceeds the dimension ratio threshold corresponding to the target device, splitting the input data according to the dimension ratio and the maximum dimension number, and determining the quantity of the split data;
and determining the number of processors required by the target equipment when the target equipment executes the operation required by the target model according to the number of the split data.
Optionally, generating, according to the device parameter, each candidate execution scheme adopted when the target device executes the operation required by the target model, specifically including:
Determining the dimension number of input data of the target model, and determining the maximum dimension number of data which can be processed by a single processor in the target device;
splitting the input data according to the maximum dimension number and the dimension number of the input data, and determining the dimension number of the split data;
determining the number of the split data which can be processed in parallel by a single processor in the target equipment according to the number of the dimensions of the split data and the maximum number of the dimensions;
determining the number of threads adopted by a single processor in the target equipment for executing parallel operation according to the determined number of the split data which can be processed in parallel by the single processor in the target equipment;
and generating each candidate execution scheme according to the thread number.
Optionally, generating, according to the device parameter, each candidate execution scheme adopted when the target device executes the operation required by the target model, specifically including:
determining each data with an operation sequence in the operation process as related data;
adjusting the operation sequence of other data except related data, and determining each operation sequence of each data processed by a processor when the target equipment executes the operation required by the target model;
And generating each candidate execution scheme according to each operation sequence.
Optionally, determining the target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme specifically includes:
acquiring a preset basic scheme;
determining an execution evaluation parameter monitored when the target equipment obtains an output result of the target model by executing the basic scheme, wherein the execution evaluation parameter is used as an execution evaluation parameter corresponding to the basic scheme;
and determining a target execution scheme according to the execution evaluation parameters corresponding to the basic scheme and the execution evaluation parameters corresponding to each candidate execution scheme.
Optionally, the method further comprises:
receiving a service request of a user;
and according to the service request, inputting service data corresponding to the service request into the target model, and executing operation required by the target model for processing the service data through the target execution scheme to obtain a service result.
The present specification provides an apparatus for data processing, comprising:
the device comprises an acquisition module, a calculation module and a calculation module, wherein the acquisition module is used for acquiring a target model and device parameters of target device, and the target device is used for executing calculation operation required by the target model;
The generation module is used for generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the equipment parameters;
the determining module is used for determining, for each candidate execution scheme, an operation required when the target model is operated through the candidate execution scheme so as to determine an execution evaluation parameter monitored when the target device obtains an output result of the target model through executing the operation, wherein the execution evaluation parameter is used as an execution evaluation parameter corresponding to the candidate execution scheme;
and the adjusting module is used for determining a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme, adjusting the operation required by the target equipment for executing the target model according to the target execution scheme, and executing data processing according to the adjusted operation and the target model.
The present specification provides a computer readable storage medium storing a computer program which when executed by a processor performs the method of data processing described above.
The present specification provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing a method of data processing as described above when executing the program.
The above-mentioned at least one technical scheme that this specification adopted can reach following beneficial effect:
in the data processing method provided in the present specification, first, a target model and device parameters of a target device for performing arithmetic operations required for the target model are acquired. And secondly, generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the equipment parameters. Then, for each candidate execution scheme, the operation required when the target model is executed through the candidate execution scheme, so as to determine the execution evaluation parameters monitored when the target device obtains the output result of the target model through the operation, and the execution evaluation parameters are used as the execution evaluation parameters corresponding to the candidate execution scheme. And finally, determining a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme, and adjusting the operation required by the target equipment for executing the target model through the target execution scheme so as to execute data processing through the target model according to the adjusted operation.
According to the method, candidate execution schemes adopted when the target equipment executes operation required by the target model can be generated according to the equipment parameters. And secondly, determining the execution evaluation parameters corresponding to the candidate execution schemes through the candidate execution schemes. And finally, determining a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme. Therefore, the target device is ensured to be capable of effectively executing data processing through the target model.
Drawings
The accompanying drawings, which are included to provide a further understanding of the specification, illustrate and explain the exemplary embodiments of the present specification and their description, are not intended to limit the specification unduly. Attached at
In the figure:
FIG. 1 is a flow chart of a method for data processing according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a target execution scheme according to an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of a data processing apparatus according to an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure.
Detailed Description
For the purposes of making the objects, technical solutions and advantages of the present specification more apparent, the technical solutions of the present specification will be clearly and completely described below with reference to specific embodiments of the present specification and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present specification. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.
The following describes in detail the technical solutions provided by the embodiments of the present specification with reference to the accompanying drawings.
