CN108919938A - A kind of power consumption of processing unit optimization method suitable for Android game - Google Patents
A kind of power consumption of processing unit optimization method suitable for Android game Download PDFInfo
- Publication number
- CN108919938A CN108919938A CN201810696874.3A CN201810696874A CN108919938A CN 108919938 A CN108919938 A CN 108919938A CN 201810696874 A CN201810696874 A CN 201810696874A CN 108919938 A CN108919938 A CN 108919938A
- Authority
- CN
- China
- Prior art keywords
- gpu
- cpu
- game
- android
- voltage frequency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3234—Power saving characterised by the action undertaken
- G06F1/3287—Power saving characterised by the action undertaken by switching off individual functional units in the computer system
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Computing Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Power Sources (AREA)
Abstract
The invention discloses a kind of power consumption of processing unit optimization methods suitable for Android game, construct the power managed frame for Android platform game application, including game features data collection module and CPU-GPU collaboration electric voltage frequency regulate and control module;Performance data collection module completes the collection of user interactive data and GPU rendering data when Android game running, provides support for CPU the and GPU load evaluation based on Android game features, improves timeliness and accuracy;It cooperates with electric voltage frequency regulation module to change using hierarchical states machine comprehensive description game running state change and processor operating status, regulates and controls logic for electric voltage frequency and indicate opportunity and direction;Regulating and controlling logic based on the producer between CPU-GPU in Android game --- the design of consumer's cooperation relation realizes the collaboration power managed between CPU-GPU.The efficiency of CPU-GPU is improved under the premise of guaranteeing user experience.
Description
Technical field
The present invention relates to a kind of power consumption of processing unit optimization methods, more particularly to a kind of place suitable for Android game
Manage device power consumption optimization method.
Background technique
With popularizing for the smart phone with good human-computer interaction characteristic, moving game is increasingly becoming to be liked by user
A kind of application, wherein complexity 3D game is particularly subject to favor.This has benefited from smart phone CPU (Central Processing
) and the raising of GPU (Graphics Processing Unit) computing capability Unit.High-performance often brings high energy consumption, for
Battery powered intelligent mobile phone platform, the power consumption that game application makes Smartphone device intrinsic the high request of computing resource with
Contradiction between battery capacity highlights.The study found that existing when current Android intelligent platform game application operation
The phenomenon that computing resource and energy consumption waste.
Dynamic voltage frequency adjusts (Dynamic Voltage and Frequency Scaling, abbreviation DVFS), is to work as
Before be generally used carry out power managed a kind of effective mechanism.Its core concept is according to processor load change information dynamic
The voltage and frequency for adjusting processor operation are to achieve the purpose that energy saving, so load evaluation and electric voltage frequency regulate and control algorithm
It is the key content in its implementation strategy.In conventional measures, Android platform generallys use the Interactive plan for CPU
Slightly and therewith similar corresponding GPU DVFS is tactful.
Traditional DVFS strategy be suitable for Android game application the problem is that:(1) have when Android game running
There are significant user interaction characteristics and CPU-GPU cooperative nature, conventional frame is difficult to be utilized these characteristics and carries out optimised power consumption.
(2) the CPU/GPU load evaluation method based on hardware utilization feedback is unable to satisfy the timeliness of Android game load evaluation
And accuracy requirement, cause user experience reduction or energy consumption waste problem.(3) completely self-contained CPU/GPU electric voltage frequency regulation
Cooperation relation when logic can not utilize Android game running between CPU-GPU is to further increase system energy efficiency.
Summary of the invention
For the above technical problems, object of the present invention is to:It provides a kind of suitable for Android game application
Power consumption of processing unit management method, the present invention is based on android system characteristics and game features building to be suitable for Android game
Power managed frame collects performance data when game running, provides support for CPU/GPU load evaluation promptly and accurately;?
Bottom connection CPU and GPU DVFS module provides unified electric voltage frequency regulation logic, the cooperation relation pair based on CPU-GPU
The electric voltage frequency of CPU and GPU carries out Collaborative Control, under the premise of guaranteeing user experience, when further increasing game running
The efficiency of android system.
