CN112631849B - Power consumption detection model construction method, power consumption detection device and electronic equipment - Google Patents

Abstract

The application provides a power consumption detection model construction method, a power consumption detection device and electronic equipment, wherein the method comprises the following steps: simulating the target functional module in a first appointed number of application scenes to obtain first turnover data of data signals in each application scene of the first appointed number; screening the data signals according to the first overturn data to determine a target observation signal group; determining second overturn data of the target observation signal group in each application scene from the first overturn data; calculating to obtain correction parameters and signal weights of all target observation signals of the target observation signal group according to the second overturn data; and determining a power consumption detection model according to the correction parameter and the signal weight by taking the turning quantity of each target observation signal in the target observation signal group as an independent variable. The detection of the power consumption of the chip can be more objective, and the efficiency of the detection of the power consumption of the chip can be improved.

Description

Power consumption detection model building method, power consumption detection method, device and electronic equipment

technical field

The present application relates to the technical field of chip detection, in particular, to a power consumption detection model building method, power consumption detection method, device and electronic equipment.

Background technique

The power consumption of the chip is one of the important factors limiting the design of the chip. The power consumption of the chip can be divided into two parts: dynamic power consumption and static power consumption, where the dynamic power consumption is related to the inversion of clock and data. Dynamic power consumption is related to clock and data toggling, the function of the chip, and the operating state of the chip.

Existing technical solutions for estimating data inversion generally select observation signals based on the work experience of relevant technical personnel. Then, based on the inversion of the observed signal, the power consumption of the chip based on the inversion is determined. The accuracy of the power consumption determined by this power consumption determination method depends on the experience of relevant technical personnel.

Contents of the invention

The purpose of the present application is to provide a power consumption detection model building method, power consumption detection method, device and electronic equipment, which can solve the problem of insufficient objectivity of power consumption detection.

In the first aspect, the embodiment of the present application provides a method for constructing a power consumption detection model, including:

Simulating the target functional module in a first specified number of application scenarios to obtain the first flip data of the data signal in each application scenario in the first specified number of application scenarios, the target functional module is a chip A functional module in the design;

Screening data signals according to the first flip data to determine a target observation signal group, where the target observation signals include a plurality of target observation signals;

From the first inversion data, determine the second inversion data of each target observation signal in the target observation signal group in each application scenario in the first specified number of application scenarios;

Calculate and obtain correction parameters and signal weights corresponding to each target observation signal of the target observation signal group according to the second flip data;

Taking the inversion amount of each target observation signal in the target observation signal group as an independent variable, and determining a power consumption detection model according to the correction parameter and the signal weight, the power consumption detection model is used to control the power consumption of the chip Calculation.

In an optional implementation manner, the screening of data signals according to the first flip data to determine a target observation signal group includes:

performing primary screening on the data signals used in the first specified number of application scenarios according to a preset value interval to obtain an initial observation signal group, the initial observation signal group including a plurality of initial observation signals;

determining the signal error of each initial observation signal according to the initial observation signal group;

The target observation signal group is determined from the initial observation signal group according to the signal error of each initial observation signal.

In the above embodiment, the selected initial observation signal is filtered based on the signal error of the selected initial observation signal, so that the observation signal in the selected target observation signal group can better characterize the chip Therefore, the power consumption detection of the chip based on the target observation signal group can also be more accurate.

In an optional implementation manner, the first specified number of application scenarios includes N application scenarios, and the N application scenarios include M sub-scenes; The data signals used in the specified number of application scenarios are initially screened to obtain the initial observation signal group, including:

In each of the sub-scenes, the observation signals whose flipping amount is in the preset value range are screened out to obtain the observation signal group corresponding to each sub-scene;

For each application scenario, taking the intersection of the observation signal groups corresponding to each of the sub-scenes in the application scenario to obtain the observation signal groups corresponding to the application scenarios, so as to obtain the N observation signal groups corresponding to the N application scenarios, The N observation signal groups are the initial observation signal groups, where M and N are positive integers greater than one.

In the above implementation manner, the observation signal groups in the preset value range are screened out, so that the observation signals in the selected observation signal group can better represent the inversion of most observation signals.

In an optional implementation manner, the preset value range is determined by the following steps:

calculating the flipping amount per unit time of each data signal in each of the sub-scenes;

Determining the non-zero intermediate number of inversion amount per unit time of each data signal in each sub-scene;

The preset value interval is determined according to the non-zero intermediate number of the flip amount.

In the above embodiment, the preset value interval determined based on the non-zero inversion value middle number can better represent the possible interval of each observation signal, so that the observation determined based on the preset value interval can be The signal is better able to represent the flipping of most observed signals.

In an optional implementation manner, the determining the target observation signal group from the initial observation signal group according to the signal error of each initial observation signal includes:

In the target application scenario among the N application scenarios, select a second specified number of observation signals with the smallest signal error to obtain a screening observation signal group, and the target application scenario is any of the N application scenarios. an application scenario;

The target observation signal group is obtained by unioning the N screening observation signal groups corresponding to the N application scenarios.

In an optional implementation manner, the signal error of each initial observation signal is determined by the following formula:

表示第i个应用场景中的单位时间内的信号总翻转量均值，tf_i,j,k表示第i个应用场景中的第j个子场景中的第k个初始观测信号的单位时间内的翻转量，

表示第i个应用场景中的第k个初始观测信号的翻转量均值。Among them, M represents the number of sub-scenes contained in the i-th application scene, W _i,k represents the signal error of the k-th initial observation signal in the i-th application scene, tF _i,j represents the i-th application scene The total signal turnover per unit time in the jth sub-scene in ,

Represents the mean value of the total signal turnover per unit time in the i-th application scenario, tf _i,j,k represents the reversal per unit time of the k-th initial observation signal in the j-th sub-scene in the i-th application scenario quantity,

Indicates the mean value of the flipping amount of the kth initial observation signal in the i-th application scenario.