Fig. 1 is a flow chart of a data processing method in the present specification, specifically including the following steps:
s100: and acquiring a target model and equipment parameters of target equipment, wherein the target equipment is used for executing operation required by the target model.
In the embodiment of the present disclosure, the execution subject of the data processing method may be a target device, and the target device may refer to an electronic device such as a server or a desktop computer. For convenience of description, a method of data processing provided in the present specification will be described below with only a target device as an execution subject.
In the embodiment of the present specification, the target device may acquire the target model and device parameters of the target device, where the target device is used to perform an arithmetic operation required by the target model. The object model mentioned here may refer to various deep learning models for executing the service, and may refer to network structures inside various deep learning models, for example, batch normalization (batch Normalization, BN), layer normalization (layer Normalization, LN), and the like. The present specification does not limit the network structure inside the deep learning model.
The device parameters of the target device mentioned herein may refer to hardware parameters such as parameters of CPU cores, the number of CPU cores, parameters of GPU cores, the number of GPU cores, parameters of registers, the number of registers, and the like.
Of course, the device parameters of the target device mentioned herein may also refer to the distribution situation of each hardware in the target device, the power consumption situation of each hardware, and so on.
Specifically, an equipment parameter configuration table of the target equipment can be preset in the target equipment, and the equipment parameter configuration table can record equipment parameters corresponding to the target equipment with different models. The target device can determine the device parameters corresponding to the target device from a pre-configured device parameter configuration table through the model of the target device.
Of course, the manufacturer corresponding to the target device may provide a query interface for querying the device parameter corresponding to the target device, so that the target device may call the query interface of the device parameter configuration table configured in advance to obtain the device parameter corresponding to the target device. For example, the vendor corresponding to the target device may provide an API interface for querying the device parameter corresponding to the target device, so that the device parameter corresponding to the target device may be obtained through the API interface provided by the vendor corresponding to the target device.
S102: and generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the equipment parameters.
In practical applications, the target model usually needs to rely on the target device for data processing. In order to improve the efficiency of training or applying the target model, an execution scheme applied by the target device during data processing is generally required to be adjusted so as to improve the operation efficiency of the target device during the operation required by the target model. However, the implementation scheme generally requires an expert to write the algorithm based on the target model manually in a large amount of time, and is low in efficiency.
Because the execution scheme written manually by an expert cannot adjust the operation process on different target devices, and cannot adjust the operation required by different target models, the efficiency of determining the execution scheme is low.
Based on this, the target device may determine the target execution scheme by generating each candidate execution scheme to determine the execution evaluation parameters after executing each candidate execution scheme.
In the embodiment of the present specification, the target device may generate, according to the device parameters, each candidate execution scheme adopted when the target device performs the arithmetic operation required by the target model. Since the actual representation of the object model is code, deployed in the object device, the candidate execution scenario referred to herein may refer to the object device executing the object model in a different manner, that is, running the code in a different manner by the object device.
The candidate execution scheme comprises a plurality of influence factors which influence the execution evaluation parameters monitored when the output result of the target model is influenced. For example, the number of processors employed when the target device performs the arithmetic operation required for the target model, the number of data that can be processed in parallel by a single processor when the target device performs the arithmetic operation required for the target model, the arithmetic order of each data when the target device performs the arithmetic operation required for the target model, and the like. The target device may generate each candidate execution scheme adopted when the target device executes the arithmetic operation required by the target model according to at least one of the above influencing factors.
In practical applications, when data processing is performed on the target model, only one processor is usually started to execute the operation required by the data processing on the target model, however, the operation time required by the method is long, which results in long training time of the target model. Based on the method, the target device can split the input data into a plurality of data with the same dimension number, operate the plurality of data with the same dimension number through a plurality of processors to obtain a plurality of intermediate operation results, and determine output data according to the plurality of intermediate operation results so as to improve the efficiency of processing the data of the target model by a method of starting the plurality of processors simultaneously.
In embodiments of the present disclosure, the target device may determine the number of dimensions of the input data of the target model, as well as the maximum number of dimensions of the data that can be processed by a single processor in the target device.
Second, the target device may determine the number of processors required by the target device to perform the arithmetic operation required by the target model according to the number of dimensions and the maximum number of dimensions of the input data of the target model.
For example, if the number of dimensions of the input data of the target model is one thousand dimensions and the maximum number of dimensions of the data that can be processed by a single processor is one hundred dimensions, at least ten processors are required to determine the number of processors required by the target device to perform the arithmetic operation required by the target model. Of course, the number of processors required by the target device to perform the computational operations required by the target model may be determined based on at least one of the influencing factors described above.
