JP2012190447A - Power consumption analysis system and application development tool - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily analyze power consumption of a terminal.SOLUTION: A power consumption analysis system 1 according to the invention includes: a resource consumption ratio calculation part 11 for acquiring a trace log consisting of execution history of respective methods invoked by an application program and a resource log in which the amount of resource consumption is recorded for respective hardware during the execution of a method and calculating a resource consumption ratio to respective hardware for respective methods on the basis of the trace log and resource log; a resource power consumption calculation part 12 for calculating power consumption for respective hardware on the basis of the resource log; and a method power consumption calculation part 13 for, on the basis of the power consumption for respective hardware and resource consumption ratio to respective hardware for respective methods, calculating power consumption for respective methods.

Description

  The present invention relates to a power consumption analysis system and application development tool for analyzing power consumption of a terminal.

  Battery-powered wireless communication terminals are designed based on a pre-evaluation of power consumption so that the limited power of the battery can be efficiently consumed from the time of hardware and software development to the time of use by the end user. It is preferable. Conventionally, attempts have been made to estimate power consumption at the source code level in order to evaluate power consumption of application programs used in wireless communication terminals. Patent Document 1 below discloses a method for calculating power consumption based on hardware specified by a registry that becomes a trigger when an application program is executed. Patent Document 2 below discloses a method for determining the power consumption of each module by analyzing the source code of the program and setting the power consumption of one operation instruction to 1, and summing the power consumption in the modules. ing.

JP 2009-75965 A JP 2008-186163 A

  However, in the power consumption evaluation method disclosed in Patent Document 1, it is necessary to analyze the relationship between the registry and hardware. It is also necessary to prepare a table for determining the power consumption for each hardware. For this reason, it is necessary to perform analysis each time the wireless communication terminal to be evaluated is changed. In the power consumption evaluation method disclosed in Patent Document 2, it is necessary to analyze the source code of the program. For this reason, even when the source code of the program is slightly changed, it is necessary to analyze the source code each time. In these evaluation methods, when various application programs are developed in various wireless communication terminals, it is not possible to easily evaluate desired power consumption.

  Accordingly, an object of the present invention is to provide a power consumption analysis system and an application development tool that can easily analyze the power consumption of a terminal in order to solve such problems.

  In order to solve the above problems, a power consumption analysis system of the present invention is a power consumption analysis system for analyzing power consumption of an application program executed on a terminal, and includes an execution history of each module called by the application program. Obtain the trace log and the resource log that records the resource consumption for each hardware when the module is executed, and calculate the resource consumption rate for each hardware for each module based on the trace log and resource log Based on resource consumption ratio calculation means, resource power consumption calculation means for calculating power consumption for each hardware based on resource log, power consumption for each hardware, and resource consumption ratio for each hardware for each module , Module Characterized in that it and a module power consumption calculation means for calculating the power consumption.

  According to the present invention, the power consumption for each module can be calculated based on the resource log and the trace log. Both the resource log and the trace log are information that does not depend on the terminal. Therefore, it is possible to calculate the power consumption in units of modules for various hardware and application programs.

  In the power consumption analysis system of the present invention, the resource log is preferably a first resource log in which resource consumption for the entire terminal is recorded. Thus, by using the first resource log, it is possible to acquire the resource consumption amount for the entire terminal.

  In the power consumption analysis system of the present invention, it is preferable that the resource consumption rate calculation means calculates a resource consumption rate for each module based on the trace log. The calculation method of the resource consumption ratio for each module includes a method in which the ratio of the module's total CPU time to the resource consumption ratio is used, and a method in which the ratio of the target module to the total number of executions is set to the resource consumption ratio. is there. The module CPU time and the resource consumption recorded in the first resource log are in a proportional relationship. Therefore, the CPU time of each module can be calculated from the trace log, and the ratio of the module CPU time to the total CPU time can be used as the resource consumption ratio of the module. Similarly, the number of module executions and the resource consumption recorded in the first resource log are in a proportional relationship. Therefore, the module execution count can be calculated from the trace log, and the reciprocal thereof can be used as the resource consumption ratio of the module.

  In the power consumption analysis system of the present invention, the resource log is preferably a second resource log in which resource consumption for each piece of hardware is recorded for each module. In this way, by using the second resource log, it is possible to acquire the resource consumption for each piece of hardware for each module.

  Further, in the power consumption analysis system of the present invention, the resource consumption ratio calculation means corrects the resource consumption for each hardware recorded for each module recorded in the second resource log based on a coefficient corresponding to the hardware. preferable. In the second resource log, it may not be possible to acquire all of the hardware resource consumption used by the application program. In such a case, the resource consumption for each hardware stored for each module stored in the second resource log is multiplied by a coefficient corresponding to the hardware or added to each hardware resource for each module. The consumption can be corrected.

  In the power consumption analysis system of the present invention, it is preferable to further include display means for displaying the power consumption for each module. According to this, the power consumption for each module can be visualized.

  In the power consumption analysis system of the present invention, the display means includes a power consumption call tree display means for creating a call tree based on the trace log and displaying the power consumption for each module superimposed on the call tree. preferable. Thus, by displaying the power consumption in the form of a call tree, it is possible to display the power consumption together with the parent-child relationship of the modules. For this reason, the module which becomes a bottleneck about power consumption can be clarified.

  In the power consumption analysis system of the present invention, it is preferable that the power consumption call tree display means displays the power consumption of the module in a specified period. In this way, the power consumption call tree for a specified period can be displayed, so that the module that becomes the bottleneck can be further clarified.

  In the power consumption analysis system of the present invention, it is preferable that the display means includes a graph display means for displaying the power consumption for each hardware in a predetermined time in a predetermined graph. By displaying the power consumption for each hardware in a predetermined time in a graph, it is possible to clarify the hardware that has become a bottleneck in terms of power consumption.