In an optional implementation manner, the calculation according to the second flip data to obtain the correction parameter and the signal weight corresponding to each target observation signal of the target observation signal group includes:

Construct a first matrix according to D sets of first-type inversion data in the second inversion data, the D represents the number of sub-scenes in the specified number of application scenarios, and the first-type inversion data represents a target observation signal The turnover amount of each target observation signal of the group under each sub-scene in the specified number of application scenarios;

Constructing a second matrix according to D sets of second-type inversion data in the second inversion data, the second-type inversion data indicating that the target observation signal group is in each application scenario in the specified number of application scenarios The total amount of turnover;

Signal weights and correction parameters corresponding to the target observation signal group are calculated according to the first matrix and the second matrix.

In an optional embodiment, it also includes:

calculating a model error of the power consumption detection model according to the second flip data of each target observation signal;

judging whether the model error is greater than a preset error value;

If the model error is greater than a preset error value, reselect the latest target observation signal group to update the constructed power consumption detection model.

In an optional implementation manner, the model error of the power consumption detection model is calculated by the following formula:

Among them, e represents the model error of the power consumption detection model, D represents the number of application scenarios involved in the simulation, stF _i represents the total turnover of the target observation signal group in the ith scene, etF _i represents The flip amount in the i-th scene calculated by the model.

In the foregoing implementation manners, calculation and verification are performed through model errors of the power consumption detection model, so that the accuracy of the power consumption detection model can be improved.

In the second aspect, the embodiment of the present application also provides a power consumption detection method, including:

Obtain the inversion value of each observation signal of the chip to be detected in the target observation signal group in unit time;

Based on the power consumption detection model obtained by the method provided in the first aspect or any optional implementation manner of the first aspect, the power consumption of the chip to be detected is calculated according to the reversal value of each observation signal per unit time .

In the third aspect, the embodiment of the present application also provides a device for constructing a power consumption detection model, including:

A simulation module, configured to simulate the target function module in a first specified number of application scenarios, so as to obtain the first flip data of the data signal in each application scenario in the first specified number of application scenarios, the target The functional module is a functional module in a chip design;

A screening module, configured to screen data signals according to the first flip data to determine a target observation signal group;

A first determination module, configured to determine, from the first inversion data, the second inversion of each target observation signal in the target observation signal group under each application scenario in the first specified number of application scenarios data;

A first calculation module, configured to calculate correction parameters and signal weights corresponding to each target observation signal of the target observation signal group according to the second flip data;

The second determination module is configured to use the inversion amount of each target observation signal in the target observation signal group as an argument, and determine a power consumption detection model according to the correction parameter and the signal weight, and the power consumption detection model It is used to calculate the power consumption of the chip.

In a fourth aspect, the embodiment of the present application further provides a power consumption detection device, including:

The obtaining module is used to obtain the inversion value of each observation signal in the target observation signal group of the chip to be detected in unit time;

The second calculation module is used to calculate the power consumption detection model obtained based on the method provided in the first aspect or any optional implementation manner of the first aspect, and calculate the calculated value according to the reversal value of each observed signal within a unit time. Describe the power consumption of the chip to be tested.

In the fifth aspect, the embodiment of the present application further provides an electronic device, including: a processor and a memory, the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the machine-readable The steps of the above method are performed when the instructions are executed by the processor.

In a fifth aspect, the embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is run by a processor, the steps of the above method are executed.

The beneficial effect of the embodiment of the present application is: through the selection based on the observation signal, and based on the selected target observation signal group, a power consumption detection model capable of calculating the power consumption of the chip is constructed, so that the calculation of the power consumption of the chip does not depend on relevant technical personnel Based on our experience, the chip power consumption calculated based on this power consumption detection model can be more objective. Further, the power consumption efficiency of the chip calculated based on the power consumption detection model may also be higher.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the accompanying drawings that are required in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and thus It should be regarded as a limitation on the scope, and those skilled in the art can also obtain other related drawings based on these drawings without creative work.

FIG. 1 is a schematic block diagram of an electronic device provided by an embodiment of the present application.

FIG. 2 is a flow chart of a method for constructing a power consumption detection model provided by an embodiment of the present application.

FIG. 3 is a flow chart of another implementation of the method for constructing a power consumption detection model provided in the embodiment of the present application.

FIG. 4 is a schematic diagram of functional modules of a device for constructing a power consumption detection model provided in an embodiment of the present application.

FIG. 5 is a flowchart of a method for detecting power consumption provided by an embodiment of the present application.

FIG. 6 is a schematic diagram of functional modules of a power consumption detection device provided by an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.

It should be noted that like numerals and letters denote similar items in the following figures, therefore, once an item is defined in one figure, it does not require further definition and explanation in subsequent figures. Meanwhile, in the description of the present application, the terms "first", "second" and the like are only used to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

The power consumption of the chip can be divided into dynamic power consumption and static power consumption. The dynamic power consumption is related to the inversion of the clock and data signals, while the static power consumption is related to the leakage of the device. Static power consumption is generally relatively small and can be evaluated based on die area and temperature. However, dynamic power consumption is related to the flipping of clock and data signals, the function of the chip and the operating state of the chip.