Finally, the target device may generate each candidate execution scheme adopted when the target device executes the operation required by the target model according to the number of processors required by the target device when executing the operation required by the target model.
Further, the target device may determine, according to input data and output data of the target model, a dimension ratio corresponding to the target model, where the dimension ratio is used to represent a ratio between a dimension number of the input data and a dimension number of the output data of the target model.
If the dimension ratio exceeds the dimension ratio threshold corresponding to the target device, splitting the input data according to the dimension ratio and the maximum dimension number, and determining the quantity of the split data.
The target device may then determine the number of processors needed by the target device to perform the arithmetic operations needed by the target model based on the number of data split.
Specifically, the target device may determine various intermediate operation results, and the number of dimensions of the intermediate operation results in each candidate execution scheme may be different, where one candidate execution scheme corresponds to the number of dimensions of one intermediate operation result.
Secondly, for each intermediate operation result, the target device can determine the dimension ratio corresponding to the intermediate operation result according to the ratio between the dimension of the input data of the target model and the dimension number of the intermediate operation result.
Then, if the dimension ratio of the intermediate operation result is smaller than the dimension ratio threshold corresponding to the target device, the target device can split the input data according to the dimension number of the intermediate operation result, and determine the number of the split data.
Finally, the target device may determine, according to the number of the split data, the number of processors required by the target device to perform the arithmetic operation required by the target model.
For example, the input data is one-thousand-dimensional data, the output data is one-dimensional data, if the number of dimensions of the intermediate operation result is ten-dimensional, the maximum number of dimensions that can be processed by a processor in the target device at a time is one hundred, and the dimension ratio corresponding to the intermediate operation result is not greater than the dimension ratio threshold corresponding to the target device, the target device determines the number of enabled processors (one to ten) according to the number of dimensions (ten) corresponding to the intermediate operation result and the number of dimensions (one thousand) of the input data. That is, by the number of processors enabled, ten-dimensional intermediate operation results are determined from one thousand-dimensional input dimensions, and one-dimensional output dimensions are determined from the ten-dimensional intermediate operation results.
The intermediate operation result obtained here may be stored in a global memory, and all processors may obtain data from the global memory. The intermediate operation result may also be stored in the shared memory. For example, a target device having 16 processors may constitute 4 computing clusters, and the shared memory within each computing cluster may share data.
In practical application, different intermediate operation results correspond to different dimension numbers of the split data, and the operation amounts required for processing the different split data are not the same. Therefore, the target device needs to determine the number of threads adopted by a single processor in the target device to execute parallel operation according to the dimension number of the split data.
In embodiments of the present disclosure, the target device may determine the number of dimensions of the input data of the target model, as well as the maximum number of dimensions of the data that can be processed by a single processor in the target device.
And secondly, the target equipment can split the input data according to the maximum dimension number and the dimension number of the input data, and determine the dimension number of the split data.
Then, the target device can determine the number of the split data which can be processed in parallel by a single processor in the target device according to the number of the dimensions and the maximum number of the dimensions of the split data.
Then, the target device can determine the number of threads adopted by the single processor in the target device to execute the parallel operation according to the determined number of the split data which can be processed by the single processor in the target device in parallel.
Finally, the target device may generate each candidate execution scheme according to the thread number.
For example, the input data is one thousand-dimensional data, the output data is one-dimensional data, and if the dimension number of the intermediate operation result is determined to be one hundred-dimensional, the maximum dimension number which can be processed by the processor in the target device at a time is determined to be one hundred, then the number of the split data which can be processed by the single processor in the target device in parallel is determined to be ten. Thus, the number of threads employed by a single processor in the target device to perform the parallel operation is determined to be ten. Of course, the number of dimensions of different split data determines that the number of threads used by a single processor to perform parallel operation is different.
Furthermore, intermediate operation results obtained by threads adopted by the processor to execute the parallel operation are required to be stored in the register, but the memory capacity of the register is limited, if more intermediate operation results are stored, the stored data exceeds the content capacity of the register, so that the operation time of the processor to execute the operation required by the target model is increased. Based on the above, the target device cannot directly determine the number of threads adopted by the processor to execute the parallel operation according to the device parameter, and the number of threads adopted by the processor to execute the parallel operation in each candidate execution scheme adopted when the target device executes the operation required by the target model needs to be determined, so that in the subsequent execution, the target execution scheme is selected according to the operation time corresponding to each candidate execution scheme.
In practical application, in the process of executing the target model by the target device, each data has an operation sequence relationship, for example, data a needs to be operated first, and then data B needs to be operated based on the operation result of data a. However, if there are two data that do not have an operational sequence, the target device adjusts the operational sequence of the two data, which may reduce the operational time when the target device performs the operation required by the target model.