  In the power consumption analysis system of the present invention, the display means further includes power consumption call tree display means for creating a call tree based on the trace log and displaying the power consumption for each module superimposed on the call tree. The power consumption call tree display means preferably displays the call tree in accordance with the portion selected in the graph displayed by the graph display means. In this way, by displaying the call tree of the selected part, it is possible to further clarify the module that becomes the bottleneck. It is also possible to display a call tree on which the power consumption of the selected hardware is superimposed, and to present a more detailed power consumption analysis result.

  In the power consumption analysis system of the present invention, the graph display means preferably displays a bar graph or a pie graph. Thus, by displaying the power consumption for each hardware in a predetermined time in a bar graph format or a pie graph format, it is possible to clarify the hardware that is a bottleneck with respect to power consumption. The graph display means may display a graph in real time while executing an application program whose power consumption is to be analyzed. The graph display means may display the graph of the part selected by the developer after executing the application program.

  In the power consumption analysis system of the present invention, the terminal is preferably a wireless communication terminal. It is not easy for an end user who uses an application program on a wireless communication terminal to recognize how much power is consumed by executing the application program or transmitting / receiving data. Moreover, it is difficult for an end user to suitably use limited power in a battery-driven wireless communication terminal. However, according to the present invention, it is possible to calculate the power consumption for each module of the application program executed on the actual wireless communication terminal by using the resource log or trace log of the wireless communication terminal. In this way, it is possible to provide an evaluation result of power consumption when using the wireless communication terminal.

  In the power consumption analysis system of the present invention, the terminal is preferably a virtual terminal. According to this, on the information processing terminal on which the power consumption analysis system is executed, a virtual terminal having virtual hardware can be reproduced by software, and the resource log and trace log of the virtual terminal can be used. is there. For this reason, it is possible to calculate the power consumption of the application program without preparing an actual wireless communication terminal. In this way, it is possible to provide an evaluation result of power consumption when developing an application program installed in a wireless communication terminal.

  In the power consumption analysis system of the present invention, it is preferable that the second resource log is acquired from the low-level API of the virtual terminal. According to this, by processing in advance so that the second resource log can be output to the low-level API used by the virtual terminal, the resource consumption amount used by each module can be accurately calculated. As a result, the accuracy of power consumption calculated for each module is improved.

  The application development tool according to the present invention is received by an input unit that receives user input, a source code editing unit that edits a source code of an application program based on an input received by the input unit, and an input unit. Source code execution means for executing the source code edited by the source code editing means based on the input, and the power consumption analysis system for analyzing the power consumption of the source code executed by the source code execution means It is characterized by. According to this, power consumption can be evaluated at the time of development of an application program installed in a wireless communication terminal.

  According to the present invention, power consumption of a terminal can be easily analyzed.

It is a block diagram which shows the function structure of the power consumption analysis system which concerns on embodiment of this invention. It is a figure which shows an example of a 1st resource log. It is a figure which shows an example of a 2nd resource log. It is a figure which shows an example of the data communication volume by CPU time and Wi-Fi of each method. It is a figure which shows an example of the GPS operation time of each method. It is a figure which shows an example of the output of a resource consumption ratio calculation part. (A) A figure showing an example of a call tree, (b) A figure showing an example of CPU time. It is a figure which shows an example of the power consumption evaluation value which shows the power consumption of the whole terminal. It is a figure which shows an example of the output of a method power consumption calculation part. It is a block diagram which shows the function structure of a display part. It is a figure which shows an example of a pie chart. It is a figure which shows an example of a bar graph. It is a figure which shows an example of a power consumption call tree. It is a flowchart which shows the process sequence of the power consumption analysis system of FIG. It is a block diagram which shows the function structure of an application development tool.

  DESCRIPTION OF EMBODIMENTS Hereinafter, a preferred embodiment of a power consumption analysis system according to the present invention will be described in detail with reference to the drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a block diagram showing a functional configuration of a power consumption analysis system 1 of the present embodiment. The power consumption analysis system 1 is a system that analyzes power consumption of an application program executed in a terminal 2 such as a wireless communication terminal. The power consumption analysis system 1 calculates the power consumption for each module used in the application program to be evaluated using the resource log and the trace log for each module acquired from the terminal 2. The power consumption analysis system 1 is an information processing device such as a server device, for example.

  Here, the module is a program component for realizing a predetermined function in the application program, and the application program includes a plurality of modules. This module may be called by other modules and may call other modules to implement the module's functionality. Depending on the type of programming language used to describe the application program, a module is called a class, method, subroutine or function. For example, in Java (registered trademark), modules are called classes and methods. A class includes one or more methods and a data structure, and uses these to realize one or more functions. As described above, the modules used in the application program have different names and granularities depending on the program language, but there is no essential difference in the present invention. Therefore, in the present embodiment, description will be made using a method as a module.

  First, the configuration of the terminal 2 will be described. The terminal 2 is a wireless communication terminal, for example, and physically includes a terminal HW22 (corresponding to “hardware” in the claims). The terminal HW22 is a hardware component of the terminal 2 (hereinafter may be simply referred to as “hardware”), and includes, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access Memory). A communication device for transmitting / receiving data to / from an external device (for example, power consumption analysis system 1), a GPS (Global Positioning System) for measuring the current position of the terminal 2, a storage for storing various data, and a predetermined And a camera for imaging the subject. The CPU of the terminal HW22 controls the terminal 2 in an integrated manner by loading a predetermined computer program stored in a built-in memory such as a ROM of the terminal HW22 into the RAM of the terminal HW22 and executing it. The communication device of the terminal HW22 is, for example, Wi-Fi (Wireless Fidelity) that performs wireless communication.

  Functionally, the terminal 2 includes a software execution unit 21, a software execution state management unit 23, and a component state management unit 24. The software execution unit 21, the software execution state management unit 23, and the component state management unit 24 are functions realized when the CPU of the terminal 2 executes a computer program stored in a built-in memory such as a ROM of the terminal 2. is there.