The technical scheme is that:
A kind of power consumption of processing unit optimization method suitable for Android game, which is characterized in that be based on android system
Frameworks layers of Application and Linux kernel layer building Android game power managed frame, including trip
Characterisitic parameter tracking (Parameters Tracer) module of playing and CPU-GPU cooperate with DVFS module;
The game features parameter tracking module, customer interaction information and GPU when acquiring game running render game frame
Data;
The CPU-GPU cooperates with DVFS module, the cpu load assessment comprising user's perception, based on game frame complexity
GPU load evaluation is cooperateed with using the parameter description of hierarchical states machine and the CPU-GPU based on CPU-GPU cooperation relation in game
Electric voltage frequency regulates and controls part, carries out the CPU-GPU collaboration power managed based on game features;
It is described in game features parameter tracking module in preferred technical solution, in android system
The identification and acquisition of user's interaction message are carried out in Frameworks layers of Application of main thread Looper object, specifically
Ground utilizes the frame refreshing message Android.view.Choreographer $ of Message Processing class Choreographer object handles
FrameHandler carries out the judgement and data acquisition of user mutual behavior, provides data for accurate evaluation cpu load and supports.
It is described in game features parameter tracking module in preferred technical solution, in android system
The acquisition of GPU frame rendering data is carried out in the EGL that Frameworks layers of Application, the data of acquisition include three classes, point
Be not game frame picture constitute in vertex data, data texturing and control command number, they all pass through OpenGL | ES API and
Parameter analysis of electrochemical is acquired;Specifically, the trace mechanism based on android system, respectively with glDrawElements () letter
Number, glDrawArrays () function obtain vertex data, with glTexImage2D () function, glBindTexture () function
Texture dimensions data are obtained, total life needed for drawing present frame every time is obtained by the TraceGL () function in Trace mechanism
Number is enabled, data is provided for accurate evaluation GPU load and supports.
In preferred technical solution, the cpu load assessment of user's perception is calculated by formula 1 and is completed, specifically will
User's interaction is independent when Android game running assesses, and it is obtained user with cpu busy percentage Weighted Fusion and is handed over
The cpu load mutually perceived, the timeliness and accuracy of cpu load assessment when improving Android game running.
Formula 1For calculating Android game
Cpu load when operation is related to through weight coefficient wBAnd wUCpu busy percentage and user's interaction intensity weighted are obtained respectively,
Delta_time and delta_idle is respectively the CPU IDLE time in sampling period and sampling period, and ∑ msg_int is to know
Not counted with user's interaction message msg_int of acquisition, M is normalization coefficient --- user once interacts maximum message quantity,
wB,wU∈ [0,1], ∑ msg_int ∈ [0, M], wB,wUIt is determined with M by experiment.
In preferred technical solution, the GPU load evaluation based on game frame complexity is calculated by formula 2 and is completed,
Specifically calculating demand of game frame when Android game running to GPU is characterized with game frame complexity defined in method, together
When to assess GPU load, utilize producer when game running in CPU-GPU data flow --- Consumer relationships, " on
Assessment GPU load in advance, the timeliness and accuracy of GPU load evaluation when greatly improving Android game running at trip " CPU.
2 FC of formulatot(v,t,c)=∑ Wj*FCj(xj), (j=v, t, c), game when for calculating Android game running
Frame complexity FCtot(v,t,c), while the assessed value as GPU load, it is related to through weight coefficient wj(j=v, t, c) is by frame structure
At three classes data complexity FCj(xj) be weighted amalgamation mode and obtain, wherein xjIndicate three classes data statistics value, specific weight
The calculation method of value and three classes component complexity can be calculated with limitation according to actual needs.
It is described to carry out user experience variation and CPU/GPU load variation using hierarchical states machine in preferred technical solution
Description, the user experience characterized with frame rate of game variation is abstracted as four super states, is user experience respectively from meeting
Foot is unsatisfactory for, from be unsatisfactory for meeting and from be unsatisfactory for being unsatisfactory for from meeting, and will cause the CPU/ of these super state reasons
GPU load variation is abstracted as corresponding sub- state, has CPU and GPU load to be all larger than upper limit threshold, respectively less than lower threshold etc.,
Under specific condition, if CPU and GPU load is not more than or less than upper limit threshold and lower threshold, at this time by the way which judges
Variation finally obtains the regulation of processor voltage frequency by the hierarchical classification of similar fashion more greatly come the next state for taking out sub- state
Opportunity and regulation direction.