In the embodiment of the present disclosure, the target device may determine, as the related data, each data having an operational precedence relationship in the operational process.
And secondly, the target device can adjust the operation sequence of other data except the related data, and determine each operation sequence of the processor for processing each data when the target device executes the operation required by the target model.
Finally, the target device may generate each candidate execution scheme according to each operation order.
In practical application, in the process of executing the target model by the target device, each data has an operation precedence relationship. However, the partial object model may separate the two parts to perform calculation, which may result in repeated access, and if each data with an operation precedence relationship is put into the same processor to perform operation, the number of times of memory access can be reduced, and the operation duration can be reduced.
In the embodiment of the present disclosure, the target device may determine, as the related data, each data having an operational precedence relationship in the operational process.
Second, the target device may put the relevant data into the same processor for operation.
Specifically, the target device may put the related data into the same processor, determine an operation result of the first related data, and store the operation result of the first related data into a register of the processor, so as to be used for operation of the next related data, and determine the operation result of the next related data until an output result is calculated.
For example, the present specification describes the above method in detail using layer normalization as an example.
The forward propagation formula for layer normalization is shown below.
In the above formula, ex may be used to represent an average. Var [ x ] can be used to represent the variance, which refers to the average of the squared values of the differences between each sample value and the average of the population of sample values. That is, there is an operation sequence relationship between Var [ x ] and Ex, and it is necessary to operate the code corresponding to Ex first and then operate the code corresponding to Var [ x ]. The target device can put the codes corresponding to Var x and Ex into the same processor for operation, thereby reducing the times of memory access and reducing the operation time.
The layer normalized back propagation formula is shown below.
In the above formula, x i May be used to represent the data of the ith input. Mu may be used to represent the average. Delta 2 May be used to represent the variance, which refers to the average of the squared values of the differences between each sample value and the average of the population of sample values. x may be used to represent the gradient of the input. That is, mu, delta 2 And x has an operation sequence relation, and the code corresponding to mu is required to be operated firstly, and delta is required to be operated 2 And (3) corresponding codes, and finally, calculating codes corresponding to the x. The target device can compare the mu corresponding code and delta 2 The corresponding codes and the codes corresponding to x are put into the same processor for operation, so that the number of memory access times is reduced, and the operation time is shortened.
In practical application, in the process of executing the target model by the target device, each data has an operation precedence relationship. However, if there are two data that do not have an operation precedence relationship, the target device performs parallel operation on the two data, so that the operation duration of the target device when executing the operation required by the target model is reduced.
In the embodiment of the present disclosure, the target device may determine, as the related data, each data having an operational precedence relationship in the operational process.
And secondly, the target equipment can perform parallel operation on each data which does not have an operation precedence relationship in the operation process.
It should be noted that the target device may select one or more of the above candidate execution schemes to obtain the target execution scheme. For example, the target execution scheme may include: at least one of the number of processors employed when the target device performs the arithmetic operation required by the target model, the number of parallel arithmetic data of a single processor when the target device performs the arithmetic operation required by the target model, the arithmetic order in which the processors process the respective data when the target device performs the arithmetic operation required by the target model, and the like.
Also, different candidate implementations may not be exactly the same. For example, the number of processors used when the target device in the candidate execution scheme a performs the operation required by the target model is the same as that in the candidate execution scheme B, the number of parallel operation data of a single processor when the target device performs the operation required by the target model is the same, and the operation order in which the processors process the respective data when the target device performs the operation required by the target model is different.
S104: and for each candidate execution scheme, executing operation required by running the target model through the candidate execution scheme to determine an execution evaluation parameter monitored when the target equipment obtains an output result of the target model through executing the operation, wherein the execution evaluation parameter is used as an execution evaluation parameter corresponding to the candidate execution scheme.
In the embodiment of the present disclosure, the target device may execute, for each candidate execution scheme, an operation required when the target model is executed by the candidate execution scheme, so as to determine, as an execution evaluation parameter corresponding to the candidate execution scheme, an execution evaluation parameter monitored when the target device obtains an output result of the target model by executing the operation. The execution evaluation parameter mentioned here may refer to parameters such as an operation time period, an operation resource consumption amount, and the like.
The output result mentioned here may refer to an output result corresponding to the test data obtained by the target device inputting the test data into the target model through each candidate execution scheme in the process of determining the execution evaluation parameter corresponding to each candidate execution scheme.
S106: and determining a target execution scheme according to execution evaluation parameters corresponding to each candidate execution scheme, adjusting operation required by the target equipment to execute the target model according to the target execution scheme, and executing data processing according to the adjusted operation through the target model.