  The software execution unit 21 executes an application program (a computer program that can be executed by the CPU of the terminal HW22) stored in a built-in memory such as a ROM of the terminal 2 (or a recording medium that is detachable from the terminal 2). For the application program executed by the software execution unit 21, the software execution state management unit 23 acquires information on a method called by the application program from the software execution unit 21. Then, the software execution state management unit 23 generates a trace log indicating the execution history of the method called by the application program. In the trace log, as the execution history of each method, for example, the time when the method is called and the time when the method is finished (return time to the calling method) are recorded.

  The component state management unit 24 acquires information indicating the operation state of the terminal HW22, and generates a resource log indicating the operation state of the terminal HW22 based on the acquired information. This resource log is information provided from the OS, for example, and is not dependent on the terminal model. The resource log includes, for example, the operation state of each hardware component of the terminal HW22, such as CPU time, Wi-Fi transmission / reception data amount, storage read / write data amount, time when the GPS and camera are turned on and off, and the like. This is a log that records resource consumption such as processing amount.

  In the present embodiment, the resource log includes, for example, two types of logs: a first resource log and a second resource log. The first resource log is a log in which the resource consumption amount in the entire terminal 2 is recorded. That is, the first resource log is a log in which the resource consumption of the terminal 2 as a whole is recorded for each hardware component of the terminal HW22. The first resource log is, for example, at regular intervals (for example, every second) based on information provided from an OS (Operating System) (for example, / proc information when the OS is Linux (registered trademark)). To be acquired.

FIG. 2 is a diagram illustrating an example of the first resource log. As shown in FIG. 2, the first resource log includes a plurality of pieces of first log information. The first log information includes, for example, the time when the first log information is acquired, the CPU operation rate, the LCD (Liquid Crystal Display) luminance, the amount of storage read / write data, the amount of wireless transmission / reception data, the GPS operation time, and the camera operation Resource consumption such as time. The first log information L1 in FIG. 2 is log information for one second at 15:43:48 on Dec. 6, 2010. The CPU operation rate is 100%, the LCD brightness is 20 cd / m ² , and reading from the storage is performed. This indicates that the amount of data is 0 bytes, the amount of data written to the storage is 0 bytes, the amount of wireless transmission data is 36716 bytes, and the amount of wireless reception data is 242490 bytes. The unit of the first log information is not limited to the above, and a different unit can be used as necessary.

  The second resource log is a log that records, for each method call that occurs during execution of the application program, the amount of resource consumed by the method or whether the hardware is being used. By processing a low-level API (Application Program Interface) corresponding to a specific method so that data handled in the API is output as a log, the component state management unit 24 allows the specific method to be executed. The resource consumption amount is acquired, and a second resource log is generated.

  FIG. 3 is a diagram illustrating an example of the second resource log. The second resource log is composed of a plurality of pieces of second log information. The second log information includes a time when the second log information is acquired, a resource consumption amount such as a communication data amount per unit time, and a method name. For example, the second log information L2 in FIG. 3 is log information for one second at 12:34:56, and the method “org / apache / harmony / luni / platform / OSNetworkSystem.sendStream” is 111 bytes / ms. Indicates that communication has been performed.

  Next, the configuration of the power consumption analysis system 1 will be described. Functionally, the power consumption analysis system 1 includes a resource consumption rate calculation unit 11 (corresponding to “resource consumption rate calculation means” in the claims) and a resource power consumption calculation unit 12 (“resources in the claims” Equivalent to "power consumption calculating means"), method power consumption calculating section 13 (corresponding to "module power consumption calculating means" in claims), and display section 14 (corresponding to "display means" in claims). It is comprised including.

The resource consumption rate calculation unit 11 functions as a resource consumption rate calculation unit that acquires a trace log and a resource log, and calculates a resource consumption rate for each hardware for each method based on the trace log and the resource log. More specifically, the resource consumption rate calculation unit 11 first determines the resource consumption C for specific hardware of the target method X (X is an identification number assigned to the method of the application program) from the second resource log. _{Get method_X} . Further, the resource consumption ratio calculation unit 11 acquires the resource consumption amount C _all of _all methods from the second resource log. Then, the resource consumption ratio calculation unit 11 calculates the resource consumption ratio α _{resource_X} for the specific hardware of the target method X based on the following resource consumption ratio calculation formula (1).
Resource consumption rate α _{resource_X} = C _{method_X} / C _all … (1)

α _{resource —} X represents the ratio of the amount of the amount of consumption of predetermined hardware by the method X of the application program to the entire method. For example, when the hardware is a CPU, the resource consumption C _{method_X} is the CPU time used by the method X in a certain period. The resource consumption C _all represents the CPU time used by the entire application program in a certain period.

FIG. 4 is a diagram illustrating an example of the CPU time of each method and the data communication amount by Wi-Fi within a certain period acquired from the second resource log. In the example of FIG. 4, the CPU time for method A is 200 ticks. The CPU time for the entire method is 600 ticks. Therefore, the CPU consumption ratio α _{CPU_A} of the method A is obtained as 1/3. Note that method A, method B, and method C are all low-level APIs operating in different threads.

FIG. 5 is a diagram illustrating an example of the GPS operation time of each method acquired from the second resource log. In the example of FIG. 5, the GPS operating time of method A is 1 and the GPS operating time of method B is 0.5 in a unit time (a period of 1 to 2). Therefore, the GPS consumption rate α _{GPS_A} of method A is obtained as 1 / 1.5 = 2/3. Note that method A, method B, and method C are all low-level APIs operating in different threads.