In preferred technical solution, the GPU collaboration voltage frequency based on CPU-GPU cooperation relation in Android game
Rate regulation logic follows following four principles:
(1) CPU and GPU working voltage frequency collaborative variation;
(2) fully consider that other side's operating status carries out electric voltage frequency regulation;
(3) different goals of regulation and control is set under different operating statuses;
(4) the compatible independent DVFS logic of CPU/GPU.
In preferred technical solution, CPU-GPU collaboration DVFS logic step is described as follows:
(a) last user experience information, CPU and GPU operating status and load information are obtained;
(b) super state is judged according to active user's experience information;And carry out the CPU-GPU electricity of different target respectively according to super state
Voltage-frequency rate coordinated regulation step (c) or (d) or (e) or (f);
If (c) super state is user experience from meeting satisfaction, goal of regulation and control is to attempt frequency reducing to reduce power consumption, and judgement is at this time
Whether CPU and GPU load is less than the lower threshold of electric voltage frequency regulation, if cpu load is less than lower threshold, the two collaboration drop
Frequently, if only GPU is less than lower threshold, only GPU frequency reducing;
If (d) to be user experience be unsatisfactory for from meeting super state, goal of regulation and control is to attempt raising frequency to guarantee user experience, first
First judge whether the trial frequency reducing logic as described in (c) causes super state, if it is, restoring last time CPU and GPU electric voltage frequency
Value, if it is not, whether judgement CPU and GPU load at this time is greater than the upper limit threshold of electric voltage frequency regulation, if cpu load is big
In upper limit threshold, the two cooperates with raising frequency, if only GPU is greater than upper limit threshold, only GPU raising frequency;If the two load value is not
Greater than upper limit threshold, then determining CPU or GPU by judgement variation size, which is to cause the Main change factor of super state, and right
It carries out raising frequency operation;
If (e) super state is user experience from being unsatisfactory for satisfaction, goal of regulation and control is that user experience is kept to meet state,
At this time without electric voltage frequency control manipulation;
If (f) super state is user experience from be unsatisfactory for being unsatisfactory for, goal of regulation and control is to attempt gradually frequency or directly rise to
Most high frequency first determines whether this super state has accumulated repeatedly to guarantee user experience, if it is, directly by CPU and GPU electricity
Pressure frequency values rise to system peak, if it is not, then stepping up the electric voltage frequency value of CPU and GPU;
(g) user experience change information, CPU and the information such as GPU load change information and the variation of super state are updated;
(h) interface provided using driving by coordinated regulation as a result, cpu frequency and GPU frequency exported respectively to CPU and
Work is arranged by the electric voltage frequency that processor is completed in driving in the drive module of GPU.
Compared with prior art, it is an advantage of the invention that:
It 1, can suitable for the power consumption of processing unit management method of Android game completely in Android operation system layer building
To close source game application towards commercialization.
2, compared to conventional processors load evaluation method, the cpu load appraisal procedure of user's interactive perception and based on game
The GPU load evaluation method of frame complexity reduces delay, improves accuracy;
3, the electric voltage frequency regulation logic synthesis consideration user experience and CPU-GPU operating status of CPU-GPU collaboration, and with
Cooperation relation when Android game running between CPU-GPU carries out the design of electric voltage frequency regulation logic, relatively current system
Default scheme reaches average 12% under the premise of guaranteeing user experience, and the efficiency of highest 34% is promoted.
4, it in realization, is sufficiently multiplexed suitable for the power consumption of processing unit management method of Android game traditional mutually indepedent
CPU/GPU power managed module, make full use of upper layer Android game features and android system characteristic to carry out power consumption pipe
Reason has fully ensured that the practicality for user configuration selection.
Detailed description of the invention
The invention will be further described with reference to the accompanying drawings and embodiments:
Fig. 1 is the structural framing schematic diagram of the power consumption of processing unit management method suitable for Android game;
Fig. 2 is that user experience variation and CPU-GPU variation schematic diagram are described using hierarchical states machine;
Fig. 3 is the collaboration DVFS algorithm core logic figure based on CPU-GPU cooperation relation in Android game;
Fig. 4 is that the efficiency the present invention is based on ODROID-XU3 experiment porch with respect to default scheme promotes effect picture;
Fig. 5 is the present invention and the user experiences of other research achievements (be averaged frame per second characterization with game) and average frame energy consumption cross
To comparison diagram.
Specific embodiment
Above scheme is described further below in conjunction with specific embodiment.It should be understood that these embodiments are for illustrating
The present invention and be not limited to limit the scope of the invention.Implementation condition used in the examples can be done according to the condition of specific producer
Further adjustment, the implementation condition being not specified is usually the condition in routine experiment.
It is as shown in Figure 1 the structural framing schematic diagram of the power consumption of processing unit management method suitable for Android game, mainly
(Parameters Tracer) module and CPU-GPU collaboration electric voltage frequency regulation are acquired including game features data
(Governor) module;The game features data acquisition module completes the acquisition of two parts game features data, is game respectively
Game frame picture needed for user's interaction generates when operation message count and GPU rendering frame picture constitutes data, includes vertex
Number, texture dimensions and control command number.CPU-GPU collaboration electric voltage frequency regulation module include cpu load assessment calculate,
GPU load evaluation is calculated, is described using the Parameters variation of hierarchical states machine (Hierarchy State Machine, abbreviation HSM)
DVFS algorithm is cooperateed with the CPU-GPU based on CPU-GPU cooperation relation in game;Wherein cpu load assessment calculation method is public affairs
The cpu load of user's interactive perception described in formula 1 is assessed;GPU load evaluation calculation method is described in formula 2 based on game frame
The GPU load evaluation of complexity;Parameters variation description carries out comprehensive description to game running state and processor load with determination
The variation of the user experience of frame per second characterization is specifically abstracted as 4 and surpassed by the opportunity and direction that CPU and GPU electric voltage frequency is adjusted
The CPU and GPU that cause these super states load variation are abstracted as sub- state and second son state with clear opportunity and direction by state;The base
The collaboration DVFS algorithm of the producer in Android game between CPU-GPU --- consumer's cooperation relation breaks traditional phase
Mutual independent power consumption of processing unit management algorithm accomplishes the electric voltage frequency regulation of CPU-GPU collaboration, further increases system energy efficiency.
Wherein, game features parameter tracking cooperates with DVFS algorithm to provide the data branch of CPU and GPU load evaluation for bottom
Support, Application Frameworks layers of application management and Linux in making full use of Android operation system
Kernel layers of resource management function is completed.Specifically, the user interaction characteristics of Android game and game frame picture characteristics.In number
According to tracking it is as follows:
1. carrying out user in Frameworks layers of Application of android system of main thread Looper object
The identification and acquisition of interaction message specifically utilize the frame refreshing message of Message Processing class Choreographer object handles
Android.view.Choreographer FrameHandler carries out the judgement and data acquisition of user mutual behavior, subject to
Really assessment cpu load provides data and supports.
2. carrying out adopting for GPU frame rendering data in Frameworks layers of Application of android system of EGL
Collection, the data of acquisition include three classes, be respectively game frame picture constitute in vertex data, data texturing and control command number,
They all pass through OpenGL | and ES API and Parameter analysis of electrochemical are acquired;Specifically, the trace mechanism based on android system,
Vertex data is obtained with glDrawElements () function, glDrawArrays () function respectively, with glTexImage2D ()
Function and glBindTexture () function obtain texture dimensions data, are obtained by the TraceGL () function in Trace mechanism
Total command number provides data for accurate evaluation GPU load and supports.
CPU and GPU load evaluation (CPU/GPU Workload Estimation) shown in figure is in the characteristic tracked
It in data basis, is calculated according to formula 1 and formula 2, parameter therein is obtained according to game features by testing.
Fig. 2, which is shown, changes schematic diagram using the variation of hierarchical states machine (HSM) description user experience and CPU-GPU, will be with
The user experience variation of frame rate of game characterization is abstracted as four super states, is user experience respectively from meeting satisfaction (QOS_MEET_
TO_MEET), from meet be unsatisfactory for (QOS_MEET_TO_LOSS), from being unsatisfactory for satisfaction (QOS_LOSS_TO_MEET) and
From be unsatisfactory for being unsatisfactory for (QOS_LOSS_TO_LOSS), the CPU/GPU for causing these super state reasons load variation is abstracted as
Corresponding sub- state has CPU and GPU load to be all larger than upper limit threshold, be respectively less than lower threshold etc., under specific circumstances, such as CPU and
GPU load is not more than or less than upper limit threshold and lower threshold, at this time by judging which variation takes out son more greatly
The next state of state finally obtains opportunity and the regulation direction of the regulation of processor voltage frequency by the hierarchical classification of similar fashion.
Collaboration DVFS algorithm core logic figure based on CPU-GPU cooperation relation in Android game shown in Fig. 3, specifically
Process is described as follows:
(a) last user experience information, CPU and GPU operating status and load information are obtained;
(b) super state is judged according to active user's experience information;And carry out the CPU-GPU electricity of different target respectively according to super state
Voltage-frequency rate coordinated regulation step (c) or (d) or (e) or (f);
If (c) super state is user experience from meeting satisfaction, goal of regulation and control is to attempt frequency reducing to reduce power consumption, and judgement is at this time
Whether CPU and GPU load is less than the lower threshold of electric voltage frequency regulation, if cpu load is less than lower threshold, the two collaboration drop
Frequently, if only GPU is less than lower threshold, only GPU frequency reducing;
If (d) to be user experience be unsatisfactory for from meeting super state, goal of regulation and control is to attempt raising frequency to guarantee user experience, first
First judge whether the trial frequency reducing logic as described in (c) causes super state, if it is, restoring last time CPU and GPU electric voltage frequency
Value, if it is not, whether judgement CPU and GPU load at this time is greater than the upper limit threshold of electric voltage frequency regulation, if cpu load is big
In upper limit threshold, the two cooperates with raising frequency, if only GPU is greater than upper limit threshold, only GPU raising frequency;If the two load value is not
Greater than upper limit threshold, then determining CPU or GPU by judgement variation size, which is to cause the Main change factor of super state, and right
It carries out raising frequency operation;
If (e) super state is user experience from being unsatisfactory for satisfaction, goal of regulation and control is that user experience is kept to meet state,
At this time without electric voltage frequency control manipulation;
If (f) super state is user experience from be unsatisfactory for being unsatisfactory for, goal of regulation and control is to attempt gradually frequency or directly rise to
Most high frequency first determines whether this super state has accumulated repeatedly to guarantee user experience, if it is, directly by CPU and GPU electricity
Pressure frequency values rise to system peak, if it is not, then stepping up the electric voltage frequency value of CPU and GPU;
(g) user experience change information, CPU and the information such as GPU load change information and the variation of super state are updated;
(h) interface provided using driving by coordinated regulation as a result, cpu frequency and GPU frequency exported respectively to CPU and
Work is arranged by the electric voltage frequency that processor is completed in driving in the drive module of GPU.
Fig. 4 show suitable for the power consumption of processing unit optimization method (COOPFSM is abbreviated as figure) of Android game according to
Different classes of game application is covered to CPU and GPU demand height, with current Android platform default power managed scheme (in figure
Be abbreviated as ORG) comparative situation;The index compared is efficiency income (q).In the case where guaranteeing user experience, shown in Fig. 4
Data in as can be seen that COOPFSM have different degrees of efficiency income with respect to ORG, for high CPU and high GPU calculating demand
Game application (in figure number be 18,19,22,27,30 and No. 36 test samples) effect it is more preferable, counted especially for high GPU
The game application of calculation demand can achieve 7%~34% efficiency income.The game application low to CPU and GPU demand (is compiled in figure
Number for 8,9,23 and No. 29 application) almost without efficiency income.
Fig. 5 show the across comparison situation of the present invention with other achievements, chooses the correlative study for being published in ISLPED'16
Achievement carries out lateral comparison.ORG, PAT15, CO-Cap16 and HiCAP are the four kinds of schemes compared in original text, wherein ORG
It is android system default scheme, Default is expressed as in original text, HiCAP is the close based on FSM and CPU/GPU of author's proposition
Improvement result of collection property on the basis of Co-Cap16.PAT15 is achievement earlier.The mode compared be by the FPS data of ORG and
Average frame energy consumption (Energy per Frame, EpF) is used as benchmark (being normalized to 100), and other schemes are according to relative efficacy meter
Calculate corresponding numerical value.The correspondence numeric renderings of (being denoted as COOPFSM) of the invention are used as one in figure.It can be seen from the figure that institute
The power consumption of processing unit optimization method for being suitable for Android game is stated on power consumption income (EpF) by better than ORG, PAT15 and CO-
Cap16, slightly worse than HiCAP, but it is better than PAT15 and HiCAP on user experience (FPS), and opposite ORG and CO-Cap16 is only
There is small (1%) gap.From the point of view of across comparison, the power consumption of processing unit optimization method suitable for Android game reaches
Preferable effect.
In conclusion the power consumption of processing unit optimization method for being suitable for Android game is guaranteeing game with respect to default method
There is apparent performance boost under the premise of user experience.
The foregoing examples are merely illustrative of the technical concept and features of the invention, its object is to allow the person skilled in the art to be
It cans understand the content of the present invention and implement it accordingly, it is not intended to limit the scope of the present invention.It is all smart according to the present invention
The equivalent transformation or modification that refreshing essence is done, should be covered by the protection scope of the present invention.
Claims (10)
1. a kind of power consumption of processing unit optimization method suitable for Android game, which is characterized in that pass through android system structure
The power consumption of processing unit management framework for building characteristic when based on Android game running realizes that power consumption of processing unit management framework includes trip
Play performance data acquisition module and CPU-GPU collaboration electric voltage frequency regulate and control module;Data by system call from upper layer transfers to
Linux kernel;
The game features data acquisition module, user's interaction message and GPU when for acquiring Android game running render
Data needed for game frame are used for CPU and GPU load evaluation;
The CPU-GPU collaboration electric voltage frequency regulates and controls module, and the load evaluation for respectively including CPU and GPU calculates and CPU-
GPU cooperates with electric voltage frequency to regulate and control logic, for completing the CPU-GPU collaboration power managed using Android game features.
2. the power consumption of processing unit optimization method according to claim 1 suitable for Android game, which is characterized in that institute
Game features data acquisition module is stated, in the Application Frameworks module data acquisition of android system,
User's interaction generates the acquisition of message based on the MessageQueue principle and Looper object and message in android system
Class Choreographer is handled to complete;GPU rendering data acquires the EGL module in system based on the trace of android system
Mechanism is completed;Data are transferred to bottom by system calling.
3. the power consumption of processing unit optimization method according to claim 1 suitable for Android game, which is characterized in that institute
Show in CPU-GPU collaboration electric voltage frequency regulation module, the load that the load evaluation of CPU generates user's interaction is independent, makees
Cpu load is because the variable quantity that user's interaction generates is assessed and is quantified when for game running, and cpu load when game running
It is obtained by cpu busy percentage and user's interaction load Weighted Fusion.
4. the power consumption of processing unit optimization method according to claim 1 suitable for Android game, which is characterized in that institute
It states in CPU-GPU collaboration electric voltage frequency regulation module, data flow relation of the load evaluation of GPU according to GPU in game running,
It is assessed using vertex needed for the rendering game frame picture as claimed in claim 2 obtained in " upstream ", texture and order data
GPU load when game running, specific method are that three classes data normalization rear weight is merged the assessed value loaded as GPU.
5. the power consumption of processing unit optimization method according to claim 1 suitable for Android game, which is characterized in that institute
It states in CPU-GPU collaboration electric voltage frequency regulation module, CPU-GPU cooperates with electric voltage frequency regulation algorithm synthesis to consider game running shape
State and processor operating status are guiding with user experience, describe user experience variation using hierarchical states machine and CPU-GPU is negative
Variation is carried, determines opportunity and the direction of the regulation of CPU and GPU electric voltage frequency.
6. the power consumption of processing unit optimization method according to claim 1 suitable for Android game, which is characterized in that institute
It states in CPU-GPU collaboration electric voltage frequency regulation module, CPU-GPU cooperates with electric voltage frequency regulation algorithm to transport based on Android game
Producer when row between CPU-GPU --- consumer's cooperation relation design regulates and controls with only electric voltage frequency simultaneously is carried out to the two
Without considering that the method for cooperation relation between the two has the difference of essence.
7. the power consumption of processing unit optimization method according to claim 3 suitable for Android game, which is characterized in that institute
The load evaluation for stating CPU includes the following steps:
(1)It carries out the counting of user's interaction message and normalizes;
(2)Calculate cpu busy percentage;
(3)It determines the two weighting parameters, calculates the cpu load value that fusion obtains.
8. the power consumption of processing unit optimization method according to claim 4 suitable for Android game, which is characterized in that institute
The load evaluation for stating GPU includes the following steps:
(a)Count number of vertex, texture dimensions and the command number in game frame rendering;
(b)Determine that each component complexity factor calculates game frame complexity according to model;
(c)Utilize game frame complexity evaluations GPU load value.
9. the power consumption of processing unit optimization method according to claim 5 suitable for Android game, which is characterized in that institute
State and describe user experience variation and CPU-GPU load variation using hierarchical states machine, by user experience variation be abstracted as four it is super
State:From meeting, be unsatisfactory for, from be unsatisfactory for meeting and from be unsatisfactory for being unsatisfactory for from meeting;According to causing super state
CPU and GPU load situation of change define sub- state:CPU or GPU load is greater than upper limit threshold, is less than lower threshold and in two
Nine kinds of variations combination between person;When CPU/GPU load change direction is identical, determine who is by comparing the variable quantity of the two
Main change factor;Pass through the above final CPU- for determining user experience change direction and causing this variation of successively classification refinement
GPU loads situation of change, determines electric voltage frequency regulation opportunity and direction.
10. the power consumption of processing unit optimization method according to claim 6 suitable for Android game, which is characterized in that institute
The electric voltage frequency regulation logic for stating CPU-GPU collaboration, according to the production between the CPU-GPU in Android game running
Person --- consumer's cooperation relation design meets following principle:
(1)CPU and GPU working voltage frequency collaborative variation;
(2)Fully consider that other side's operating status carries out electric voltage frequency regulation;
(3)Different goals of regulation and control is set under different operating statuses;
(4)The compatible independent DVFS logic of CPU/GPU.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810696874.3A CN108919938A (en) | 2018-06-29 | 2018-06-29 | A kind of power consumption of processing unit optimization method suitable for Android game |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810696874.3A CN108919938A (en) | 2018-06-29 | 2018-06-29 | A kind of power consumption of processing unit optimization method suitable for Android game |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108919938A true CN108919938A (en) | 2018-11-30 |
Family
ID=64422085
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810696874.3A Pending CN108919938A (en) | 2018-06-29 | 2018-06-29 | A kind of power consumption of processing unit optimization method suitable for Android game |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108919938A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110308784A (en) * | 2019-04-30 | 2019-10-08 | 东莞恒创智能科技有限公司 | CPU, GPU based on Nvidia TX2 combine frequency modulation energy-saving optimization method |
CN111585319A (en) * | 2020-04-20 | 2020-08-25 | 维沃移动通信有限公司 | Camera power supply method and device, electronic equipment and storage medium |
CN112163985A (en) * | 2020-09-22 | 2021-01-01 | Oppo(重庆)智能科技有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
CN113138655A (en) * | 2021-04-02 | 2021-07-20 | Oppo广东移动通信有限公司 | Processor frequency adjusting method and device, electronic equipment and storage medium |
US11803221B2 (en) | 2020-03-23 | 2023-10-31 | Microsoft Technology Licensing, Llc | AI power regulation |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102830785A (en) * | 2011-06-17 | 2012-12-19 | 上海果壳电子有限公司 | Quick response system for equipment end input interaction and implementation method thereof |
US20130290751A1 (en) * | 2012-04-27 | 2013-10-31 | Qualcomm Innovation Center, Inc. | Frequency reduction of mobile device cores based on application processing requirements |
CN105045367A (en) * | 2015-01-16 | 2015-11-11 | 中国矿业大学 | Android system equipment power consumption optimization method based on game load prediction |
-
2018
- 2018-06-29 CN CN201810696874.3A patent/CN108919938A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102830785A (en) * | 2011-06-17 | 2012-12-19 | 上海果壳电子有限公司 | Quick response system for equipment end input interaction and implementation method thereof |
US20130290751A1 (en) * | 2012-04-27 | 2013-10-31 | Qualcomm Innovation Center, Inc. | Frequency reduction of mobile device cores based on application processing requirements |
CN105045367A (en) * | 2015-01-16 | 2015-11-11 | 中国矿业大学 | Android system equipment power consumption optimization method based on game load prediction |
Non-Patent Citations (1)
Title |
---|
宋家臣: ""CPU-GPU协同的Andorid平台游戏功耗管理研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110308784A (en) * | 2019-04-30 | 2019-10-08 | 东莞恒创智能科技有限公司 | CPU, GPU based on Nvidia TX2 combine frequency modulation energy-saving optimization method |
US11803221B2 (en) | 2020-03-23 | 2023-10-31 | Microsoft Technology Licensing, Llc | AI power regulation |
CN111585319A (en) * | 2020-04-20 | 2020-08-25 | 维沃移动通信有限公司 | Camera power supply method and device, electronic equipment and storage medium |
CN111585319B (en) * | 2020-04-20 | 2021-12-14 | 维沃移动通信有限公司 | Camera power supply method and device, electronic equipment and storage medium |
CN112163985A (en) * | 2020-09-22 | 2021-01-01 | Oppo(重庆)智能科技有限公司 | Image processing method, image processing device, storage medium and electronic equipment |
CN113138655A (en) * | 2021-04-02 | 2021-07-20 | Oppo广东移动通信有限公司 | Processor frequency adjusting method and device, electronic equipment and storage medium |
CN113138655B (en) * | 2021-04-02 | 2023-11-28 | Oppo广东移动通信有限公司 | Processor frequency adjusting method and device, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108919938A (en) | A kind of power consumption of processing unit optimization method suitable for Android game | |
Qi et al. | A QoS-aware virtual machine scheduling method for energy conservation in cloud-based cyber-physical systems | |
CN103475790B (en) | A kind of intelligent mobile terminal power consumption management | |
Zhou et al. | Virtual machine placement algorithm for both energy-awareness and SLA violation reduction in cloud data centers | |
CN102819313B (en) | Operating method of terminal equipment and terminal equipment | |
CN104049716B (en) | Computer energy-saving method and system combined with temperature sensing | |
CN103823718B (en) | Resource allocation method oriented to green cloud computing | |
CN105045367A (en) | Android system equipment power consumption optimization method based on game load prediction | |
CN104505864B (en) | Demand response control strategy analogue system and method for distributed power generation of dissolving | |
CN109214095A (en) | Electric car charge and discharge Multiobjective Optimal Operation method | |
CN108196935A (en) | A kind of energy saving moving method of virtual machine towards cloud computing | |
CN104821906B (en) | A kind of energy-efficient virtual network node mapping model and algorithm | |
CN111900719B (en) | Power grid adequacy evaluation method, device and system considering flexible controllable load | |
CN107545364A (en) | The user response policy optimization method of stimulable type electricity needs response | |
Li et al. | Deep reinforcement learning and game theory for computation offloading in dynamic edge computing markets | |
CN104408663A (en) | User demand response dispatching system and method of smart grid | |
CN113852135A (en) | Virtual power plant energy scheduling method, device, storage medium and platform | |
CN114997665A (en) | Virtual power plant optimal scheduling method and system considering controllable load response performance difference | |
CN103369640B (en) | Base station electricity saving method and device | |
CN106209433A (en) | Towards the application system energy-efficient deployment device of framework under a kind of cloud environment | |
Pang et al. | Energy-saving virtual machine placement method for user experience in cloud environment | |
CN108921368A (en) | Balanced cooperative game controller based on virtual power plant | |
CN105808355A (en) | Binary linear regression equation-based dynamic frequency modulation method | |
CN111914000B (en) | Server power capping method and system based on power consumption prediction model | |
CN112527409A (en) | Mobile medical health management-oriented edge cloud computing unloading method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181130 |
|
RJ01 | Rejection of invention patent application after publication |