In the embodiment of the present disclosure, the target device may determine the target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme, and adjust, according to the target execution scheme, the operation required by the target device to execute the target model, so as to execute data processing according to the adjusted operation through the target model.
Specifically, the target device may determine the target execution scheme according to the execution evaluation parameters such as the operation duration, the operation resource consumption amount, and the like of each candidate execution scheme.
Of course, the target device may determine the composite score according to the execution evaluation parameters such as the operation duration, the operation resource consumption amount, and the like of the candidate execution scheme. And determining the target execution scheme according to the comprehensive score of each candidate execution scheme. The composite score referred to herein may refer to a score determined based on the performance evaluation parameters. For example, the shorter the operation time period, the higher the score. For another example, the smaller the amount of computation resource consumption, the higher the score. And determining a formula corresponding to the comprehensive score according to the actual requirement. The present specification does not limit the formula corresponding to the composite score.
In practical applications, in order to ensure that the user can get the execution scheme in a short time. The target device can preset a basic execution scheme, so that the problem that the target execution scheme capable of executing the target model on the target device cannot be determined in a short time is avoided.
In the embodiment of the present disclosure, the target device may acquire a preset basic scheme. The basic scheme mentioned here may be obtained through expert experience.
Then, the target device can determine the execution evaluation parameters monitored when the target device obtains the output result of the target model by executing the basic scheme, and the execution evaluation parameters are used as the execution evaluation parameters corresponding to the basic scheme.
Finally, the target device may determine the target execution scheme according to the execution evaluation parameters corresponding to the basic scheme and the execution evaluation parameters corresponding to each candidate execution scheme.
Specifically, in the set time, the target device may determine the execution evaluation parameters corresponding to each candidate execution scheme and the execution evaluation parameters corresponding to the basic scheme through the operation required by each candidate execution scheme and the basic scheme to execute the target model. The target device can select the execution scheme with the shortest operation time or the highest comprehensive score from the execution schemes as the target execution scheme.
In practical application, because each implementation scheme determines a single device parameter based on an empirical rule, the influence among different device parameters is not considered, and whether the determined device parameters are optimal cannot be judged. Thus, the determined device parameter may not be the optimal parameter value. Based on the above, the target device may determine an initial value of the device parameter based on the operation process and the device parameter, and then continuously adjust the magnitude of the numerical value of the device parameter with the initial value as a reference until the determined integrated score of the execution evaluation parameter is highest, so as to obtain the target execution scheme.
In the embodiment of the present disclosure, the target device may determine the candidate execution scheme with respect to the execution evaluation parameter monitored when the target device obtains the output result of the target model by performing the arithmetic operation as the adjustment target.
It should be noted that, in this specification, a method for determining a candidate implementation may refer to a random search algorithm.
Of course, the target device may adjust the value of each device parameter individually until the determined performance evaluation parameter has the highest overall score, so as to obtain the target execution scheme.
In the embodiment of the present disclosure, after the target device adjusts the operation required by the target device to execute the target model according to the target execution scheme, the target device may receive the service request of the user, and input the service data corresponding to the service request into the target model according to the service request, so as to execute the operation required by the target model to process the service data according to the target execution scheme, thereby obtaining the service result.
In the embodiment of the present disclosure, a specific flow of determining, by the target device, the target execution scheme is shown in fig. 2.
Fig. 2 is a schematic diagram of a determining target execution scheme according to an embodiment of the present disclosure.
In fig. 2, the target device may generate, according to the device parameters, candidate execution schemes adopted when the target device performs the arithmetic operation required by the target model. The target device may then determine the execution evaluation parameters corresponding to each candidate execution scheme. And determining a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme.
According to the method, candidate execution schemes adopted when the target equipment executes operation required by the target model can be generated according to the equipment parameters. And secondly, determining the execution evaluation parameters corresponding to the candidate execution schemes through the candidate execution schemes. And finally, determining a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme. Therefore, the target device is ensured to be capable of effectively executing data processing through the target model.
The above method for processing data provided for the embodiments of the present specification further provides a corresponding apparatus, a storage medium, and an electronic device based on the same concept.