In order to realize the detection of the dynamic power consumption of the chip, the implementation of the present application provides a method for constructing a power consumption detection model, which can construct a model for detecting the power consumption of the chip. In the method provided in the embodiment of the present application, through power consumption simulation, it is possible to make statistics on the inversion of each data signal in the chip under various expected working scenarios. The following describes several embodiments.

Embodiment one

To facilitate the understanding of this embodiment, an electronic device that executes the power consumption detection model construction method and the power consumption detection method disclosed in the embodiments of the present application is first introduced in detail.

As shown in FIG. 1 , it is a schematic block diagram of an electronic device. The electronic device 100 may include a memory 111 and a processor 113 . Those skilled in the art can understand that the structure shown in FIG. 1 is only for illustration, and does not limit the structure of the electronic device 100 . For example, the electronic device 100 may also include more or fewer components than shown in FIG. 1 , or have a different configuration than that shown in FIG. 1 .

The components of the above-mentioned memory 111 and processor 113 are directly or indirectly electrically connected to each other to realize data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines. The aforementioned processor 113 is used to execute the executable modules stored in the memory.

Wherein, the memory 111 can be, but not limited to, random access memory (Random Access Memory, referred to as RAM), read-only memory (Read Only Memory, referred to as ROM), programmable read-only memory (ProgrammableRead-Only Memory, referred to as PROM) , Erasable Programmable Read-Only Memory (EPROM for short), Electric Erasable Programmable Read-Only Memory (EEPROM for short), etc. Wherein, the memory 111 is used to store a program, and the processor 113 executes the program after receiving an execution instruction, and the method performed by the electronic device 100 according to the process definition disclosed in any embodiment of the present application can be applied to processing In the device 113, or implemented by the processor 113.

The above-mentioned processor 113 may be an integrated circuit chip, which has a signal processing capability. Above-mentioned processor 113 can be general-purpose processor, comprises central processing unit (Central Processing Unit, be called for short CPU), network processor (Network Processor, be called for short NP) etc.; Can also be digital signal processor (digital signal processor, be called for short DSP) , Application Specific Integrated Circuit (ASIC for short), Field Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. Various methods, steps, and logic block diagrams disclosed in the embodiments of the present application may be implemented or executed. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, or the like.

The electronic device 100 in this embodiment may be used to execute each step in each method provided in the embodiment of this application. The implementation process of the power consumption detection model construction method and the power consumption detection method will be described in detail below through several embodiments.

Embodiment two

Please refer to FIG. 2 , which is a flowchart of a method for constructing a power consumption detection model provided by an embodiment of the present application. The specific process shown in FIG. 2 will be described in detail below.

Step 201: Simulate the target function module in a first specified number of application scenarios to obtain first inversion data of data signals in each application scenario in the first specified number of application scenarios.

The above-mentioned target function module is a function module in a chip design. The above-mentioned target function module can also be a subsystem or a chip.

Optionally, according to different functions of the target functional module, the above specified number of application scenarios may also be different. For example, the target function module may be a security coprocessor module, and the specified number of application scenarios may include: selecting a random number generation scenario, a symmetric encryption algorithm encryption operation scenario, a symmetric encryption algorithm decryption operation scenario, an asymmetric encryption algorithm encryption operation scenario, Asymmetric encryption algorithm decryption operation scenarios, hash operation scenarios, data compression operation scenarios and data decompression operation scenarios.

Exemplarily, the application scenarios corresponding to the target function modules can be respectively S ₁ , S ₂ , ..., S _N , and in the application scenario S _i , the application scenario S _i can be divided into M sub-scenes S _i,j (j=1,2,...,M), where the workload of the target function module in the sub-scene S _i,j is arranged in ascending order with j.

In this embodiment, the power consumption simulation is performed on the N*M sub-scenes S _i,j (i=1,2,...,N; j=1,2,...,M) of the target function module respectively, and the simulation is obtained by statistics The valid duration is T _i,j (i=1,2,...,N; j=1,2,...,M).

Exemplarily, the above-mentioned first inversion data may include: the total signal inversion amount F _i,j (i=1,2,...,N ; j=1,2,...,M), and the inversion amount f _i,j,k (i=1,2 ,...,N; j=1,2,...,M; k=1,2,...,K), where K represents the number of data signals in the target functional module. Wherein, N, M, and K are all positive integers greater than zero. Taking the aforementioned target function module as an example, the value of N may be 8.

In this embodiment, according to the different functions of the target functional modules, the number of observed signals in the corresponding target functional modules may also be different.

In step 202, the data signals are screened according to the first flip data, so as to determine a target observation signal group.

The target observation signal group includes a plurality of target observation signals.

As the design of integrated circuits becomes more and more complex and has more and more functions, the number of logic gates in a single functional module can reach millions or even tens of millions of gates. Therefore, it is necessary to select a suitable observation signal from a large number of data signals for evaluation The total number of data rollovers for the block. Optionally, the desired target observation signal group can be gradually screened out through two rounds of screening.

Exemplarily, step 202 may include the following steps.

Step 2021 , perform primary screening on the data signals used in the first specified number of application scenarios according to the preset value range, so as to obtain an initial observation signal group.

The initial observation signal group includes a plurality of initial observation signals.

Exemplarily, the number of the specified number of application scenarios takes N application scenarios as an example. The above step 2021 may include step a and step b.

In step a, in each of the sub-scenes, the observation signals whose inversion amount is in the preset value range are screened out to obtain an observation signal group corresponding to each sub-scene.

In one embodiment, the preset value range is determined by the following steps: calculating the inversion amount per unit time of each data signal in each sub-scene; determining the inversion per unit time of each data signal in each sub-scene The non-zero intermediate number of inversion amount in the amount; the preset value interval is determined according to the non-zero inversion amount intermediate number.