By the way, the resource consumption of each method of the target application program may not be sufficient only by the information included in the second resource log. For example, a method at the time of wireless transmission / reception cannot acquire a data amount such as a packet header. In such a case, the resource consumption rate calculation unit 11 calculates the resource consumption rate for the specific hardware of the target method in consideration of the resource consumption amount not included in the second resource log. That is, the resource consumption ratio calculation unit 11 _corrects the resource consumption C _{method_X} acquired from the second resource log based on the coefficient β _resource and the constant γ _resource . For example, the resource consumption ratio calculation unit 11 _corrects the resource consumption C _{method_X} acquired from the second resource log based on the following equation (2).
Correction resource consumption C _{method_X} '= β _resouce × C _{method_X} + γ _resouce … (2)

Here, the coefficient beta _resouce and constant gamma _resouce is a value determined depending on the hardware (resources). For example, when performing wireless communication, the resource consumption ratio calculation unit 11 obtains the data amount per packet in advance for the header and control information, and uses this as the coefficient β _WiFi . Then, the resource consumption ratio calculation unit 11 multiplies the number of packets obtained at the OS level by the coefficient β _WiFi and sets the constant γ _WiFi to 0, thereby calculating the corrected resource consumption C _{method_X} ′ for the method Wi-Fi. To do. Subsequently, the resource consumption rate calculation unit 11 calculates the resource consumption rate of the target method based on the following equation (3).
Resource consumption rate α _{resource_X} = C _{method_X} '/ C _all … (3)

Here, since all the call relationships up to the low level API are output as a log, the resource consumption rate calculation unit 11 calculates a different resource consumption rate α _{resource_X} if the call source is different even in the same low level API.

FIG. 6 is a diagram illustrating an example of the output D _rate of the resource consumption ratio α _{resource_X} calculated by the resource consumption ratio calculation unit 11. As illustrated in FIG. 6, the output D _rate includes time information indicating the start time of a period to be calculated, a resource consumption ratio α _{resource_X,} and a method name including a method calling relationship. The output D _{rate in} FIG. 6 shows that the consumption rate of the method with the method name “org / apache / harmony / luni / platform / OSNetworkSystem.sendStream” was ½ from 12:34:56 to 1 second. Show.

  On the other hand, for example, when the low-level API cannot be processed in the terminal 2, the component state management unit 24 of the terminal 2 cannot generate the second resource log. Thus, depending on the terminal 2, the resource consumption ratio calculation unit 11 may not be able to acquire the second resource log. In such a case, the resource consumption rate calculation unit 11 calculates the resource consumption rate for each method based on the first resource log and the trace log. A method for calculating the resource consumption rate for each method will be described below.

The first calculation method is a method of calculating based on the number of method calls. More specifically, the resource consumption ratio calculation unit 11 acquires the number of invocations n _t of a method that consumes specific hardware in the period t from the trace log. Then, the resource consumption ratio calculation unit 11 calculates the resource consumption ratio α _{resource_X} for each method in the period t based on the following equation (4).
Resource consumption rate α _{resource_X} = 1 / n _t … (4)

  The second calculation method is a method of calculating based on the CPU time of the method. This method differs in calculation method depending on whether or not specific hardware is a CPU.

When the specific hardware is a CPU, the resource consumption ratio calculation unit 11 acquires the CPU time T _{method_X} of each method that consumes the CPU in the period t from the trace log. In addition, the CPU time total _{Tmain_all} of the highest parent method is acquired from the trace log during the synchronization period t. Here, the CPU time of a method is the time from when the method is called until the execution is completed. Then, the resource consumption ratio calculation unit 11 calculates the resource consumption ratio for each method in the period t based on the following equation (5).
Resource consumption rate α _{resource_X} = T _{method_X} / T _{main_all} … (5)

  FIG. 7A is a diagram illustrating an example of a call tree, and FIG. 7B is a diagram illustrating an example of the CPU time of each method indicated in the call tree of FIG. In the call tree of FIG. 7A, the main method calls method A and method B. Method A calls method A1, and method A1 calls method A2 and method A3. Furthermore, method B calls method B1.

In the example of FIG. 7B, the CPU time _{Tmain_all} of the main method indicates the time during which the main method is executed within the period t. The CPU time of each child method indicates the time from the child method being called by the parent method to the end of execution within the period t. For example, the CPU time T _{method_A2} of the method A2 indicates the time from when the method A2 is called by the method A1 until the execution is completed.

Here, when the specific hardware is a CPU, the parent method operates the CPU by calculation or the like in addition to executing the child method. For this reason, the resource consumption rate (CPU consumption rate) of each method is calculated as the ratio of the CPU time T _{method_X} of each method to the CPU time T _{main_all} of the main method which is the highest method. For example, the CPU consumption rate α _{CPU_A2} of the method A2 is calculated by the following equation.
CPU consumption rate α _{CPU_A2} = T _{method_A2} / T _{main_all}

When the specific hardware is other than the CPU, the resource consumption ratio calculation unit 11 acquires the CPU time T _{method_X} of each method that consumes the specific hardware in the period t from the trace log. In addition, the CPU time T _all of all the low-level APIs that consume specific hardware during the synchronization period t is acquired. Then, the resource consumption ratio calculation unit 11 calculates the resource consumption ratio for each method in the period t based on the following equation (6).
Resource consumption rate α _{resource_X} = T _{method_X} / T _all … (6)

This will be specifically described with reference to the example of FIG. Note that the method A2 and the method B1 are low-level APIs that consume specific hardware (for example, Wi-Fi). Here, when the specific hardware is other than the CPU, the method that consumes the hardware is a low-level API. The parent method does not consume the hardware. Therefore, the resource consumption ratio of each method is calculated as the ratio of the CPU time T _{method_X} of each method to the total CPU time T _all of the low-level API that consumes the hardware. For example, resource consumption ratio alpha _{Resouce_A2} method A2 is calculated by the following equation. In this case, the total CPU time T _all for the specific hardware is the sum of the CPU time T _{method_A2} of the method A2 and the CPU time T _{method_B1} of the method B1.
Resource consumption rate α _{resouce_A2} = T _{method_A2} / (T _{method_A2} + T _{method_B1} )

Note that the method call relationship can be obtained by analyzing the trace log. Therefore, the resource consumption ratio calculation unit 11 performs output in the same format as the output D _rate shown in FIG.