Fig. 3 is a schematic structural diagram of an apparatus for data processing according to an embodiment of the present disclosure, where the apparatus includes:
an obtaining module 300, configured to obtain a target model and device parameters of a target device, where the target device is configured to execute the target model;
a generating module 302, configured to generate, according to the device parameter, each candidate execution scheme adopted when the target device executes the target model;
a determining module 304, configured to, for each candidate execution scheme, perform an operation required for running the target model according to the candidate execution scheme, so as to determine, as an execution evaluation parameter corresponding to the candidate execution scheme, an execution evaluation parameter monitored when the target device obtains an output result of the target model by performing the operation;
and the adjustment module 306 is configured to determine a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme, and adjust, according to the target execution scheme, an operation required by the target device to execute the target model, so as to perform data processing according to the adjusted operation through the target model.
Optionally, the generating module 302 is specifically configured to determine a dimension number of input data of the target model, determine a maximum dimension number of data that can be processed by a single processor in the target device, determine, according to the dimension number of input data of the target model and the maximum dimension number, a number of processors required by the target device to perform an operation required by the target model, and generate, according to the number of processors required by the target device to perform the operation required by the target model, each candidate execution scheme adopted by the target device to perform the operation required by the target model.
Optionally, the generating module 302 is specifically configured to determine, according to input data and output data of the target model, a dimension ratio corresponding to the target model, where the dimension ratio is used to represent a ratio between a dimension number of the input data and a dimension number of the output data of the target model, and if it is determined that the dimension ratio exceeds a dimension ratio threshold corresponding to the target device, split the input data according to the dimension ratio and the maximum dimension number, determine a number of split data, and determine, according to the number of split data, a number of processors required by the target device when executing an operation required by the target model.
Optionally, the generating module 302 is specifically configured to determine a dimension number of input data of the target model, determine a maximum dimension number of data that can be processed by a single processor in the target device, split the input data according to the maximum dimension number and the dimension number of the input data, determine a dimension number of the split data, determine a number of data that can be processed by the single processor in the target device in parallel according to the dimension number of the split data and the maximum dimension number, determine a number of threads that can be adopted by the single processor in the target device to execute parallel operation according to the determined number of data that can be processed by the single processor in the target device in parallel, and generate each candidate execution scheme according to the number of threads.
Optionally, the generating module 302 is specifically configured to determine each data having an operational sequence relationship in an operational process, adjust an operational sequence of other data except the related data as related data, determine each operational sequence of each data processed by a processor when the target device executes an operation required by the target model, and generate each candidate execution scheme according to each operational sequence.
Optionally, the adjustment module 306 is specifically configured to obtain a preset basic scheme, determine an execution evaluation parameter monitored when the target device obtains an output result of the target model by executing the basic scheme, and determine a target execution scheme according to the execution evaluation parameter corresponding to the basic scheme and the execution evaluation parameter corresponding to each candidate execution scheme as the execution evaluation parameter corresponding to the basic scheme.
Optionally, the adjustment module 306 is further specifically configured to receive a service request of a user, input service data corresponding to the service request into the target model according to the service request, and execute, according to the target execution scheme, an operation required by the target model to process the service data, so as to obtain a service result.
The present specification also provides a computer readable storage medium storing a computer program which when executed by a processor is operable to perform the method of data processing provided in figure 1 above.
The embodiment of the specification also provides a schematic structural diagram of the electronic device shown in fig. 4. At the hardware level, as in fig. 4, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile storage, although it may include hardware required for other services. The processor reads the corresponding computer program from the nonvolatile memory into the memory and then runs to implement the method of data processing provided in fig. 1.
Of course, other implementations, such as logic devices or combinations of hardware and software, are not excluded from the present description, that is, the execution subject of the following processing flows is not limited to each logic unit, but may be hardware or logic devices.
It should be noted that, all actions for acquiring signals, information or data in the present application are performed under the condition of conforming to the corresponding data protection rule policy of the country of the location and obtaining the authorization given by the owner of the corresponding device.
In the 90 s of the 20 th century, improvements to one technology could clearly be distinguished as improvements in hardware (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software (improvements to the process flow). However, with the development of technology, many improvements of the current method flows can be regarded as direct improvements of hardware circuit structures. Designers almost always obtain corresponding hardware circuit structures by programming improved method flows into hardware circuits. Therefore, an improvement of a method flow cannot be said to be realized by a hardware entity module. For example, a programmable logic device (Programmable Logic Device, PLD) (e.g., field programmable gate array (Field Programmable Gate Array, FPGA)) is an integrated circuit whose logic function is determined by the programming of the device by a user. A designer programs to "integrate" a digital system onto a PLD without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Moreover, nowadays, instead of manually manufacturing integrated circuit chips, such programming is mostly implemented by using "logic compiler" software, which is similar to the software compiler used in program development and writing, and the original code before the compiling is also written in a specific programming language, which is called hardware description language (Hardware Description Language, HDL), but not just one of the hdds, but a plurality of kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), lava, lola, myHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog are currently most commonly used. It will also be apparent to those skilled in the art that a hardware circuit implementing the logic method flow can be readily obtained by merely slightly programming the method flow into an integrated circuit using several of the hardware description languages described above.