Exemplarily, the turnover per unit time of each data signal in the sub-scene can be expressed as:

tf _i,j,k =f _i,j,k /T _i,j ;

Among them, f _i,j,k represents the inversion amount of the kth data signal in the jth sub-scene in the i-th application scene, T _i,j represents the flipping amount of the j-th sub-scene in the i-th application scene The effective duration of the simulation, tf _i,j,k represent the flipping amount per unit time of the k-th data signal in the j-th sub-scene in the i-th application scenario.

Optionally, the above-mentioned intermediate number of non-zero inversion amounts may be a mode number of non-zero inversion amounts of the K data signals in the j-th sub-scene in the i-th application scenario.

Optionally, the above-mentioned intermediate number of non-zero inversion amounts may be an average value of non-zero inversion amounts of K data signals in the j-th sub-scene in the i-th application scenario.

Optionally, the above-mentioned intermediate number of non-zero inversion amounts may be a median of non-zero inversion amounts of K data signals in the j-th sub-scene in the i-th application scenario. For example, the median may be expressed as: sort the non-zero inversion amounts per unit time of the K data signals from small to large, and the median represents the value sorted in the middle position.

Exemplarily, the determination of the preset value interval according to the non-zero inversion amount intermediate number may be: taking the difference between the non-zero inversion amount intermediate number and the fluctuation value as the small endpoint of the preset value interval, and using the non-zero inversion amount intermediate number and The sum of the fluctuation values is used as the maximum endpoint of the preset value range.

In the above example, the preset value range can be expressed as: [Mtf _i,j -dt,Mtf _i,j +dt];

Wherein, Mtf _i,j represents the middle number of non-zero inversion amounts of K data signals in the j-th sub-scene in the i-th application scenario, and dt represents the fluctuation value.

Optionally, the fluctuation value used when determining the preset value range can be set according to requirements. Exemplarily, the fluctuation value may take values such as 2, 3 and so on.

By repeating step a, the observation signal groups Q _{i ,j of all sub-scenes S i,j} of the application scene S _i can be filtered out.

Step b, for each application scenario, taking the intersection of the observation signal groups corresponding to each of the sub-scenes in the application scenario to obtain the observation signal groups corresponding to the application scenario, so as to obtain the N observation signal groups corresponding to the N application scenarios signal group.

The aforementioned N observation signal groups are used as the initial observation signal group.

Exemplarily, in the i-th application scenario, M observation signal groups Q _i,j are intersected to obtain an observation signal group PQ _i . As far as N application scenarios are concerned, N observation signal groups under N application scenarios can be obtained.

Step 2022: Determine the signal error of each initial observation signal according to the initial observation signal group.

Optionally, for the initial observation signal X _k in the initial observation signal group PQ _i of the application scenario S _i , the unit time flip tf i,j,k of the initial observation signal X _k under the sub-scene S _{i ,j} can be expressed as As the independent variable, the total data flipping number tF i, _{j in the target function module under each sub-scene S i ,j} is used as the dependent variable, and the signal error of the initial observation signal is calculated by using the linear regression model and the least square method.

Exemplarily, the signal error of each initial observation signal is determined by the following formula:

tF _i,j =F _i,j /T _i,j ;

表示第i个应用场景中的单位时间内的信号总翻转量均值，tf_i,j,k表示第i个应用场景中的第j个子场景中的第k个初始观测信号的单位时间内的翻转量，

表示第i个应用场景中的第k个初始观测信号的翻转量均值，F_i,j表示在第i个应用场景中的第j个子场景的信号总翻转量。Among them, M represents the number of sub-scenes contained in the i-th application scene, W _i,k represents the signal error of the k-th initial observation signal in the i-th application scene, and tF _i,j represents the i-th application scene The total signal turnover per unit time in the jth sub-scene in ,

Represents the mean value of the total signal turnover per unit time in the i-th application scenario, tf _i,j,k represents the reversal per unit time of the k-th initial observation signal in the j-th sub-scene in the i-th application scenario quantity,

Indicates the mean value of the inversion amount of the k-th initial observation signal in the i-th application scenario, and F _i,j indicates the total inversion amount of the signal in the j-th sub-scene in the i-th application scenario.

The signal error W _i,k of the initial observation signal X _k in the initial observation signal group PQ _i of the application scenario S _i can be calculated based on the above formula.

Step 2023: Determine the target observation signal group from the initial observation signal group according to the signal error of each initial observation signal.

In the first embodiment, in the first target application scenario among the N application scenarios, a first specified number of observation signals with the smallest signal error is selected to obtain a screening observation signal group, and the N application scenarios The corresponding N screening observation signal groups are combined to obtain the target observation signal group. Wherein, the first target application scenario is any one of the N application scenarios.

Exemplarily, W observation signals with the smallest error may be selected from each initial observation signal group. Exemplarily, W observation signals are selected from the observation signal group PQ _i corresponding to each application scenario to form a target observation signal group. Wherein, W is the above-mentioned first specified quantity.

In the second embodiment, in the second target application scenario among the N application scenarios, a second specified number of screening observation signal groups with the smallest signal error is selected, and the N screening observation signal groups are used as Target observation signal group. Wherein, the second target application scenario is any one of the N application scenarios.

Exemplarily, W observation signals with the smallest error may be selected from each initial observation signal group. Exemplarily, W observation signals are selected from the observation signal group PQ _i corresponding to each application scenario to form a target observation signal group. Wherein, W is the second specified quantity mentioned above.