Returning to FIG. 1, the resource power consumption calculation unit 12 calculates the power consumption (including the breakdown for each hardware) of the entire terminal 2 based on the first resource log acquired by the component state management unit 24 of the terminal 2. It functions as resource power consumption calculation means. More specifically, the resource power consumption calculation unit 12 is preset according to the first resource log and each parameter (p _j described later) of the first log information stored in the first resource log. Based on the value of the power coefficient (c _ij described later), the power consumption P _resource of each hardware of the terminal HW22 and the power consumption P of the entire terminal 2 according to the following power consumption evaluation formulas (7) and (8) The evaluation value of _all is calculated.

In the above formulas (7) and (8), i represents a hardware identification number, and is an integer value from 1 to N (N is the total number of components included in the terminal HW22 of the terminal 2). _{Preresource_i} represents the power consumption of the i-th hardware. In the above formula (7), j represents the identification number of the parameter included in the first log information, and 1 to M (M is the total number of parameters included in the first log information). It is an integer value. p _j represents the j-th parameter included in the first log information. c _ij represents a power coefficient indicating the degree of influence of the j-th parameter on the power consumption of the i-th hardware. The power coefficient c _ij is preset in the resource power consumption calculation unit 12. The power coefficient c _ij can be calculated by, for example, a method described in JP 2010-225133 A. The method for calculating the power coefficient c _ij is not limited to this. In the above formula (8), P _all represents the power consumption of the entire terminal 2.

FIG. 8 is a diagram illustrating an example of a power consumption evaluation value indicating the power consumption of the entire terminal 2. As illustrated in FIG. 8, the power consumption evaluation value includes time information and power consumption P _resource of each hardware of the terminal HW22. The time information indicates the start time of the period for which power consumption is calculated. The power consumption _Preresource indicates the power consumption for each hardware until the unit time elapses from the start time indicated by the time information. For example, the power consumption evaluation value Pr in FIG. 8 is calculated from 12:34:56 for 1 second. During this period, the CPU power consumption P _CPU is 0.23 W, the Wi-Fi power consumption P _WiFi is 0.1 W, the LCD power consumption P _LCD is 0.2 W, and the storage P _STORAGE power consumption is 0.1 W. , GPS power consumption P _GPS indicates 0.0 W. Further, the sum of the power consumption of each piece of hardware indicates the power consumption P _all of the terminal 2 as a whole. In this way, the resource consumption ratio calculation unit 11 calculates the power consumption P _resource for each hardware and the power consumption P _all of the terminal 2 as a whole.

Returning to FIG. 1, the method power consumption calculation unit 13 calculates the power consumption P _{method_X} for each method based on the power consumption P _resource for each hardware and the resource consumption ratio α _{resource_X} for each hardware for each method. Functions as a method power consumption calculation means. Specifically, the method power consumption calculation unit 13 calculates the power consumption P _{method_X} of each method based on the following method power consumption calculation formula (9).

In the above equation (9), P _{method_X} represents the power consumption of the method X. i represents a hardware identification number and is an integer value of 1 to N (N is the total number of components included in the terminal HW22 of the terminal 2). _{Preresource_i} represents the power consumption of the i-th hardware. Α _{resource_i_X} represents the consumption rate of the method X with respect to the i-th hardware. For example, when the hardware included in the terminal HW22 is a CPU, Wi-Fi, GPS, and camera, the power consumption of the method A is calculated by the following equation.
P _{method_A} = P _CPU × α _{CPU_A} + P _WiFi × α _{WiFi_A} + P _GPS × α _{GPS_A} + P _camera × α _{camera_A}

FIG. 9 is a diagram illustrating an example of the output D _power including the power consumption for each method in the unit time calculated by the method power consumption calculation unit 13. As illustrated in FIG. 9, the output D _power includes time information indicating the start time of the calculation target, power consumption used by the method, and a method name including a method calling relationship. The output D _{power in} FIG. 9 shows that the _power consumption of the method with the method name “org / apache / harmony / luni / platform / OSNetworkSystem.sendStream” was 0.5 W from 12:34:56 to 1 second. Show.

The method power consumption calculation unit 13 transmits this output D _power to the display unit 14. In addition, the method power consumption calculation unit 13 transmits a trace log including time information indicating the time when the method is called to the display unit 14.

  The display unit 14 functions as a display unit that displays power consumption for each method. FIG. 10 is a block diagram illustrating a functional configuration of the display unit 14. As shown in FIG. 10, the display unit 14 includes a graph display unit 141 (corresponding to “graph display means” in the claims) and a power consumption call tree display unit 142 (“power consumption call tree in the claims”). Equivalent to “display means”).

  The graph display unit 141 functions as a graph display unit that displays power consumption for each hardware in a predetermined time in a predetermined graph. The graph display unit 141 may display a graph in real time while executing an application program whose power consumption is to be analyzed. In addition, the graph display unit 141 may perform graph display of a portion selected by the developer after executing the application program. The graph display unit 141 includes, for example, a pie graph display unit 141a and a bar graph display unit 141b.

The pie chart display unit 141a displays the consumption ratio of power consumption for each hardware at a certain time in a pie chart. More specifically, the pie chart display unit 141a divides the power consumption P _resource of each hardware calculated by the resource power consumption calculation unit 12 by the power consumption P _all of the entire terminal 2 to thereby reduce the power consumption of each hardware. Calculate the consumption rate of. The pie chart display unit 141a displays the power consumption ratio of each hardware in a pie chart. FIG. 11 is a diagram illustrating an example of a pie chart displayed by the pie chart display unit 141a.

  When the part indicating the consumption rate of the predetermined hardware on the displayed pie chart is selected by the user, the pie chart display unit 141a transmits information indicating the selected hardware to the power consumption call tree display unit 142. To do.