The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer readable medium storing computer readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable logic controllers, and embedded microcontrollers, examples of which include, but are not limited to, the following microcontrollers: ARC625D, atmel AT91SAM, microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic of the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller in a pure computer readable program code, it is well possible to implement the same functionality by logically programming the method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Such a controller may thus be regarded as a kind of hardware component, and means for performing various functions included therein may also be regarded as structures within the hardware component. Or even means for achieving the various functions may be regarded as either software modules implementing the methods or structures within hardware components.
The system, apparatus, module or unit set forth in the above embodiments may be implemented in particular by a computer chip or entity, or by a product having a certain function. One typical implementation is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
For convenience of description, the above devices are described as being functionally divided into various units, respectively. Of course, the functions of each element may be implemented in one or more software and/or hardware elements when implemented in the present specification.
It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The present description is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the specification. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.
Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises the element.
It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.
The foregoing is merely exemplary of the present disclosure and is not intended to limit the disclosure. Various modifications and alterations to this specification will become apparent to those skilled in the art. Any modifications, equivalent substitutions, improvements, or the like, which are within the spirit and principles of the present description, are intended to be included within the scope of the claims of the present description.
Claims (10)
1. A method of data processing, comprising:
acquiring a target model and equipment parameters of target equipment, wherein the target equipment is used for executing operation required by the target model;
generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the equipment parameters;
for each candidate execution scheme, executing operation required by running the target model through the candidate execution scheme to determine an execution evaluation parameter monitored when the target equipment obtains an output result of the target model through executing the operation, wherein the execution evaluation parameter is used as an execution evaluation parameter corresponding to the candidate execution scheme;
And determining a target execution scheme according to execution evaluation parameters corresponding to each candidate execution scheme, adjusting operation required by the target equipment to execute the target model according to the target execution scheme, and executing data processing according to the adjusted operation through the target model.
2. The method of claim 1, generating, according to the device parameters, candidate execution schemes adopted when the target device executes the arithmetic operation required by the target model, specifically including:
determining the dimension number of input data of the target model, and determining the maximum dimension number of data which can be processed by a single processor in the target device;
determining the number of processors required by the target equipment when the target equipment executes the operation required by the target model according to the dimension number of the input data of the target model and the maximum dimension number;
and generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the number of processors required by the target equipment when executing the operation required by the target model.
3. The method according to claim 2, according to the dimension number of the input data of the target model and the maximum dimension number, determining the number of processors required by the target device to perform the arithmetic operation required by the target model, specifically comprising:
Determining a dimension ratio corresponding to the target model according to the input data and the output data of the target model, wherein the dimension ratio is used for representing the ratio between the dimension number of the input data and the dimension number of the output data of the target model;
if the dimension ratio exceeds the dimension ratio threshold corresponding to the target device, splitting the input data according to the dimension ratio and the maximum dimension number, and determining the quantity of the split data;
and determining the number of processors required by the target equipment when the target equipment executes the operation required by the target model according to the number of the split data.
4. The method of claim 1, generating, according to the device parameters, candidate execution schemes adopted when the target device executes the arithmetic operation required by the target model, specifically including:
determining the dimension number of input data of the target model, and determining the maximum dimension number of data which can be processed by a single processor in the target device;
splitting the input data according to the maximum dimension number and the dimension number of the input data, and determining the dimension number of the split data;
Determining the number of the split data which can be processed in parallel by a single processor in the target equipment according to the number of the dimensions of the split data and the maximum number of the dimensions;
determining the number of threads adopted by a single processor in the target equipment for executing parallel operation according to the determined number of the split data which can be processed in parallel by the single processor in the target equipment;
and generating each candidate execution scheme according to the thread number.
5. The method of claim 1, generating, according to the device parameters, candidate execution schemes adopted when the target device executes the arithmetic operation required by the target model, specifically including:
determining each data with an operation sequence in the operation process as related data;
adjusting the operation sequence of other data except related data, and determining each operation sequence of each data processed by a processor when the target equipment executes the operation required by the target model;
and generating each candidate execution scheme according to each operation sequence.
6. The method of claim 1, wherein determining the target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme specifically comprises:
Acquiring a preset basic scheme;
determining an execution evaluation parameter monitored when the target equipment obtains an output result of the target model by executing the basic scheme, wherein the execution evaluation parameter is used as an execution evaluation parameter corresponding to the basic scheme;
and determining a target execution scheme according to the execution evaluation parameters corresponding to the basic scheme and the execution evaluation parameters corresponding to each candidate execution scheme.