Optionally, the value of W may be determined according to the data signal used in the simulation process. Exemplarily, the more data signals used in the simulation, the more correspondingly selected observation signals may also be. Of course, the value of W may not be limited by the data signal used in the simulation. For example, the value of W may be 2, 3, 4 and so on.

In this embodiment, if the observation signals in the N screening observation signal groups are all different, the number of observation signals in the target observation signal group obtained in the above-mentioned first implementation manner and the second implementation manner is the same.

Step 203, from the first inversion data, determine second inversion data of each target observation signal in the target observation signal group in each application scenario in the first specified number of application scenarios.

Optionally, the target observation signal group is part of the digital signals used in the simulation performed in step 201 . Therefore, the second inversion data corresponding to the target observation signal group also belongs to a part of the first inversion data. Therefore, the second inversion data in each application scenario of the specified number of application scenarios of the target observation signal group may be selected from the first inversion data.

In this embodiment, the second inversion data of the target observation signal group may include inversion data of the first type and inversion data of the second type.

Exemplarily, the first type of inversion data represents the inversion amount of each target observation signal of the target observation signal group in each sub-scene in the specified number of application scenarios. Optionally, the first type of inversion data includes an inversion amount per unit time of each observation signal in the target observation signal group in each sub-scene.

Exemplarily, the second type of inversion data represents a total inversion amount of the target observation signal group in each application scenario in the specified number of application scenarios. Optionally, the second type of inversion data may also include the total inversion amount per unit time of all observation signals in the target observation signal group in each sub-scene.

Step 204, calculating correction parameters and signal weights corresponding to each target observation signal of the target observation signal group according to the second flip data.

In an implementation manner, the first matrix may first be constructed according to D sets of first-type inversion data in the second inversion data.

Wherein, D represents the number of sub-scenes in the specified number of application scenarios.

Exemplarily, the first matrix constructed by the first type of flipped data can be expressed as:

Among them, D represents the number of sub-scenes used in the simulation, R represents the number of observation signals in the target observation signal group, A represents the first matrix, stf _{D, R} represents the unit time of the R-th observation signal in the D-th sub-scene Inversion amount, stf _i,j represents the unit time inversion amount of the jth observed signal in the i-th sub-scene.

In this embodiment, the second matrix may be constructed according to D sets of second-type inversion data in the second inversion data.

Exemplarily, the construction of the second matrix of the second type of flipped data can be expressed as:

Wherein, B represents the second matrix, and stF _i represents the total turnover per unit time of the target observation signal group in the ith sub-scene.

In this embodiment, the signal weights and correction parameters corresponding to the target observation signal group can be calculated according to the first matrix and the second matrix.

Exemplarily, the relationship between the first matrix and the second matrix, the signal weight and the correction parameter can be expressed by the following formula:

P = (A ^T A) ^-1 A ^T B;

Among them, P represents the signal weight and correction parameter matrix, _ri represents the signal weight corresponding to the i-th target observation signal, and C represents the correction parameter.

In another implementation manner, the signal weights and correction parameters corresponding to the target observation signal group may be calculated by using the gradient descent method to minimize the residual function.

P _n =(1+s×ΔJ)×P _o ;

Among them, P _n represents the current updated observation group parameters, P _o represents the observation group parameters before update, s represents the progressive step, etF _i represents the turnover amount in the i-th scene calculated according to the current power consumption detection model . The above observation group parameters in the currently updated observation group parameters and the observation group parameters before update include signal weights and correction parameters corresponding to target observation signal groups in corresponding states.

In step 205, a power consumption detection model is determined according to the calibration parameters and the signal weights by using the inversion amount of each target observation signal in the target observation signal group as an independent variable.

Exemplarily, the above power consumption detection model is used to calculate the power consumption of the chip.

Exemplarily, a multivariate first-order model may be used to construct a power consumption detection model.

Optionally, a calculation formula for the total inversion amount may be constructed based on the correction parameters and the signal weights by using the inversion amount of each target observation signal in the target observation signal group as an independent variable.

Exemplarily, the calculation formula of the total turnover amount can be expressed as:

Among them, tF represents the total flip amount of the chip to be detected calculated according to the calculation formula of the total flip amount, and tf _i represents the flip amount of the i-th target observation signal per unit time under the chip to be detected.

In this embodiment, a power consumption detection model can be constructed according to the above calculation formula of the total turnover amount.

Exemplarily, the power consumption detection model can be expressed as:

Among them, d represents the process coefficient.

Exemplarily, the specific value of the process coefficient d can be set according to the cell library of the process, and the embodiment of the present application is not limited to the value of the process coefficient d.

Exemplarily, other linear models or nonlinear models may also be used to construct the power consumption detection model.

Based on the above-mentioned power consumption detection model, the chip that needs power consumption detection can be detected. Further, in order to improve the detection accuracy of the power consumption detection model, the constructed power consumption detection model may also be verified.

Optionally, as shown in Figure 3, the method for building a power consumption detection model in the embodiment of the present application may further include:

Step 206, calculating a model error of the power consumption detection model according to the second flip data of each target observation signal.

Exemplarily, the model error of the power consumption detection model is calculated by the following formula:

Among them, e represents the model error of the power consumption detection model, D represents the number of application scenarios involved in the simulation, stF _i represents the total turnover of the target observation signal group in the ith scene, etF _i represents The flip amount in the i-th scene calculated by the model.

Step 207, judging whether the model error is greater than a preset error value.

If the model error is greater than the preset error value, return to step 202 to reselect the latest target observation signal group, and perform steps 202-205 again to update the constructed power consumption detection model.

Optionally, when step 202 is re-executed and the latest target observation signal group is reselected, the number of observation signals selected in each sub-scene may be adjusted.

If the model error is not greater than the preset error value, the current power consumption detection model is output to detect the power consumption of the chip.

Optionally, the value of the above-mentioned preset error value can be set according to requirements. For example, the preset error value may be 5%, 7%, 4% and so on.

Through the power consumption detection model construction method in the embodiment of the present application, the power consumption of the chip can be detected by the determined observation signal. Compared with the existing observation signal selection, there is no uniform standard. Personal experience is used to select observation signals (such as CPU cache miss and other signals). The effectiveness of observation signals depends heavily on the designer's understanding of chip design and personal ability. In contrast, the observation signal provided by the embodiments of the present application The selection method is more objective, so that the selected observation signal is also more reliable.

Further, the method provided by the embodiment of the present application solves the problem of estimating the inversion amount of the data signal. Using the method provided by the embodiment of the present application can streamline and standardize the estimation of dynamic power consumption, and can improve the accuracy of the estimation result of chip power consumption. Furthermore, the method provided by the embodiment of the present application can be used to estimate the power consumption of the chip without knowing the internal design details of the chip design, and can quickly provide an accurate power consumption estimation model.

Embodiment three

Based on the same application concept, the embodiment of the present application also provides a power consumption detection model construction device corresponding to the power consumption detection model construction method, because the problem-solving principle of the device in the embodiment of the present application is consistent with the aforementioned power consumption detection model construction method The embodiments are similar, so the implementation of the device in this embodiment can refer to the description in the embodiment of the above method, and the repetition will not be repeated.

Please refer to FIG. 4 , which is a schematic diagram of functional modules of an apparatus for constructing a power consumption detection model provided in an embodiment of the present application. Each module in the device for constructing a power consumption detection model in this embodiment is used to execute each step in the above method embodiment. The device for constructing a power consumption detection model includes: a simulation module 301, a screening module 302, a first determination module 303, a first calculation module 304, and a second determination module 305; wherein,

The simulation module 301 is configured to simulate the target function module in a first specified number of application scenarios, so as to obtain the first flip data of the data signal in each application scenario in the first specified number of application scenarios, the The target functional module is a functional module in a chip design;

A screening module 302, configured to screen data signals according to the first flip data to determine a target observation signal group, where the target observation signals include a plurality of target observation signals;

The first determination module 303 is configured to determine, from the first flip data, the second value of each target observation signal in the target observation signal group under each application scenario in the first specified number of application scenarios. flip data;

The first calculation module 304 is configured to calculate correction parameters and signal weights corresponding to each target observation signal of the target observation signal group according to the second flip data;

The second determination module 305 is configured to use the inversion amount of each target observation signal in the target observation signal group as an independent variable, and determine a power consumption detection model according to the correction parameter and the signal weight, and the power consumption detection The model is used to calculate the power consumption of the chip.

In a possible implementation manner, the screening module 302 includes: an initial screening unit, an error determination unit, and a signal determination unit.

An initial screening unit, configured to perform primary screening on the data signals used in the first specified number of application scenarios according to a preset value interval, so as to obtain an initial observation signal group;

an error determination unit, configured to determine the signal error of each initial observation signal according to the initial observation signal group;

A signal determining unit, configured to determine the target observation signal group from the initial observation signal group according to signal errors of the respective initial observation signals.

In a possible implementation manner, the specified number of application scenarios includes N application scenarios, and the N application scenarios include M sub-scenes; the signal determination unit is configured to:

In each of the sub-scenes, the observation signals whose flipping amount is in the preset value range are screened out to obtain the observation signal group corresponding to each sub-scene;

For each application scenario, taking the intersection of the observation signal groups corresponding to each of the sub-scenes in the application scenario to obtain the observation signal groups corresponding to the application scenarios, so as to obtain the N observation signal groups corresponding to the N application scenarios, The N observation signal groups are the initial observation signal groups, where M and N are positive integers greater than one.

In a possible implementation manner, the device for constructing a power consumption detection model in this embodiment further includes: a second value determination module 305, configured to:

calculating the flipping amount per unit time of each data signal in each of the sub-scenes;

Determining the non-zero intermediate number of inversion amount per unit time of each data signal in each sub-scene;

The preset value interval is determined according to the non-zero intermediate number of the flip amount.

In a possible implementation manner, the signal determination unit is configured to:

In the target application scenario among the N application scenarios, select a second specified number of observation signals with the smallest signal error to obtain a screening observation signal group, and the target application scenario is any of the N application scenarios. an application scenario;

The target observation signal group is obtained by unioning the N screening observation signal groups corresponding to the N application scenarios.

In a possible implementation manner, the signal error of each initial observation signal is determined by the following formula:

表示第i个应用场景中的单位时间内的信号总翻转量均值，tf_i,j,k表示第i个应用场景中的第j个子场景中的第k个初始观测信号的单位时间内的翻转量，

表示第i个应用场景中的第k个初始观测信号的翻转量均值。Among them, M represents the number of sub-scenes contained in the i-th application scene, W _i,k represents the signal error of the k-th initial observation signal in the i-th application scene, and tF _i,j represents the i-th application scene The total signal turnover per unit time in the jth sub-scene in ,

Represents the mean value of the total signal turnover per unit time in the i-th application scenario, tf _i,j,k represents the reversal per unit time of the k-th initial observation signal in the j-th sub-scene in the i-th application scenario quantity,

Indicates the mean value of the flipping amount of the kth initial observation signal in the i-th application scenario.

In a possible implementation manner, the first computing module 304 is configured to:

Construct a first matrix according to D sets of first-type inversion data in the second inversion data, the D represents the number of sub-scenes in the specified number of application scenarios, and the first-type inversion data represents a target observation signal The turnover amount of each target observation signal of the group under each sub-scene in the specified number of application scenarios;

Constructing a second matrix according to D sets of second-type inversion data in the second inversion data, the second-type inversion data indicating that the target observation signal group is in each application scenario in the specified number of application scenarios The total amount of turnover;

Signal weights and correction parameters corresponding to the target observation signal group are calculated according to the first matrix and the second matrix.

In a possible implementation manner, the device for constructing a power consumption detection model in this embodiment further includes: an error evaluation module, configured to:

calculating a model error of the power consumption detection model according to the second flip data of each target observation signal;

judging whether the model error is greater than a preset error value;

If the model error is greater than a preset error value, reselect the latest target observation signal group to update the constructed power consumption detection model.

In a possible implementation manner, the model error of the power consumption detection model is calculated by the following formula:

Among them, e represents the model error of the power consumption detection model, D represents the number of application scenarios involved in the simulation, stF _i represents the total turnover of the target observation signal group in the ith scene, etF _i represents The flip amount in the i-th scene calculated by the model.

Embodiment four

Please refer to FIG. 5 , which is a flowchart of a method for detecting power consumption provided by an embodiment of the present application. The specific process shown in FIG. 5 will be described in detail below.

Step 401, acquiring the inversion value of each observation signal of the chip to be inspected in the target observation signal group within a unit time.

Step 402: Based on the power consumption detection model obtained by the power consumption detection model construction method, the power consumption of the chip to be detected is calculated according to the inversion value of each observed signal within a unit time.

In the power consumption detection method provided in the embodiment of the present application, the detection efficiency of the chip power consumption can be improved by using a power consumption detection model to check the power consumption of the chip.

Embodiment five

Based on the same application concept, the embodiment of the present application also provides a power consumption detection device corresponding to the power consumption detection method. For the implementation of the device in the embodiment, reference may be made to the description in the embodiment of the above method, and repeated descriptions will not be repeated.

Please refer to FIG. 6 , which is a schematic diagram of functional modules of a power consumption detection device provided in an embodiment of the present application. Each module in the power consumption detecting device in this embodiment is used to execute each step in the above method embodiment. The power consumption detection device includes: an acquisition module 501 and a second calculation module 502; wherein,

The obtaining module 501 is used to obtain the inversion value of each observation signal of the chip to be detected in the target observation signal group in unit time;

The second calculation module 502 is configured to calculate the power consumption of the chip to be tested based on the power consumption detection model obtained by the power consumption detection model construction method according to the inversion value of each observation signal within a unit time.

In addition, an embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is run by a processor, the power consumption detection model construction described in the above-mentioned method embodiments is executed. method and/or steps of a power consumption detection method.

The computer program product of the power consumption detection model building method and the power consumption detection method provided in the embodiments of the present application includes a computer-readable storage medium storing program codes, and the instructions included in the program codes can be used to execute the above method embodiments For the steps of the power consumption detection model building method or the power consumption detection method, please refer to the above method embodiments for details, and details are not repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed devices and methods may also be implemented in other ways. The device embodiments described above are only illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functions and possible implementations of devices, methods and computer program products according to multiple embodiments of the present application. operate. In this regard, each block in a flowchart or block diagram may represent a module, program segment, or part of code that includes one or more Executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved. It should also be noted that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations, can be implemented by a dedicated hardware-based system that performs the specified function or action , or may be implemented by a combination of dedicated hardware and computer instructions.

In addition, each functional module in each embodiment of the present application may be integrated to form an independent part, each module may exist independently, or two or more modules may be integrated to form an independent part.

If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disc, etc., which can store program codes. . It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is a relationship between these entities or operations. There is no such actual relationship or order between them. Furthermore, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article, or apparatus comprising a set of elements includes not only those elements, but also includes elements not expressly listed. other elements of or also include elements inherent in such a process, method, article, or device. Without further limitations, an element defined by the statement "comprising..." does not exclude the presence of additional same elements in the process, method, article or device comprising said element.

The above descriptions are only preferred embodiments of the present application, and are not intended to limit the present application. For those skilled in the art, there may be various modifications and changes in the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the protection scope of this application. It should be noted that like numerals and letters denote similar items in the following figures, therefore, once an item is defined in one figure, it does not require further definition and explanation in subsequent figures.

The above is only a specific implementation of the application, but the scope of protection of the application is not limited thereto. Anyone familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the application. Should be covered within the protection scope of this application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims (14)

1. The power consumption detection model construction method is characterized by comprising the following steps of:

simulating a target functional module in a first appointed number of application scenes to obtain first turnover data of data signals in each application scene in the first appointed number of application scenes, wherein the target functional module is one functional module in a chip design;

Screening the data signals according to the first overturn data to determine a target observation signal group, wherein the target observation signals comprise a plurality of target observation signals;

determining second overturn data of each target observation signal in the target observation signal group in each application scene in the first appointed number of application scenes from the first overturn data;

calculating to obtain correction parameters and signal weights corresponding to all target observation signals of the target observation signal group according to the second overturn data;

and determining a power consumption detection model by taking the turnover amount of each target observation signal in the target observation signal group as an independent variable according to the correction parameters and the signal weights, wherein the power consumption detection model is used for calculating the power consumption of the chip.

2. The method of claim 1, wherein screening the data signals according to the first flipping data to determine a set of target observed signals comprises:

primary screening is carried out on data signals used in the first specified number of application scenes according to a preset numerical value interval, so that an initial observation signal group is obtained, wherein the initial observation signal group comprises a plurality of initial observation signals;

Determining signal errors of all initial observation signals according to the initial observation signal group;

and determining the target observation signal group from the initial observation signal groups according to the signal errors of the initial observation signals.

3. The method of claim 2, wherein the first specified number of application scenarios includes N application scenarios, and wherein the N application scenarios include M sub-scenarios; the primary screening is performed on the data signals used in the first specified number of application scenarios according to the preset numerical value interval, so as to obtain an initial observation signal group, including:

screening observation signals with turnover quantity in the preset numerical value interval from each sub-scene to obtain an observation signal group corresponding to each sub-scene;

and aiming at each application scene, taking intersections of the observation signal groups corresponding to the sub-scenes in the application scene to obtain N observation signal groups corresponding to the application scene, wherein the N observation signal groups are the initial observation signal groups, and M and N are positive integers larger than one.

4. A method according to claim 3, wherein the predetermined value interval is determined by:

Calculating the turnover amount of each data signal in each sub-scene in unit time;

determining the non-zero intermediate number of the turnover amount of each data signal in each sub-scene in unit time;

and determining the preset numerical value interval according to the non-zero turning quantity intermediate number.

5. A method according to claim 3, wherein said determining said set of target observed signals from said set of initial observed signals based on signal errors of said respective initial observed signals comprises:

selecting a second specified number of observation signals with the minimum signal error from target application scenes in the N application scenes to obtain a screening observation signal group, wherein the target application scene is any one application scene in the N application scenes;

and combining N screening observation signal groups corresponding to the N application scenes to obtain the target observation signal group.

6. The method of claim 2, wherein the signal error of each of the initial observed signals is determined by the following equation:

wherein M represents the number of sub-scenes included in the ith application scene, W _i,k Signal error, tF, representing the kth initial observed signal in the ith application scenario _i,j Representing the total signal flip amount per unit time in the jth sub-scene in the ith application scene,

representing the average value of the total signal turning quantity in unit time in the ith application scene, tf _i,j,k Representing the amount of flip per unit time of the kth initial observation signal in the jth sub-scene in the ith application scene, +.>

And representing the average value of the turning quantity of the kth initial observation signal in the ith application scene.

7. The method according to claim 1, wherein the calculating, according to the second flip data, a correction parameter and a signal weight corresponding to each target observation signal of the target observation signal group includes:

constructing a first matrix according to D groups of first-class overturn data in the second overturn data, wherein D represents the number of sub-scenes in the appointed number of application scenes, and the first-class overturn data represents the overturn amount of each target observation signal in the target observation signal group in each sub-scene in the appointed number of application scenes;

constructing a second matrix according to D groups of second-class overturn data in the second overturn data, wherein the second-class overturn data represents the total overturn amount of the target observation signal group in each application scene in the appointed number of application scenes;

And calculating to obtain the signal weight and the correction parameter corresponding to the target observation signal group according to the first matrix and the second matrix.

8. The method as recited in claim 1, further comprising:

calculating a model error of the power consumption detection model according to the second turnover data of each target observation signal;

judging whether the model error is larger than a preset error value or not;

and if the model error is larger than a preset error value, reselecting the latest target observation signal group to update the constructed power consumption detection model.

9. The method of claim 8, wherein the model error of the power consumption detection model is calculated by the formula:

where e represents a model error of the power consumption detection model, D represents the number of application scenarios involved in the simulation, stF _i Representing the total flip amount of the target observation signal group in the ith scene etF _i And representing the calculated turning quantity in the ith scene according to the power consumption detection model.

10. A power consumption detection method, comprising:

acquiring turnover values of each observation signal of a chip to be detected in a target observation signal group in unit time;

Based on the power consumption detection model obtained by the method according to any one of claims 1-9, calculating to obtain the power consumption of the chip to be detected according to the turnover value of each observation signal in unit time.

11. A power consumption detection model construction apparatus, characterized by comprising:

the simulation module is used for simulating the target functional module in a first appointed number of application scenes to obtain first turnover data of data signals in each application scene in the first appointed number of application scenes, and the target functional module is one functional module in a chip design;

the screening module is used for screening the data signals according to the first turnover data so as to determine a target observation signal group;

the first determining module is used for determining second turnover data of each target observation signal in the target observation signal group in each application scene in the first appointed number of application scenes from the first turnover data;

the first calculation module is used for calculating and obtaining correction parameters and signal weights corresponding to all target observation signals of the target observation signal group according to the second overturn data;

The second determining module is configured to determine a power consumption detection model according to the correction parameter and the signal weight by using the turnover amount of each target observation signal in the target observation signal group as an independent variable, where the power consumption detection model is used to calculate power consumption of a chip.

12. A power consumption detection apparatus, characterized by comprising:

the acquisition module is used for acquiring the turnover value of each observation signal in the target observation signal group of the chip to be detected in unit time;

the second calculation module is configured to calculate, based on the power consumption detection model obtained by the method according to any one of claims 1 to 9, the power consumption of the chip to be detected according to the turnover value of each observation signal in unit time.

13. An electronic device, comprising: a processor, a memory storing machine-readable instructions executable by the processor, which when executed by the processor perform the steps of the method of any of claims 1 to 10 when the electronic device is run.

14. A computer-readable storage medium, characterized in that it has stored thereon a computer program which, when executed by a processor, performs the steps of the method according to any of claims 1 to 10.

Images