The bar graph display unit 141b displays a power consumption consumption ratio for each hardware in a certain time as a bar graph. More specifically, the bar graph display unit 141b divides the power consumption P _resource of each hardware calculated by the resource power consumption calculation unit 12 by the power consumption P _all of the terminal 2 as a whole. Calculate the consumption rate. And the bar graph display part 141b displays the consumption ratio of the power consumption of each hardware with a bar graph. FIG. 12 is a diagram illustrating an example of a bar graph displayed by the bar graph display unit 141b. The vertical axis of FIG. 12 shows the power consumption ratio of each hardware when the power consumption P _all of the terminal 2 as a whole is 1. The horizontal axis in FIG. 12 is identification information for identifying the start time of the period to be calculated.

  When a portion indicating a predetermined hardware consumption rate on the displayed bar graph is selected by the user, the bar graph display unit 141 b transmits information indicating the selected hardware to the power consumption call tree display unit 142.

  The power consumption call tree display unit 142 functions as power consumption call tree display means for creating a call tree based on the trace log and displaying the power consumption for each method superimposed on the call tree. The power consumption call tree display unit 142 includes a call tree creation unit 142a and a power consumption superimposition unit 142b.

Based on the trace log output from the method power consumption calculation unit 13, the call tree creation unit 142a analyzes the call relationship for each method in a unit time and creates a call tree. This call tree displays the methods used by the executed application program regardless of the time axis. The power consumption superimposing unit 142b superimposes the power consumption P _{method_X} for each method calculated by the method power consumption calculating unit 13 based on the method name of the call tree created by the call tree creating unit 142a. Then, the power consumption call tree display unit 142 displays a power consumption call tree in which the power consumption P _{method_X} for each method is superimposed on the call tree.

FIG. 13 is a diagram illustrating an example of the _power consumption call tree T _power displayed by the power consumption call tree display unit 142. As shown in FIG. 13, the power consumption call tree includes a call tree T _call and a power consumption table Pm. The call tree T _call indicates the relationship of methods called by the parent method in a tree structure. In the example of FIG. 13, the call tree T _call indicates that the application program AP calls the method a and the method b. The call tree T _call indicates that the method a calls the method aa and the method ab, and the method b calls the method ba and the method bb.

The power consumption table Pm shows the power consumption of each method. In the example of FIG. 13, the power consumption table Pm indicates the power consumption of the method aa, method ab, method ba, and method bb. Specifically, in the example of FIG. 13, power consumption table Pm is displaying the power consumption _{P Method_X} of each method, the power consumption _{P Method_i_X} per hardware for each method, the time a CPU, a.

  In addition, when the power consumption call tree display unit 142 receives information about the hardware selected by the user from the graph display unit 141, the power consumption call tree display unit 142 displays a power consumption call tree related to a method using the hardware. At this time, the power consumption call tree display unit 142 may display the power consumption call tree together with the graph displayed by the graph display unit 141. By displaying the call tree and power consumption of the method using the selected hardware in a superimposed manner, the power consumption shown in the graph can be displayed by subdividing it into power consumption for each method.

  For example, as shown in FIGS. 11 and 12, when the power consumption for each hardware for an arbitrary time is displayed on the graph display unit 141, it can be seen that the ratio of the consumption output of Wi-Fi is large except for the CPU. When a portion indicating Wi-Fi in the graph of FIG. 11 or 12 is selected, the power consumption call tree display unit 142 receives information indicating that Wi-Fi has been selected from the graph display unit 141. And the power consumption call tree display part 142 displays the power consumption call tree which superimposed the power consumption regarding Wi-Fi, and a call tree. Thereby, in Wi-Fi, a method that is a bottleneck with respect to power consumption is found.

  Next, the operation of the power consumption analysis system 1 will be described with reference to FIG. FIG. 14 is a flowchart showing processing of the power consumption analysis system 1.

  The resource consumption ratio calculation unit 11 acquires a resource log from the terminal 2 (S11). Further, the resource consumption ratio calculation unit 11 acquires a trace log from the terminal 2 (S12). Then, the resource consumption ratio calculation unit 11 uses the acquired resource log and trace log, and the above formula (1), the above formula (3), the above formula (4), the above formula (5), or the above formula The resource consumption ratio for each method is calculated by (6) (S13). Next, the resource power consumption calculation unit 12 uses the resource log acquired from the terminal 2 to calculate the power consumption of the entire terminal 2 (power consumption for each hardware of the terminal HW22) according to the above formula (7) and the above formula (8). Are calculated) (S14).

  And the method power consumption calculation part 13 calculates the power consumption for every method based on the resource consumption ratio for every method and the power consumption of the whole terminal 2 by said Formula (9) (S15). Next, the call tree creation unit 142a of the display unit 14 analyzes the parent-child relationship of the methods based on the trace log acquired from the terminal 2 and creates a call tree (S16). Subsequently, the power consumption superimposing unit 142b creates a power consumption call tree by superimposing power consumption for each method on the call tree. Then, the power consumption call tree display unit 142 displays the power consumption call tree (S17).

  Next, the effect of the power consumption analysis system 1 of this embodiment is demonstrated. In the power consumption analysis system 1, the resource consumption ratio calculation unit 11 includes a trace log that includes an execution history of each method called by the application program, and a resource log that records the resource consumption for each terminal HW 22 when the method is executed. And the resource consumption rate for each terminal HW22 is calculated for each method based on the trace log and the resource log. Further, the resource power consumption calculation unit 12 calculates the power consumption for each terminal HW22 based on the resource log. And the method power consumption calculation part 13 calculates the power consumption for every method based on the power consumption for every terminal HW22 and the resource consumption ratio with respect to each terminal HW22 for every method.

  Both the first resource log and the trace log are terminal-independent information that can be acquired from the OS level of the terminal. The second resource log is information that can be acquired from the API or the application program and does not depend on the terminal. Therefore, it is possible to calculate the power consumption in units of methods for various hardware and application programs.

  In the power consumption analysis system 1, the resource log is a first resource log in which the resource consumption amount in the entire terminal 2 is recorded. Depending on the terminal 2, there may be a case where the second resource log in which the resource consumption amount consumed by the method is recorded cannot be generated. In such a case, by using the first resource log, it is possible to acquire the resource consumption for the entire terminal 2 and the terminal HW22.

  In the power consumption analysis system 1, the resource consumption rate calculation unit 11 calculates the ratio of the method execution count calculated based on the trace log or the total CPU time of the method calculated based on the trace log. To calculate the resource consumption rate for each method. The first resource log cannot be acquired with real-time characteristics, and only the resource consumption per fixed time can be acquired. Therefore, assuming that the method execution count and the resource consumption recorded in the first resource log are in a proportional relationship, the first resource log (resource log of the entire terminal 2) and the trace log (method execution count) Resource consumption for each method can be estimated.

  Similarly, assuming that the CPU time of the method and the resource consumption recorded in the first resource log are in a proportional relationship, the resource consumption for each method can be estimated from the first resource log and the trace log. For this reason, information equivalent to the second resource log can be estimated, and the resource consumption ratio for each method can be obtained. This makes it possible to estimate the resource consumption of all methods during the same period and the resource consumption for each method call.

  In the power consumption analysis system 1, the resource log is a second resource log in which the resource consumption for each hardware for each method is recorded. In this way, by using the second resource log, it is possible to acquire the resource consumption for each piece of hardware for each method.

  Further, in the power consumption analysis system 1, the resource consumption rate calculation unit 11 corrects the resource consumption for each hardware recorded for each method recorded in the second resource log based on a coefficient corresponding to the hardware. The second resource log may not be able to acquire all the hardware resource consumption used by the application program. In such a case, multiply the resource consumption for each hardware stored in the second resource log for each hardware by a coefficient corresponding to the hardware, or add the resource for each hardware for each method. The consumption can be corrected. As a result, it is possible to calculate the power consumption for each method with higher accuracy.

  In the power consumption analysis system 1, the display unit 14 displays the power consumption for each method. Thereby, the power consumption for each method can be visualized, and the developer and the end user can easily grasp the power consumption.

  In the power consumption analysis system 1, the power consumption call tree display unit 142 of the display unit 14 creates a call tree based on the trace log and displays the power consumption for each method superimposed on the call tree. By displaying the power consumption in the form of a call tree, it is possible to display the power consumption together with the parent-child relationship of the method. For this reason, it becomes possible to clarify a method that becomes a bottleneck in terms of power consumption.

  In the power consumption analysis system 1, the power consumption call tree display unit 142 displays the power consumption of the method in the specified period. Thereby, the time zone of evaluation object can be specified and the analysis according to the time zone can be performed.

  In the power consumption analysis system 1, the graph display unit 141 of the display unit 14 displays the power consumption for each hardware in a predetermined time in a predetermined graph. By displaying the power consumption for each hardware in a graph, it is possible to clarify the hardware that is the bottleneck in terms of power consumption.

  In the power consumption analysis system 1, the power consumption call tree display unit 142 of the display unit 14 creates a call tree based on the trace log and displays the power consumption for each method superimposed on the call tree. Then, the power consumption call tree display unit 142 displays a call tree in accordance with the selected part in the graph displayed by the graph display unit 141. By displaying the call tree of the selected part, the bottleneck method can be further clarified.

  The graph display unit 141 displays, for example, a bar graph or a pie graph. By displaying the power consumption for each hardware in a predetermined time in a bar graph format or a pie chart format, it is possible to clarify the hardware that is a bottleneck with respect to power consumption.

  In the power consumption analysis system 1, the terminal 2 is a wireless communication terminal. It is not easy for an end user who uses an application program on a wireless communication terminal to recognize how much power is consumed by execution of the application program or data transmission / reception. It is difficult to suitably use the generated power. However, by using the resource log or trace log of the wireless communication terminal, it is possible to calculate the power consumption for each method of the application program executed on the actual wireless communication terminal. In this way, it is possible to provide an evaluation result of power consumption when using the wireless communication terminal.

  The power consumption analysis system according to the present invention is not limited to the one described in this embodiment. The power consumption analysis system according to the present invention may be modified or applied to other power consumption analysis systems according to the embodiments without changing the gist described in each claim. In the above embodiment, “power consumption (W)” is described, but “power consumption (Wh)” may be used.

  For example, in the above embodiment, the terminal 2 is a wireless communication terminal, but is not limited to this. The terminal 2 can also be a virtual terminal, for example. In this case, a virtual terminal having the virtual terminal HW22 may be reproduced by software on the information processing terminal on which the power consumption analysis system 1 operates. Then, an application program to be evaluated may be executed on the virtual terminal, and its resource log or trace log may be used. For this reason, it is possible to calculate the power consumption of the application program without preparing an actual wireless communication terminal. In this way, it is possible to provide an evaluation result of power consumption when developing an application program installed in a wireless communication terminal.

  In the above embodiment, the second resource log may be acquired from the low level API of the virtual terminal. By processing so that the second resource log can be output in advance to the low-level API used by the virtual terminal, the resource consumption amount used by each method can be accurately calculated. As a result, it is possible to improve the accuracy of power consumption calculated for each method.

In the above embodiment, the resource consumption ratio calculation unit 11 calculates the corrected resource consumption C _{method_i} ′ based on the above equation (2), but is not limited thereto. For example, resource consumption ratio calculating unit 11, the resource consumption _{C method} acquired from the second resource log is divided by the factor beta _resouce, or, such as by subtracting a constant gamma _resouce from resource consumption _{C method,} the correction resources The consumption C _{method_i} ′ may be calculated.

  The graph display unit 141 may display the power consumption for each hardware over the entire operation time of the application program. Further, the graph display unit 141 may display the power consumption for each hardware for the period specified by the user or the like.

  Further, the power consumption call tree display unit 142 may display the methods used by the application program executed within the time specified by the developer, the end user, etc. in the form of a call tree. By doing in this way, the analysis according to a time slot | zone is attained.

  In the above-described embodiment, the power consumption call tree display unit 142 displays only the power consumption of the method at the end, but may display the power consumption of the parent method. The power consumption call tree display unit 142 may sort and display the methods in the order of power consumption of predetermined hardware in the power consumption call tree. By rearranging in ascending order of power consumption, it is possible to find a method that becomes a bottleneck as a total during execution of the application program.

  In addition, an application development tool (application development apparatus) can be mounted using the power consumption analysis system 1. FIG. 15 is a block diagram showing a functional configuration of the application development tool 100. As shown in FIG. 15, the application development tool 100 includes a power consumption analysis system 1, an input unit 110 (corresponding to “input means” in the claims), and a source code editing unit 120 (in the claims “ Source code editing means ”) and a source code execution unit 130. The input unit 110 functions as an input unit that receives input from a user (for example, a developer). The source code editing unit 120 functions as source code editing means for editing the source code of the application program based on the input received by the input unit 110. The source code execution unit 130 functions as a source code execution unit that executes the source code edited by the source code editing unit 120 based on the input received by the input unit 110. The power consumption analysis system 1 analyzes the power consumption for each method with respect to the source code of the application program executed by the source code execution unit 130, and displays the result on the display unit 14.

  According to this application development tool, a developer can develop an application program while confirming power consumption in a wireless terminal and a virtual wireless terminal. And it becomes possible to implement | achieve power saving by reducing the method and hardware used as a bottleneck.

  In the above embodiment, the case where the application program uses a method as a module has been described. However, instead of the method, another module such as a subroutine, a function, or a class may be used.

  In addition, the above embodiment may be realized by a program module for executing each function in the power consumption analysis system 1. In other words, a resource consumption ratio calculation module corresponding to the resource consumption ratio calculation unit 11, a resource consumption power calculation module corresponding to the resource power consumption calculation unit 12, and a method power consumption calculation module corresponding to the method power consumption calculation unit 13 It is a power analysis program, and a function equivalent to the power consumption analysis system 1 described above can be realized by causing the computer system such as a server device to read the program. The above-described power consumption analysis program is provided by being stored in a storage medium such as a flexible disk, a CD-ROM, a DVD, or a ROM, or a semiconductor memory, for example. The above-described power consumption analysis program may be provided via a network as a computer data signal superimposed on a carrier wave.

DESCRIPTION OF SYMBOLS 1 ... Power consumption analysis system, 2 ... Terminal, 11 ... Resource consumption ratio calculation part (Resource consumption ratio calculation means), 12 ... Resource power consumption calculation part (Resource power consumption calculation means), 13 ... Method power consumption calculation part (Module) (Power consumption calculation means), 14 ... display section (display means), 141 ... graph display section (graph display means), 142 ... power consumption call tree display section (power consumption call tree display means), 22 ... terminal HW (hardware) ), 100... Application development tool, 110... Input unit (input means), 120... Source code editing section (source code editing means), 130... Source code execution section (source code execution means)

Claims (16)

A power consumption analysis system for analyzing power consumption of an application program executed on a terminal,
The trace log including the execution history of each module called by the application program, and the resource log in which the resource consumption for each hardware at the time of execution of the module is recorded, and the trace log and the resource log Based on the above, resource consumption rate calculation means for calculating a resource consumption rate for each hardware for each module,
Resource power consumption calculating means for calculating power consumption for each hardware based on the resource log;
Module power consumption calculating means for calculating the power consumption for each module based on the power consumption for each hardware and the resource consumption ratio for each hardware for each module;
A power consumption analysis system comprising:   The power consumption analysis system according to claim 1, wherein the resource log is a first resource log in which resource consumption of the entire terminal is recorded.   The resource consumption ratio calculation means uses the number of executions of the module calculated based on the trace log or the ratio of the total CPU time of the module calculated based on the trace log for each module. The power consumption analysis system according to claim 2, wherein a resource consumption ratio is calculated.   The power consumption analysis system according to claim 1, wherein the resource log is a second resource log in which a resource consumption amount for each piece of hardware for each module is recorded.   5. The resource consumption ratio calculation unit corrects the resource consumption amount for each piece of hardware recorded in the second resource log based on a coefficient corresponding to the hardware. The power consumption analysis system described.   The power consumption analysis system according to any one of claims 1 to 5, further comprising display means for displaying power consumption for each module.   7. The power consumption call tree display means for creating a call tree based on the trace log and displaying the power consumption of each module superimposed on the call tree. The power consumption analysis system described.   The power consumption analysis system according to claim 7, wherein the power consumption call tree display means displays the power consumption of the module in a specified period.   The power consumption analysis system according to claim 6, wherein the display unit includes a graph display unit that displays power consumption for each piece of hardware in a predetermined time in a predetermined graph. The display means further comprises a power consumption call tree display means for creating a call tree based on the trace log and displaying the power consumption for each module superimposed on the call tree.
The power consumption analysis system according to claim 9, wherein the power consumption call tree display unit displays the call tree according to a portion selected in the graph displayed by the graph display unit.   The power consumption analysis system according to claim 9 or 10, wherein the graph display means displays a bar graph or a pie chart.   The power consumption analysis system according to any one of claims 1 to 11, wherein the terminal is a wireless communication terminal.   The power consumption analysis system according to any one of claims 1 to 11, wherein the terminal is a virtual terminal.   The power consumption analysis system according to claim 13, wherein the second resource log is acquired from a low-level API of the virtual terminal.   The power consumption analysis system according to claim 1, wherein the module is a method. Input means for accepting user input;
Source code editing means for editing the source code of the application program based on the input received by the input means;
Source code execution means for executing the source code edited by the source code editing means based on the input received by the input means;
The power consumption analysis system according to any one of claims 1 to 15, wherein the power consumption of the source code executed by the source code execution unit is analyzed.
An application development tool characterized by comprising:

Images