7. The method of claim 1, the method further comprising:
receiving a service request of a user;
and according to the service request, inputting service data corresponding to the service request into the target model, and executing operation required by the target model for processing the service data through the target execution scheme to obtain a service result.
8. An apparatus for data processing, comprising:
the device comprises an acquisition module, a calculation module and a calculation module, wherein the acquisition module is used for acquiring a target model and device parameters of target device, and the target device is used for executing calculation operation required by the target model;
the generation module is used for generating each candidate execution scheme adopted when the target equipment executes the operation required by the target model according to the equipment parameters;
The determining module is used for determining, for each candidate execution scheme, an operation required when the target model is operated through the candidate execution scheme so as to determine an execution evaluation parameter monitored when the target device obtains an output result of the target model through executing the operation, wherein the execution evaluation parameter is used as an execution evaluation parameter corresponding to the candidate execution scheme;
and the adjusting module is used for determining a target execution scheme according to the execution evaluation parameters corresponding to each candidate execution scheme, adjusting the operation required by the target equipment for executing the target model according to the target execution scheme, and executing data processing according to the adjusted operation and the target model.
9. A computer readable storage medium storing a computer program which, when executed by a processor, implements the method of any of the preceding claims 1-7.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any of the preceding claims 1-7 when the program is executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310164810.XA CN116384505A (en) | 2023-02-13 | 2023-02-13 | Data processing method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310164810.XA CN116384505A (en) | 2023-02-13 | 2023-02-13 | Data processing method and device, storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116384505A true CN116384505A (en) | 2023-07-04 |
Family
ID=86962362
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310164810.XA Pending CN116384505A (en) | 2023-02-13 | 2023-02-13 | Data processing method and device, storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116384505A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116842060A (en) * | 2023-08-30 | 2023-10-03 | 之江实验室 | Reasoning query optimization method and device based on agent model rearrangement technology |
-
2023
- 2023-02-13 CN CN202310164810.XA patent/CN116384505A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116842060A (en) * | 2023-08-30 | 2023-10-03 | 之江实验室 | Reasoning query optimization method and device based on agent model rearrangement technology |
| CN116842060B (en) * | 2023-08-30 | 2024-01-09 | 之江实验室 | Reasoning query optimization method and device based on agent model rearrangement technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116225669B (en) | Task execution method and device, storage medium and electronic equipment | |
| CN115981870B (en) | Data processing method and device, storage medium and electronic equipment | |
| CN116521380A (en) | Resource self-adaptive collaborative model training acceleration method, device and equipment | |
| CN116432778B (en) | Data processing method and device, storage medium and electronic equipment | |
| JP2024529206A (en) | Distributed Training for Container Scheduling for Intelligent Computing | |
| CN117312394B (en) | Data access method and device, storage medium and electronic equipment | |
| CN109656946B (en) | Multi-table association query method, device and equipment | |
| CN116384505A (en) | Data processing method and device, storage medium and electronic equipment | |
| CN116932175B (en) | Heterogeneous chip task scheduling method and device based on sequence generation | |
| CN116402165B (en) | Operator detection method and device, storage medium and electronic equipment | |
| CN116107636B (en) | Hardware acceleration method and device, storage medium and electronic equipment | |
| CN117424827A (en) | Communication method and device based on distributed deep learning cache system | |
| CN116204324A (en) | Task execution method and device, storage medium and electronic equipment | |
| CN116108498A (en) | Program execution method, program execution device, storage medium and electronic equipment | |
| CN110008112B (en) | Model training method and device, service testing method and device | |
| CN114676132A (en) | Data table association method and device, storage medium and electronic equipment | |
| CN109614388B (en) | Budget deduction method and device | |
| CN111880922A (en) | Processing method, device and equipment for concurrent tasks | |
| CN116109008B (en) | Method and device for executing service, storage medium and electronic equipment | |
| CN117171577B (en) | Dynamic decision method and device for high-performance operator selection | |
| CN118466863B (en) | Data storage method and device, storage medium and electronic equipment | |
| CN117522669B (en) | Method, device, medium and equipment for optimizing internal memory of graphic processor | |
| CN116909744B (en) | Thread pool parameter adjusting method and device, storage medium and electronic equipment | |
| CN116954954B (en) | Method and device for processing multi-task queues, storage medium and electronic equipment | |
| CN117455015B (en) | Model optimization method and device, storage medium and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |