WO2017016357A1 - 运算资源的散热控制方法、运算控制系统和存储介质 - Google Patents

运算资源的散热控制方法、运算控制系统和存储介质 Download PDF

Info

Publication number
WO2017016357A1
WO2017016357A1 PCT/CN2016/087145 CN2016087145W WO2017016357A1 WO 2017016357 A1 WO2017016357 A1 WO 2017016357A1 CN 2016087145 W CN2016087145 W CN 2016087145W WO 2017016357 A1 WO2017016357 A1 WO 2017016357A1
Authority
WO
WIPO (PCT)
Prior art keywords
computing resource
computing
resource
open state
closed
Prior art date
Application number
PCT/CN2016/087145
Other languages
English (en)
French (fr)
Inventor
张文彦
Original Assignee
深圳市万普拉斯科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 深圳市万普拉斯科技有限公司 filed Critical 深圳市万普拉斯科技有限公司
Priority to US15/747,508 priority Critical patent/US10488900B2/en
Priority to EP16829723.2A priority patent/EP3330852A4/en
Publication of WO2017016357A1 publication Critical patent/WO2017016357A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5094—Allocation of resources, e.g. of the central processing unit [CPU] where the allocation takes into account power or heat criteria
    • 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
    • G06F1/16—Constructional details or arrangements
    • G06F1/20—Cooling means
    • G06F1/206—Cooling means comprising thermal management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the present invention relates to the field of mobile terminal technologies, and in particular, to a heat dissipation control method, an operation control system, and a storage medium for computing resources.
  • processing chips are placing more and more computing resources on smaller and smaller areas. Because the high-density design will cause the computing resources to start together with other computing resources, the interaction will not only make the heat difficult to dissipate, but also cause the overall temperature to rise. The thermal interaction of various computing resources will be more serious, and the heat will not be scattered. Out.
  • the central processing unit of the existing processor for the intelligent mobile terminal includes a plurality of cores.
  • the typical practice is to reduce the operating frequency of the core, and when the core operating frequency decreases, the impact will be affected. Processor performance.
  • embodiments of the present invention provide a heat dissipation control method, an operation control system, and a storage medium for computing resources, which can effectively reduce multi-processors.
  • the structured operation controls the temperature of the system and maintains the efficient operation of the multi-processor architecture of the operational control system.
  • the embodiment of the invention provides a heat dissipation control method for a computing resource, and the method includes:
  • the computing resource to be opened is determined according to the distance between each computing resource in the closed computing resource and the computing resource in the open state. , the computing resource to be turned on is turned on;
  • the computing resource to be closed is determined according to the distance between the computing resources in the open state, and the computing resource to be closed is closed.
  • the step of determining the computing resource to be opened according to the distance between each computing resource in the closed state and the computing resource in the open state includes:
  • the step of determining the computing resource to be shut down according to the distance between the computing resources in the open state includes:
  • the computing resource to be closed is selected from the computing resource in the open state, or is from the open state.
  • the part of the computing resource selects the computing resource to be closed, wherein the load rate based on the computing resource, the usage duration of the computing resource in the open state, and one of the running tasks of the computing resource or a combination of two or more parameters And defining, in the computing resource of the open state, a portion of the computing resource in the open state.
  • the step of spatial location layout based on the respective computing resources in the current operating state includes:
  • the location of the computing resource in the open state and the computing resource in the closed state are obtained to form the spatial location layout.
  • the step of determining the computing resource to be opened according to the distance between each computing resource in the closed state and the computing resource in the open state includes:
  • the computing resource of the computing resource with at least one closed state separated from the computing resource of each open state is selected from the computing resources in the closed state, and the computing resource is the computing resource to be opened.
  • the step of determining the computing resource to be opened according to the distance between each computing resource and the computing resource in the open state according to the closed state includes:
  • the computing resource to be started includes multiple candidate computing resources, searching for an alternative computing resource that satisfies one of a minimum current temperature and a longest closed time As a computing resource that is opened preferentially.
  • the step of determining the computing resource to be shut down according to the distance between the computing resources in the open state includes:
  • Determining that when there are two computing resources of at least the open state in the operation group of the same category The distance between the at least two computing resources and the computing resources currently running in the computing group of the other class determines the computing resource corresponding to the minimum value, and uses the computing resource as the computing resource to be closed.
  • the step of determining the computing resource to be shut down according to the distance between the computing resources in the open state includes:
  • the computing resource to be closed which is determined according to the distance between the computing resources in the open state, includes multiple computing resources
  • the selection meets the minimum load rate, the load rate is within the preset range, the usage duration is the longest, and the related person is not executed.
  • the computing resources adjacent to the computing resources of the at least two computing resources that are to be closed are selected as the computing resources that are preferentially closed.
  • An embodiment of the present invention further provides an operation control system, including:
  • the resource monitoring module is configured to determine, according to a preset condition, whether the computing resource needs to be turned on or off;
  • the heat management module is configured to: when the computing resource needs to be turned on, based on the spatial position layout of each computing resource in the current running state, according to the distance between each computing resource in the closed state computing resource and the computing resource in the open state Determining the computing resource to be opened, and turning on the computing resource to be opened; and configuring the spatial location of each computing resource in the current running state according to the distance between the computing resources in the open state when the computing resource needs to be closed. Leave, to determine the computing resources to be shut down, and close the computing resources to be shut down.
  • system further comprises:
  • a first timer configured to record a shutdown duration of the shutdown state computing resource in the shutdown mode
  • a temperature sensor configured to detect a current temperature of the shutdown state computing resource
  • the heat management module is further configured to: when the computing resource to be opened is a plurality of computing resources, find the largest distance that satisfies one of the two conditions of the lowest current temperature and the longest closing time, as the computing resource to be opened;
  • system further includes:
  • a second timer configured to record a usage duration of the open state computing resource in an open mode
  • a load monitor configured to detect a load rate of the open state computing resource
  • a task process manager configured to monitor a running task of the open state computing resource
  • the heat management module is further configured to: when the computing resource to be closed, which is determined according to the distance between the computing resources in the open state, includes multiple computing resources, the priority is that the load rate is the lowest, the load rate is within the preset range, and the usage time is long.
  • the computing resources adjacent to the computing resources with more open states among the at least two computing resources to be closed are preferentially selected as the computing resources that are preferentially closed.
  • the embodiment of the invention further provides a computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions are used to execute the heat dissipation control method of the computing resource according to the embodiment of the invention.
  • the above-mentioned heat dissipation control method, operation control system and storage medium of the computing resource can effectively select an operation resource (such as a processor) to be turned on and/or off based on the spatial position layout of the plurality of processors, so that the computing resources are interposed.
  • the influence of thermal interaction is reduced, and the traditional thermal management is forced to reduce the efficiency at high temperatures, so that the multi-processor structure of the arithmetic control system can operate efficiently.
  • FIG. 1 is a schematic diagram showing the structure of a multiprocessor system according to an embodiment of the present invention
  • FIG. 2 is a diagram showing an example of a heat dissipation control method according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of a calibration position of a multiprocessor system structure according to an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of a running state 1 of a multiprocessor system according to an embodiment of the present invention.
  • FIG. 5 is a schematic diagram of a running state 2 of a multiprocessor system according to an embodiment of the present invention.
  • FIG. 6 is a diagram showing an example of a heat dissipation control method according to another embodiment of the present invention.
  • FIG. 7 is a schematic structural diagram of a system in an embodiment of the present invention.
  • the computing resource may be a core of a central processing unit (CPU) in the processor, where the processor is a single system on chip (SoC) chip including multiple computing resources, for example, typical processor for a mobile terminal, such as Qualcomm's Snapdragon TM multi-core processors for mobile phones, comprising a central processing unit may have four, six or eight cores.
  • the computing resource may also be a Graphics Processing Unit (GPU) in the processor. It will be understood by those skilled in the art that the computing resource is not limited to the foregoing.
  • the computing resource may be a core, a main core, a sub-core, a hardware engine, or the like in the processor.
  • the above computing resources may be a single one of the above, or a combination of the above.
  • each computing resource can be turned on (powered on) to perform operations, undertake task processes, etc., and can be turned off (power off) to reduce power consumption, so each computing resource can be There are at least two modes, on and off.
  • FIG. 1 schematically shows an SoC processing chip for an intelligent mobile terminal, which includes three heterogeneous operation groups (11, 12, 13) for a total of twelve computing resources (1 to 12). Each computing resource can be turned on or off.
  • the computing resources in each computing group may be a Central Processing Unit (CPU), a Graphic Processing Unit (GPU), or a Visual Processing Unit (VPU).
  • the opening/closing control of a plurality of computing resources is mainly performed, and the plurality of computing resources are divided into a plurality of computing groups.
  • each computing group ie, the processing unit group below
  • the conditions for opening or closing can be determined according to the number of tasks of the system or the computing resource load situation currently working. Once it is decided to dynamically open new computing resources to help coordinate operations to improve performance, or to select an open operation.
  • the resource is closed to save power, and the method and system provided by the present invention are needed to implement optimization selection of computing resources, including "selecting new computing resources" and "selecting closed computing resources.”
  • a method for controlling computing resources is provided in one embodiment of the present invention.
  • step 100 based on the preset conditions, it is determined whether the computing resource needs to be turned on or off.
  • the preset condition may be whether the number of tasks of the system meets a predetermined condition, whether a workload of the currently running computing resource satisfies a predetermined condition, and the like.
  • the usage of computing resources can be used to measure the size of the workload. It can be understood that the preset condition It is not limited to the foregoing two.
  • other parameters may be monitored to determine whether to turn on or off the computing resource, for example, monitoring the temperature of the currently running computing resource, and if the temperature continues to exceed or fall below a preset threshold for a predetermined time, Then turn other computing resources on or off.
  • the use of the new computing resource condition includes: the number of working tasks increases and the computing resource load becomes heavy, that is, the number of tasks of the system reaches the set value, and the system load rate exceeds the set value, etc.
  • the closed computing resource condition used is opposite to the above: the number of working tasks is reduced or the load is lightened, so that it is no longer necessary to keep so many open computing resources to maintain the working state, and the most suitable one can be selected through the present invention.
  • the computing resources are closed.
  • the number of tasks and processes of the system can be monitored by the task process manager.
  • the workload condition of the computing resource can be measured by the load ratio, for example, whether the workload of the currently running computing resource satisfies a predetermined condition, that is, whether the load rate of the currently running computing resource satisfies a predetermined condition.
  • the load rate of the computing resource is the statistics of the usage status of the computing resources in a time period. Through this indicator, it can be seen that the computing resources are occupied in a certain period of time. If the occupied time is high, then it is needed. Considering whether the computing resources are already in overload operation, long-term overload operation will keep the electronic equipment in a high temperature state for a long time, which is a kind of damage to the electronic equipment itself, so the load rate of the computing resources must be controlled to a certain extent. Under the ratio, the operating temperature of the computing resource is kept within a certain range to ensure the normal operation of the electronic device.
  • dynamically loaded elements in an operating system of an electronic device may be employed to detect the load rate of each computing resource in the multi-computing resource structure in real time.
  • These elements may be drives for detecting the load rate of computing resources, which may be hardware or a collection of software.
  • step 200 when the computing resource needs to be turned on, based on the spatial location layout of each computing resource in the current running state, the distance between each computing resource in the closed state computing resource and the computing resource in the open state is determined.
  • the computing resource to be opened will be opened.
  • the computing resource is enabled; when the computing resource needs to be shut down, based on the spatial location layout of each computing resource in the current running state, the computing resource to be closed is determined according to the distance between the computing resources in the open state, and the computing resource to be closed is shut down.
  • step 100 when it is determined in step 100 that the computing resource needs to be turned on, only the computing resource to be turned on is determined in step 200, and then the computing resource to be turned on is turned on; when it is determined in step 100 that the computing resource needs to be closed, in step 200 Only the computing resources to be closed are determined, and then the computing resources to be shut down are closed.
  • the spatial position layout in this embodiment refers to the layout of the computing resources in the current running state on the physical space, which can be obtained in the following manner.
  • the foregoing steps of spatial location layout based on respective computing resources in a current operating state include:
  • Step 410 Based on the physical layout of all the computing resources, obtain the computing resources of the open state and the computing resources of the closed state according to the current running state of each computing resource, to form the spatial location layout.
  • the distance between the computing resources can be predetermined and stored in the distance relationship for use in determining the computing resources to be turned on/off. For example, based on the spatial position layout of the plurality of computing resources described above, a Cartesian coordinate system is established, and the position of each computing resource can be accurately determined. For example, in the coordinate system shown in FIG.
  • the position (1, 1) corresponds to the computing resource 7
  • the position (1, 2) corresponds to the computing resource 8
  • the position (1, 2) corresponds to the computing resource 8
  • the location (2, 1) corresponds to the computing resource 5
  • the location (2, 2) corresponds to the computing resource 6
  • the location (2, 3) corresponds to the computing resource 11
  • the location (2, 4) corresponds to the computing resource 12
  • the position (3, 1) corresponds to the operation resource 3
  • the position (3, 2) corresponds to the operation resource 4
  • the position (3, 3) corresponds to the operation resource 9
  • the position (1, 4) corresponds to the operation.
  • Resource 10 position (1, 4) corresponds to computing resource 1
  • location (2, 4) corresponds to computing resource 2.
  • the position is determined in a polar coordinate manner by computing resources arranged according to a circle.
  • the distance between each two computing resources can be determined, and the above distance relationship is stored.
  • the running state of each computing resource is corresponding to the location, and the spatial location layout is obtained.
  • the above operating states include an open or closed state.
  • the category of computing resources that need to be turned on or off may also be further confirmed.
  • the computing resources in the same computing group can process tasks of the same type, and the computing resources in different computing groups process different types of tasks.
  • the first operation group includes a plurality of CPUs
  • the second operation group includes a plurality of GPUs, and the like, and then the first operation group and the second operation group are respectively two categories, and the operation resources therein are different. category.
  • an computing resource can be opened to share a part of the workload.
  • the newly opened computing resource belongs to the same category as the currently running computing resource, that is, belongs to the operation group 11.
  • the category of the newly opened computing resource should belong to the operation group. 13.
  • the computing group to be opened or the computing group category to which the computing resource to be closed belongs is determined according to the determined category. Therefore, in an embodiment, when the computing resource needs to be turned on or when the computing resource needs to be shut down, the following steps are further included:
  • the computing resource to be opened is selected from the computing resource in the closed state in the computing group in the same category.
  • the second operation group includes a plurality of graphics processing units, and when the computing resources that need to be turned on, the one or more graphics processing units are selected to be turned on in the computing resources in the closed state in the second computing group. Resources.
  • the computing resource to be closed is selected from the computing resources in the open state according to the distance between the computing resource in the open state and the computing resource in the open computing state in the computing group of the same type. For example, if the first computing group includes multiple CPUs, then the computing resource in the open state needs to be closed, based on the distance between the computing resource in the open state in the first computing group and the GPU in the second computing group. To select which of the computing resources in the first computing group is turned off.
  • the computing resource to be shut down or its alternative can be selected according to the distance between the computing resources, the operation is convenient, the precision is high, and the heat management is more convenient, and the distance is determined to determine the most favorable.
  • the computing resource of heat accumulation realizes heat dissipation management by turning off the computing resource.
  • the following embodiments may also be used.
  • the operation in the open state The part of the resource that defines the open computing resource is used as a backup, and then the distance between the computing resource in the open state and the computing resource in the open computing state in the computing group of the same category is from the open state.
  • the part of the computing resource selects the computing resource to be closed.
  • the distance parameter can be combined according to any one of the load rate, the usage duration, and the running task, and the priority of the computing resource to be closed can be set, so that the heat dissipation management is more accurate and effective.
  • the computing resource to be turned on is selected from the computing resources in the off state, and may be selected from the closed state and the computing resources of the specific category.
  • the computing resource to be closed may be selected from an operating resource in an open state, and may be selected from an open state and a specific class of computing resources.
  • the computing resource to be turned on when determining the computing resource to be turned on, may be determined by one of the following rules.
  • the computing resource 1 and the current operation are selected among the computing resources 1, 2, and 3 to be selected according to the above rules.
  • the computing resources 4 and 11 are larger than the distance between the computing resources 2 and 3 and the computing resources 4 and 11, so that the computing resource 1 is determined to be the computing resource to be turned on. Compared to turning on the computing resource 2 or 3, turning on the computing resource 1 is most advantageous for the heat dissipation of the currently running computing resources 4 and 11.
  • the computing resource to be opened is selected from other computing resources to be opened, that is, the selection of the above step 1) is performed after the minimum distance is excluded from the computing resources in the closed state.
  • FIG. 4 it is assumed that the computing resources in one computing group 12 need to be opened. Compared with the computing resources 5, 7, and 8, the computing resources 6 and the running computing resources 4 and 11 have the smallest distance. It is the most disadvantageous for the heat dissipation of the computing resources 4 and 11. Therefore, one of the computing resources 5, 7, and 8 to be selected is turned on.
  • the selected criteria may be considered based on the temperature and/or the duration of the off state. For example, the computing resource with the lowest temperature may be selected, the computing resource with a long time in the closed state may be selected, or one computing resource may be randomly selected.
  • the computing resource is the computing resource to be opened. Specifically, if the selected computing resources are the same computing resource, Then, the computing resource is an computing resource to be opened. As shown in FIG. 4, both the computing resource 7 and the computing resources 4 and 11 in the open state have at least one computing resource in the off state. If the computing resource 7 is turned on, the heat dissipation of the computing resources 4 and 11 is optimal without generating heat. Gather.
  • the foregoing predetermined rules are not limited to the foregoing. Since there may be many changes in the layout of the computing resources in practice, the appropriate rules may be set according to actual conditions.
  • the main criteria for setting the rules include: using the distance as a criterion to select a computing resource that has the least influence on the heat dissipation of the currently running computing resource. When there are multiple objects to be selected, other factors can be considered. For example, based on the spatial location layout of each computing resource in the current running state, when the computing resource needs to be turned on, the operation to be turned on is determined according to the distance between each computing resource in the closed state computing resource and the computing resource in the open state. Resources.
  • the candidate computing resource may be the plurality of distances obtained by the above method.
  • the candidate computing resource may be a plurality of computing resources of the computing resource selected from the computing resources in the closed state and separated from the computing resources in each open state by at least one closed state. Therefore, the above predetermined rule may further include the following conditions:
  • An alternative computing resource that satisfies one of the two conditions of the current minimum temperature and the longest closed time is searched as a computing resource that is preferentially turned on.
  • the basis of the foregoing search process is given by detecting the current temperature of each computing resource in the closed computing resource and/or the closing duration of the closed mode.
  • the long-term unused computing resources and/or the lower-temperature computing resources are selected as the priority initiators, and the coordinated operation after the startup is implemented, so that the load can be high. The later the case, the later the time when the performance is suppressed by the heat management.
  • a timer may be employed to determine in real time the off duration of the computing resource in the off state. At the same time, a timer can also be used to determine the duration of use of the computing resource in the open state. Then one or two timers can be equipped for one computing resource. Used to record the closing time or opening time separately. Of course, it is also possible to equip multiple computing resources with a timer.
  • the method further includes: detecting a current temperature of each computing resource, and using a temperature sensor to detect a current temperature of each computing resource.
  • the current processing chip for a smart mobile terminal is configured with a temperature sensor (Sermary Sensor) for detecting the temperature of the corresponding computing resource next to each computing resource.
  • the temperature of the plurality of computing resources can be collected in real time using a temperature sensor.
  • the temperature sensor here may be one or more, for example, each computing resource is equipped with a temperature sensor, or the temperature of a plurality of computing resources is respectively detected by a temperature sensor.
  • the computing resource to be shut down when determining the computing resource to be shut down, may be determined by the following rules.
  • the load of the running computing resources of the same category is about the same. If at least two computing resources in the same type of computing resources are available, the distance between the at least two computing resources and the computing resources in the other categories is determined, and the computing resource corresponding to the minimum distance is determined. Is a resource to be closed among the at least two computing resources. Taking FIG. 5 as an example, the computing resources 9, 11, and 12 in the computing group 13 can be turned off, and the distance between the computing resource 9 and the computing resource 4 in the computing group 11 is the smallest, and the help of turning off the heat dissipation of the computing resource 4 is maximized. It will be appreciated that in other embodiments, other factors such as temperature and runtime may also be considered.
  • the operation resources 9, 11, 12 in the operation group 13 can be turned off, and the operation time of the operation resource 11 is the longest.
  • the distance from the operation resource 4 is not the smallest, the operation resource 11 can be selected as the operation resource to be closed.
  • the computing resource 12 has the highest temperature, and the computing resource 11 can also be selected as the computing resource to be turned off, because this helps the cooling of the computing resources 12 and 4.
  • the computing resource to be shut down is determined based on the distance between computing resources currently running. Specifically, in the above step 200, The steps of determining the computing resources to be shut down according to the distance between the computing resources in the open state include:
  • the computing resources in the group are adjacent to the more open state.
  • the computing resource as a computing resource that is preferentially closed, is preferentially closed.
  • the distance relationship between the computing resources can be used to set the priority of the computing resource to be closed, and the distance between the computing resources is first used to determine the computing resource that helps the heat dissipation.
  • the step of determining the computing resource to be shut down according to the distance between the computing resources in the open state further includes:
  • the computing resource to be closed which is determined according to the distance between the computing resources in the open state, includes multiple computing resources
  • the preference is to satisfy the lowest load ratio
  • the load ratio is within the preset range
  • the usage duration is the longest
  • the execution is not performed.
  • the computing resource of at least one condition in the running task of the human-computer interaction processing is used as the computing resource that is preferentially closed.
  • the detection rate by the load rate, the duration of use, and the running task, etc. Therefore, to select the alternately closed computing resources, in the case of high load, by turning off the maximum computing resources for cooling, to achieve cooling as soon as possible, rather than forcing the performance to achieve cooling.
  • the closed state computing resource can be selected according to the following priority levels: 1) the computing resource that is idle and has the greatest help to the heat dissipation is closed, that is, the distance between the computing resources according to the open state is determined.
  • the computing resource to be shut down, and the computing resource has the lowest load rate; 2) turning off the computing resource that is used for a long time and has the greatest help to the heat dissipation, that is, the distance to be closed determined by the distance between the computing resources according to the above-mentioned open state is determined.
  • the computing resource has the longest duration of use; 3) the computing resource that is not idle but the least busy, and which has the greatest help to the heat dissipation, that is, the computing resource to be closed that meets the distance between the computing resources according to the open state,
  • the running task on the human-computer interaction processing is not executed.
  • the biggest definition of heat dissipation is to select the object that is most likely to generate heat sharing with its own group or other computing group.
  • the load rate of the computing resource in the open state and the operation duration and the open state of the computing resource in the open state are monitored.
  • the second candidate computing resource of the running task of the human-computer interaction processing realizes the heat-dissipation management by shutting down the selected second candidate computing resource one by one.
  • the load rate of the computing resource in the open state, the usage duration of the computing resource in the open state, and one of the running tasks of the computing resource in the open state or more than two parameters are monitored.
  • the operation resource that uses the longest running time and does not execute the running task related to the human-computer interaction processing is selected again to perform the shutdown processing, and then it is determined whether the shutdown condition is still satisfied;
  • the computing resource whose load rate is within the preset range and the running task for the human-computer interaction processing is not performed is closed again, and then it is determined whether the shutdown condition is still satisfied;
  • the computing resources are only allowed to run in low performance mode.
  • the distance from other computing groups is mentioned in the foregoing embodiment to achieve the optimization of the computing resources to be shut down, but the technical solution of the present invention is not limited thereto.
  • the following embodiments may be used to select the alternative based on the load rate and the execution task. If two or more computing resources are available at the same time, the selected and other open states are selected. The neighbor of the computing resource is the computing resource to be closed, and the more the computing resource adjacent to the open state, the higher the priority.
  • the step of determining the computing resource to be shut down according to the distance between the computing resources according to the open state may also be implemented in the following manner:
  • the computing resources adjacent to the computing resources of the at least two operating resources that are to be closed are preferentially selected as the computing resources that are preferentially closed.
  • the load rate of the computing resource, the usage duration of the computing resource in the open state, and one of the running tasks of the computing resource or a combination of two or more parameters may be first used in the computing resource in the open state.
  • the demarcated part is used as an alternative, and then, the distance between the computing resource of the open state according to the operation group of the same category and the computing resource of the open state of the other type of operation group is selected from the part of the computing resource of the open state.
  • the above-mentioned operational resources to be closed are closed.
  • the priority of the shutdown can also be as low as the load rate is satisfied, the load rate is within the preset range, the usage time is the longest, and the human-computer interaction processing is not performed. Set the combination conditions in the task to run.
  • the least impact on the user experience can help to cool down as quickly as possible, that is, the task load is the lightest, the transfer cost is the smallest, and the temperature is the lowest after being turned off because the static is not working, then according to the characteristics of "hot to low temperature running", gathered nearby Heat can dissipate in this low temperature direction.
  • a new startup resource optimization scheme and a shutdown optimization scheme of the computing resource are provided, which include “selecting new computing resources”.
  • the normal computing resources When the normal computing resources are not used, they will enter the shutdown mode to reduce power consumption. Opening a new computing resource means turning on a closed computing resource, so that it can start to provide computing power.
  • the above method can adopt "selecting the coolest and least affecting other computing resources as the priority object", and the most cool can be the computing resource that is not used for the longest time or the computing resource with the lowest temperature, so the object can Helps to achieve the later time when performance is suppressed by Thermal Management under high load conditions.
  • the least affecting other computing resources means that the opened object starts to heat up once it is started, but its fever has the least impact on other computing resources. It also includes “Choose Closed Computational Resources”, designed to cool down as quickly as possible without compromising the user experience, rather than forcing sacrifices to achieve cooling. When the computing resources are turned off, the user's experience is least affected, and the cooling and cooling as much as possible can be helped as a priority.
  • a heat dissipation control method of the computing resources that can realize the shutdown management of the computing resources by using the spatial position layout of the plurality of computing resources in the computing control system is also provided.
  • the spatial location layout of multiple computing resources in the computing control system is used to realize the opening and closing management of the computing resources.
  • the two methods are the opposite two processes, which are complementary and do not have to be executed at the same time. In the same execution process, a new computing resource cooling control method is formed.
  • the technical solution of the present invention which is essential or contributes to the prior art, may be embodied in the form of a software product carried on a non-transitory computer readable storage carrier (eg ROM, disk, optical disk, server storage space, including a plurality of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to execute the system structure and method described in various embodiments of the present invention .
  • a non-transitory computer readable storage carrier eg ROM, disk, optical disk, server storage space, including a plurality of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to execute the system structure and method described in various embodiments of the present invention .
  • an operation control system is further provided, which is applied to the above-mentioned chip having a plurality of computing resource layouts.
  • the system includes:
  • the resource monitoring module 302 is configured to determine, according to a preset condition, whether the computing resource needs to be turned on or off;
  • the heat management module 309 is configured to: when the computing resource needs to be turned on, the distance between each computing resource and the computing resource in the open state according to the spatial location layout of each computing resource in the current running state, The computing resource to be opened is determined, and the computing resource to be opened is enabled; and when the computing resource needs to be closed, the spatial location of each computing resource in the current running state is based on the distance between the computing resources in the open state. To determine the computing resource to be shut down, and to close the computing resource to be shut down.
  • the heat dissipation management module 309 is further configured to determine, when the computing resource needs to be turned on or when the computing resource needs to be shut down, the class of the operation group to which the computing resource of the currently open state belongs; The distance between each computing resource and the computing resource in the open state is selected from the computing resources in the closed state in the computing group in the same category, or the thermal management module 309 is configured to be based on the same category.
  • the distance between the operation resource of the open state in the operation group and the operation resource of the open state in the operation group of the other category, and the operation resource to be closed is selected from the operation resource of the open state, or from the operation resource of the open state Partially selecting the computing resource to be shut down, wherein the load rate based on the computing resource, the usage duration of the computing resource in the open state, and one of the running tasks of the computing resource or a combination of two or more parameters are in an open state.
  • the computing resource of the computing resource section is selected from the operation resource of the open state, or from the operation resource of the open state Partially selecting the computing resource to be shut down, wherein the load rate based on the computing resource, the usage duration of the computing resource in the open state, and one of the running tasks of the computing resource or a combination of two or more parameters are in an open state.
  • the system further includes a location determining module 301 configured to acquire, according to a physical layout of all computing resources, a computing resource in an open state and a computing resource in a closed state according to a current operating state of each computing resource. Forming the above spatial position layout to provide the thermal management module as an alternative reference.
  • the location determining module 301 may be configured to perform the method for forming a spatial location layout in the foregoing step 100. For details, refer to the related description in the foregoing.
  • the heat dissipation management module 309 is further configured to determine a distance between each of the computing resources in the closed state and the computing resources in the open state, and select and calculate each of the computing resources in the closed state. If the distance of the computing resource in the open state is the largest, if the largest distance is the same computing resource, the computing resource is the computing resource to be opened; or, the computing resource from the closed state is selected and the operation is started. When the resource is separated from the computing resource of the computing resource of the closed state, the computing resource is the computing resource to be opened.
  • the heat dissipation management module 309 is further configured to determine a distance between the computing resource in the closed state and the computing resource in the open state, and the computing resource from the closed state. The source selects the smallest distance from the computing resource of each open state. If the minimum distance is the same computing resource, the minimum distance is excluded from the closed computing resource.
  • system further includes:
  • the first timer 305 is configured to record the closing time of the closed resource in the closed mode
  • a temperature sensor 304 configured to detect a current temperature of the computing resource in the off state
  • the heat management module 309 is further configured to: when the computing resource to be opened includes a plurality of candidate computing resources, search for an alternative computing resource that satisfies one of a minimum current temperature and a longest closing time, as a priority open operation Resources.
  • the heat dissipation management module 309 is further configured to determine a category of the operation group to which the operation resource of the current open state belongs, and determine at least two operations when the operation group of the same category has at least two computing resources.
  • the distance between the resource and the currently running computing resource in the operation group of the other category determines the computing resource corresponding to the minimum value, and uses the computing resource as the computing resource to be closed.
  • system further includes:
  • the second timer 306 is configured to record the usage duration of the computing resource in the open state in the open mode
  • the load monitor 307 is configured to detect a load rate of the computing resource in the open state
  • the task process manager 303 is configured to monitor a running task of the computing resource in the open state.
  • the heat management module 309 is further configured to monitor a load rate of the computing resource in the open state, a use duration of the computing resource in the open state, and a running parameter of the computing resource in the open state, or a combination of two or more parameters;
  • the priority is that the load rate is the lowest, the load rate is within the preset range, the longest usage time, and the human machine is not executed.
  • Interactive processing An operation resource of at least one condition in a running task, as an operation resource to be shut down; or
  • the computing resources adjacent to the computing resources of the at least two computing resources that are to be closed are preferentially selected as the computing resources to be closed.
  • the resource monitoring module 302 is configured to perform the above-mentioned step 100
  • the heat-dissipation management module 309 is configured to perform the above-mentioned step 200. Therefore, for specific details, refer to the foregoing related description, and no further description is provided herein.
  • the first timer and the second timer may be one or more.
  • the location determining module 301 and the heat dissipation management module 309 may be implemented by using a single added computing resource or multiple computing resources, or may be implemented by one or more of the plurality of computing resources 400.
  • the task process manager 303 provides information about the performance of the computer and displays detailed information about the programs and processes running on the computer, for example, can be used to monitor the current state of programs running on all computing resources.
  • the task process manager 303 provides information about the performance of the computer and displays detailed information about the programs and processes running on the computer, for example, can be used to monitor the current state of programs running on all computing resources.
  • the load monitor 307 can detect the load rate of each computing resource in real time using dynamically loaded components in the operating system of the electronic device.
  • These elements may be drives for detecting the load rate of computing resources, which may be hardware or a collection of software.
  • the system further includes a storage module 308.
  • the heat dissipation management module 309 is further configured to simultaneously detect one or two or more modules according to the load monitor 307 and the temperature sensor 304, the first timer, the second timer, the position determining module, the task process manager, and the like, respectively.
  • the load rate, the duration of use, the duration of the shutdown, the current temperature, and the like are high to low.
  • the computing resource 400 performs sorting in descending order, obtains an index table of the computing resource 400 and its corresponding parameters such as temperature and load rate, and stores the index table in the storage module 308.
  • the heat management module 309 can find a suitable computing resource from the computing resource according to the index table. This allows real-time monitoring of the usage of each computing resource.
  • the control device for the computing resource can be implemented by a personal computer in practical applications.
  • the resource monitoring module 302, the heat dissipation management module 309, the location determining module 301, the first timer 305, the second timer 306, the load monitor 307, and the task process manager 303 in the control device of the computing resource are actually applied.
  • the central processing unit (CPU), the digital signal processor (DSP), the micro control unit (MCU) or the programmable gate array (FPGA, Field-) can be used in the device.
  • the Programmable Gate Array is implemented; the storage module 308 in the control device of the computing resource can be implemented by a memory in the device in an actual application.
  • the heat dissipation control method and the operation control system of the above computing resources can efficiently select an operation resource (such as an operation resource) to be turned on and/or off based on a spatial position layout of a plurality of computing resources, so that the computing resources are thermally interacted with each other.
  • the effect is reduced, and the traditional thermal management is forced to reduce the efficiency at high temperatures, so that the arithmetic control system that maintains the multi-computing resource structure can operate efficiently.
  • the disclosed apparatus and method may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the unit is only a logical function division.
  • there may be another division manner such as: multiple units or components may be combined, or Can be integrated into another system, or some features can be ignored or not executed.
  • the components shown or discussed are mutually
  • the coupling, or direct coupling, or communication connection may be an indirect coupling or communication connection through some interface, device or unit, and may be electrical, mechanical or otherwise.
  • the units described above as separate components may or may not be physically separated, and the components displayed as the unit may or may not be physical units, that is, may be located in one place or distributed to multiple network units; Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may be separately used as one unit, or two or more units may be integrated into one unit;
  • the unit can be implemented in the form of hardware or in the form of hardware plus software functional units.
  • the foregoing program may be stored in a computer readable storage medium, and the program is executed when executed.
  • the foregoing storage device includes the following steps: the foregoing storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
  • ROM read-only memory
  • RAM random access memory
  • magnetic disk or an optical disk.
  • optical disk A medium that can store program code.
  • the above-described integrated unit of the present invention may be stored in a computer readable storage medium if it is implemented in the form of a software function module and sold or used as a standalone product.
  • the technical solution of the embodiments of the present invention may be embodied in the form of a software product in essence or in the form of a software product stored in a storage medium, including a plurality of instructions.
  • a computer device (which may be a personal computer, server, or network device, etc.) is caused to perform all or part of the methods described in various embodiments of the present invention.
  • the foregoing storage medium includes various media that can store program codes, such as a mobile storage device, a ROM, a RAM, a magnetic disk, or an optical disk.
  • the technical solution of the embodiment of the present invention can effectively select an operation resource (such as a processor) to be turned on and/or off based on a spatial location layout of multiple processors, so that the influence of thermal interaction between computing resources is reduced, and the slowdown is high.
  • an operation resource such as a processor
  • the traditional thermal management forces the performance to be reduced, so that the multi-processor structure of the arithmetic control system can operate efficiently.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Power Sources (AREA)

Abstract

本发明实施例公开了一种运算资源的散热控制方法、运算控制系统和存储介质，所述运算控制系统包括：资源监控模块，配置为基于预设的条件，确定是否需要开启或关闭运算资源；散热管理模块，配置为当需要开启运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源，将待开启的运算资源开启；还配置为当需要关闭运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照开启状态的运算资源之间的距离，来确定待关闭的运算资源，将待关闭的运算资源关闭。

Description

运算资源的散热控制方法、运算控制系统和存储介质 技术领域
本发明涉及移动终端技术领域，特别是涉及一种运算资源的散热控制方法、运算控制系统和存储介质。
背景技术
随着技术的不断发展，处理芯片在越来越小的面积上摆放越来越多的运算资源。因为高密度的设计会造成运算资源启动之后，与其他运算资源一同发热，如此交互影响除了让热不易消散导致整体温度上升，各种运算资源的热交互影响也就更加严重，热也更不易散出。
对于应用于智能移动终端(例如手机、平板电脑)中的处理器，上述的发热问题更加突出，因为智能移动终端因体积的限制，只能采用被动散热的方式，无法采用风扇等装置进行主动散热。现有的用于智能移动终端的处理器的中央处理单元包括多个核心，在核心温度较高时的典型做法是将核心的运行频率降低，当核心的运行频率的降幅较大时，会影响处理器的效能。
发明内容
基于此，为了解决各种运算资源高密度集成所带来的热量交互影响的问题，本发明实施例提供了一种运算资源的散热控制方法、运算控制系统和存储介质，能够有效降低多处理器结构的运算控制系统的温度，并能够保持多处理器结构的运算控制系统的高效运转。
本发明实施例提供了一种运算资源的散热控制方法，所述方法包括：
基于预设的条件，确定是否需要开启或关闭运算资源；
当需要开启运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源，将待开启的运算资源开启；
当需要关闭运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照开启状态的运算资源之间的距离，来确定待关闭的运算资源，将待关闭的运算资源关闭。
在其中一个实施例中，所述当需要开启运算资源或需要关闭运算资源时，还包括以下步骤：
确定当前开启状态的运算资源所属的运算组的类别；
所述依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源的步骤包括：
依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，在位于同一类别的运算组中从关闭状态的运算资源选取所述待开启的运算资源，或者，
所述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤包括：
依据同一类别的运算组中开启状态的运算资源与其他类别的运算组中开启状态的运算资源之间的距离，从开启状态的运算资源中选取所述待关闭的运算资源，或从开启状态的运算资源中的部分选取所述待关闭的运算资源，其中，基于运算资源的负载率、运算资源处于开启状态下的使用时长和运算资源的运行任务中的其中一个参量或两个以上参量的组合，在开启状态的运算资源中划定所述开启状态的运算资源中的部分。
在其中一个实施例中，所述基于各个运算资源在当前运行状态下的空间位置布局的步骤包括：
基于全部运算资源的物理布局，依据各个运算资源的当前运行状态， 获取开启状态的运算资源和关闭状态的运算资源的位置，形成所述空间位置布局。
在其中一个实施例中，所述依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源的步骤包括：
确定关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，从关闭状态的运算资源中选出与每个开启状态的运算资源的距离最大者，如果所述距离最大者为同一个运算资源，则该运算资源为待开启的运算资源；或者，
从关闭状态的运算资源中选出与每个开启状态的运算资源间隔至少一个关闭状态的运算资源的运算资源，则该运算资源为待开启的运算资源。
在其中一个实施例中，所述依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离来确定待开启的运算资源的步骤还包括：
确定关闭状态的运算资源中与开启状态的运算资源之间的距离，从关闭状态的运算资源中选出与每个开启状态的运算资源的距离最小者，如果所述距离最小者为同一个运算资源，则将该距离最小者从关闭状态的运算资源中排除选择。
在其中一个实施例中，所述方法中，当所述待开启的运算资源包括多个备选运算资源时，查找满足当前温度最低和关闭时长最长这两个条件之一的备选运算资源，作为优先开启的运算资源。
在其中一个实施例中，所述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤中包括：
确定当前开启状态的运算资源所属的运算组的类别，
当同一类别的运算组存在至少开启状态的两个运算资源时，确定所述 至少两个运算资源与其他类别的运算组中当前正在运行的运算资源的距离，确定距离最小值所对应的运算资源，将该运算资源作为待关闭的运算资源。
在其中一个实施例中，所述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤包括：
监控开启状态的运算资源的负载率、运算资源处于开启状态下的使用时长和开启状态的运算资源的运行任务中的其中一个参量或两个以上参量的组合；
当依照开启状态的运算资源之间的距离来确定的待关闭的运算资源包括多个运算资源时，则选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为优先关闭的运算资源；或者，
选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为待关闭的运算资源，当同一类别的运算组中存在至少两个待关闭的运算资源时，则选择所述至少两个待关闭的运算资源中与越多开启状态的运算资源相邻的运算资源，作为优先关闭的运算资源。
本发明实施例还提供了一种运算控制系统，其包括：
资源监控模块，配置为基于预设的条件，确定是否需要开启或关闭运算资源；
散热管理模块，配置为当需要开启运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源，将待开启的运算资源开启；还配置为当需要关闭运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照开启状态的运算资源之间的距 离，来确定待关闭的运算资源，将待关闭的运算资源关闭。
在其中一个实施例中，所述系统还包括：
第一定时器，配置为记录所述关闭状态运算资源处于关闭模式的关闭时长，
温度传感器，配置为检测所述关闭状态运算资源的当前温度，
散热管理模块还配置为当所述待开启的运算资源为多个运算资源时，查找满足当前温度最低和关闭时长最长这两个条件之一的距离最大者，作为待开启的运算资源；
和/或所述系统还包括：
第二定时器，配置为记录所述开启状态运算资源处于开启模式下的使用时长；
负载监控器，配置为检测所述开启状态运算资源的负载率；
任务进程管理器，配置为监控所述开启状态运算资源的运行任务；
散热管理模块还配置为当依照开启状态的运算资源之间的距离来确定的待关闭的运算资源包括多个运算资源时，则优先选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为优先关闭的运算资源；或者，
还配置为选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为待关闭的运算资源，当同一类别的运算组中存在至少两个待关闭的运算资源时，则优先选择所述至少两个待关闭的运算资源中与越多开启状态的运算资源相邻的运算资源，作为优先关闭的运算资源。
本发明实施例还提供了一种计算机存储介质，所述计算机存储介质中存储有计算机可执行指令，所述计算机可执行指令用于执行本发明实施例所述的运算资源的散热控制方法。
上述的运算资源的散热控制方法、运算控制系统和存储介质，能够基于多个处理器的空间位置布局，有效地选取运算资源(如处理器)做开启和/或关闭的处理，让运算资源间热交互的影响降低，减缓因为高温度下传统散热管理强迫降低效能的情况，从而保持多处理器结构的运算控制系统能够高效运转。
附图说明
图1为本发明的一个实施例的多处理器系统结构的示意图；
图2为本发明的一个实施例的散热控制方法的示例图；
图3为本发明的一个实施例的多处理器系统结构定标位置的示意图；
图4为本发明的一个实施例的多处理器系统结构运行状态1的示意图；
图5为本发明的一个实施例的多处理器系统结构运行状态2的示意图；
图6为本发明的另一个实施例的散热控制方法的示例图；
图7为本发明的一个实施例中系统结构示意图。
具体实施方式
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
本发明的实施例中提供了一种运算资源的控制方法。所述运算资源可以为处理器中的中央处理单元(Central Processing Unit,CPU)的核心，所述处理器为包含多个运算资源的单一片上系统(System on Chip，SoC)芯片，例如，现有典型的用于移动终端的处理器，如美国高通公司的用于手机的多核心的骁龙TM处理器，包含的中央处理单元可以具有4个、6个或8个核心。所述运算资源还可以为处理器中的图形处理单元(Graphics Processing Unit,GPU)。可以理解地，本领域技术人员会意识到，所述运算 资源并不局限于前述二者，在其他实施方式中，该运算资源可以为处理器中的核心、主核心、子核心、硬件引擎等具有计算能力的组件。上述运算资源可以为上述的单独一种，或是上述的多种之组合。
另外，在本发明的其中一些实施例中，每个运算资源可以被开启(加电)来进行运算、承接任务进程等，可以被关闭(断电)来降低耗电，因此每个运算资源可以至少有开启和关闭两种模式。图1示意性的显示一种用于智能移动终端的SoC处理芯片，其包括三种异质运算组(11、12、13)共十二个运算资源(1至12)。每个运算资源可以是开启或关闭的运作状态。每种运算组中的运算资源可能为中央处理单元(Central Processing Unit,CPU)、图形处理单元(Graphic Processing Unit,GPU)、或视觉处理单元(Visual Processing Unit,VPU)。本实施例中主要是针对多个运算资源的开启或关闭控制，并且多个运算资源被分为多个运算组。为了电量与效能考虑，各个运算组(即下文的处理单元组)都有布置软件调适器，根据实时的使用情况来决定组内哪些运算资源应该做开启或关闭。开启或关闭的条件，可以是依据系统的任务数量或是目前正在工作的运算资源负载情形来决定，一旦决定要动态开启新的运算资源帮助协同运作来增进效能，或是选取一个开启中的运算资源做关闭来节省电量，则需要利用本发明提供的方法和系统来实现运算资源的优化挑选，包括“挑选新运算资源”与”挑选被关闭运算资源”。
如图2所示，本发明的其中一个实施例中提供了一种运算资源的控制方法。
在步骤100中，基于预设的条件，确定是否需要开启或关闭运算资源。
在一个实施方式中，所述预设条件可以为系统的任务数量是否满足预定条件、当前正在运行的运算资源的工作负载是否满足预定条件等。可以运算资源的使用率来衡量工作负载的大小。可以理解的，所述预设的条件 并不局限于前述两者，在需要时，可以监控其他参数来决定是否开启或关闭运算资源，例如，监测当前运行的运算资源的温度，如果温度持续超过或低于预设阈值达预定时间，则开启或关闭其他的运算资源。
本发明实施例中，使用的开启新运算资源条件包括:工作任务数变多与运算资源负载变重，也就是说，系统的任务数量到达设定值、系统负载率超过设定值等需要通过开启新的运算资源来协同运算的条件。本发明实施例中，使用的关闭运算资源条件则与上相反：工作任务数减少或负载变轻，因此不再需要这么多开启的运算资源保持工作状态，可以透过本发明来挑选最合适的运算资源做关闭。本实施例中，系统的任务数量和进程等可以通过任务进程管理器来监控。运算资源的工作负载情况可以通过负载率来衡量，例如，当前正在运行的运算资源的工作负载是否满足预定条件，即当前正在运行的运算资源的负载率是否满足预定条件。
运算资源的负载率，顾名思义就是对一个时间段内运算资源的使用状况的统计，通过这个指标可以看出在某一个时间段内运算资源被占用的情况，如果被占用时间很高，那么就需要考虑运算资源是否已经处于超负荷运作，长期超负荷运作会使得电子设备长期保持一种高温的状态，这对电子设备本身来说是一种损害，因此必须将运算资源的负载率控制在一定的比例下，即将运算资源的工作温度保持在一定范围内，以保证电子设备的正常运作。在本发明的其中一些实施例中，可以采用电子设备的操作系统中动态加载的元件来实时地检测多运算资源结构中的每个运算资源的负载率。这些元件可以是用于检测运算资源的负载率的驱动(drive)，该驱动可以是硬件也可以是软件集合。
在步骤200中，当需要开启运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源，将待开启的 运算资源开启；当需要关闭运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照开启状态的运算资源之间的距离，来确定待关闭的运算资源，将待关闭的运算资源关闭。可以理解的，当步骤100中确定需要开启运算资源时，在步骤200中仅确定待开启的运算资源，然后将待开启的运算资源开启；当步骤100中确定需要关闭运算资源时，在步骤200中仅确定待关闭的运算资源，然后将待关闭的运算资源关闭。
本实施例中的空间位置布局是指当前运行状态下的运算资源在物理空间上的布局，具体可以通过以下方式来获取。
在本发明的其中一个实施例中，如图6所示，上述基于各个运算资源在当前运行状态下的空间位置布局的步骤包括：
步骤410，基于全部运算资源的物理布局，依据各个运算资源的当前运行状态，获取开启状态的运算资源和关闭状态的运算资源的位置，形成所述空间位置布局。
各个运算资源之间的距离可以预先确定及将上述距离关系存储起来，以备在确定待开启/关闭的运算资源时使用。例如，基于上述运多个运算资源的空间位置布局，建立直角坐标系，，可以精确确定每个运算资源的位置。例如图3所示的坐标系中，位置(1,1)对应的是运算资源7，位置(1,2)对应的是运算资源8，位置(1,2)对应的是运算资源8，位置(2,1)对应的是运算资源5，位置(2,2)对应的是运算资源6，位置(2,3)对应的是运算资源11，位置(2,4)对应的是运算资源12，位置(3,1)对应的是运算资源3，位置(3,2)对应的是运算资源4，位置(3,3)对应的是运算资源9，位置(1,4)对应的是运算资源10，位置(1,4)对应的是运算资源1，位置(2,4)对应的是运算资源2。同样还可以依据多个运算资源的空间位置，设置三维空间坐标来确定位置。还可以采用更加简单的方式，直接多个运算资源进行编号，例如图1中的运算资源1,……,运算资源12。还可 以依据圆形排列的运算资源按照极坐标的方式来确定位置。当位置确定后，可以确定每两个运算资源之间的距离，并且存储上述距离关系。同时，根据监控的各个运算资源的当前运行状态，将各个运算资源的运行状态与位置对应，获得上述空间位置布局。上述运行状态包括开启或关闭状态。
在一个实施方式中，对于步骤100，还可以进一步地确认需要开启或关闭的运算资源的类别。以图1为例，同运算组中的运算资源可以处理同类型的任务，不同运算组中的运算资源处理不同类型的任务。例如，第一运算组中包括多个CPU，第二运算组中包括多个GPU，等等，那么第一运算组和第二运算组分别为两个类别，而其中的运算资源则分属不同类别。
假设运算组11中有一个运算资源的负载超过了预定阈值达一定时间，可以开启一个运算资源来分担其一部分工作负载。此时，需要新开启的运算资源与当前运行的这个运算资源属于同一类别即均属于运算组11。又如，假设运算组13中有一个运算资源的负载超过了预定阈值达一定时间，当有新的任务需要运算组13中的运算资源处理时，则新开启的运算资源的类别应该属于运算组13。当需要开启或关闭的运算资源的类别确定后，步骤200中会依照上述确定的类别来确定待开启的运算资源或待关闭的运算资源所属的运算组类别。因此，在一个实施方式中，所述当需要开启运算资源时或当需要关闭运算资源时，还包括以下步骤：
首先，确定当前正在运行的运算资源所属的运算组的类别；
然后，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，在位于同一类别的运算组中从关闭状态的运算资源选取待开启的运算资源。例如，第二运算组包括多个图形处理单元，那么需要开启的运算资源时，那么就在第二运算组中处于关闭状态的运算资源中选取开启一个或多个图形处理单元作为待开启的运算资源。
或者，在上述依照开启状态的运算资源之间的距离来确定待关闭的运 算资源的步骤中：依据同一类别的运算组中开启状态的运算资源与其他类别的运算组中开启状态的运算资源之间的距离，从开启状态的运算资源中选取上述待关闭的运算资源。例如，第一运算组中包括多个CPU，那么其中开启状态的运算资源需要关闭时，基于第一运算组中开启状态的运算资源与第二运算组中处于开启状态的GPU之间的距离，来选择关闭第一运算组中开启状态的运算资源中的哪一个。本实施例中可以依据运算资源之间的距离，来选择上述待关闭的运算资源或其备选项，运算方便，精确度高，而且也更加利于有效地进行散热管理，通过距离识别来确定最利于热量聚集的运算资源，通过关闭该运算资源来实现散热管理。
当然还可以采用以下实施例，首先，基于运算资源的负载率、运算资源处于开启状态下的使用时长和运算资源的运行任务中的其中一个参量或两个以上参量的组合，在开启状态的运算资源中划定开启状态的运算资源中的部分作为备选项，然后，依据同一类别的运算组中开启状态的运算资源与其他类别的运算组中开启状态的运算资源之间的距离，从开启状态的运算资源中的部分选取所述待关闭的运算资源。本实施例中可以依据负载率、使用时长和运行任务中的任意一项参数来结合距离参量，设定待关闭的运算资源的优先级，使得散热管理更加精确、有效。
在本发明的一个实施方式中，待开启的运算资源从关闭状态的运算资源中选取，其可以是从关闭状态的且特定类别的运算资源中选取。所述待关闭的运算资源可以从开启状态的运算资源中选取，其可以是从开启状态的且特定类别的运算资源中选取。
在本发明的一个实施方式中，在确定待开启的运算资源时，可通过下述规则之一来确定待开启的运算资源。
1)确定每一个关闭状态的运算资源与正在运行的运算资源(即开启状态的运算资源)之间的距离，从关闭状态的运算资源中选出与每个正在运 行的运算资源的距离最大者，如果上述距离最大者为同一个运算资源，则该运算资源为待开启的运算资源。由于新开启的运算资源与所有的正在运行的运算资源的距离都是最大的，其对所有的正在运行的运算资源的散热影响最小。否则，如果新开启的运算资源与一个或多个正在运行的运算资源的距离很近，其被开启后会发热，距离越近，越不利于彼此的散热。以图4为例，运算资源4和11当前运行，若需要新开启一个运算组11中的运算资源，则依照上述规则，待选的运算资源1、2、3中，运算资源1与当前运行的运算资源4和11距离比运算资源2和3与运算资源4和11的距离都大，因此确定运算资源1为待开启的运算资源。相较于开启运算资源2或3，开启运算资源1对当前运行的运算资源4和11的散热最有利。
2)确定每一个待选的运算资源与正在运行的运算资源之间的距离，从中选出与每个正在运行的运算资源的距离最小者，如果上述距离最小者为同一个运算资源，则将该距离最小者排除，从其他待选的运算资源中选择待开启的运算资源，即，从将该距离最小者从关闭状态的运算资源中排除选择后再执行上述步骤1)的选择。以图4为例，假设需要开启一个运算组12中的运算资源，相较于运算资源5、7、8，运算资源6与正在运行的运算资源4和11距离都最小，若开启运算资源6，对运算资源4和11的散热最不利。因此，从待选的运算资源5、7、8中择一开启。选择的标准可以基于温度和/或处于关闭状态的时长来考虑，例如，可以选择温度最低的运算资源，可以选择处于关闭状态的时长长的运算资源，亦可随机选择一个运算资源。
3)确定关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离间隔，从关闭状态的运算资源中选出与每个开启状态的运算资源间隔至少一个关闭状态的运算资源的运算资源，则该运算资源为待开启的运算资源。具体的还可以是，如果选出的运算资源为同一个运算资源， 则该运算资源为待开启的运算资源。如图4所示，运算资源7与开启状态的运算资源4和11均存在至少一个关闭状态的运算资源，若开启运算资源7，对运算资源4和11的散热最佳，而不会产生热量聚集。
可以理解地，前述的预定规则并不局限与前文所述，由于实际中的运算资源的布局可能有很多变化，可以依据实际情况来设定合适的规则。设定所述规则的主要标准包括：以距离为判断基准，选择对当前运行的运算资源的散热影响最小的运算资源。当存在多个待选对象时，可以考虑其他因素。比如，基于各个运算资源在当前运行状态下的空间位置布局，当需要开启运算资源时，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离来确定待开启的运算资源。
在本发明的其中一个实施例中，如果上述过程中确定的待开启的运算资源包括多个备选运算资源时，例如，该备选运算资源可以是通过上述方法获得的多个距离最大者，或者该备选运算资源还可以是，从关闭状态的运算资源中选出的、与每个开启状态的运算资源间隔至少一个关闭状态的运算资源的多个运算资源。于是上述预定规则还可以包括以下条件：
查找满足当前温度最低和关闭时长最长这两个条件之一的备选运算资源，作为优先开启的运算资源。其中通过检测所述关闭状态的运算资源中各个运算资源的当前温度、和/或处于关闭模式的关闭时长，来给出上述查找过程的依据。在本发明的实施例中通过当前温度和关闭时长的检测，来挑选长期未使用的运算资源和/或温度较低的运算资源作为优先启动者，实现启动后的协同运算，这样可以在高负载的情况下越晚达到效能被散热管理抑制的时间点。
在本发明的其中一些实施例中，可以采用定时器来实时确定运算资源处于关闭状态的关闭时长。同时还可以采用定时器来确定运算资源处于开启状态下的使用时长。那么对于一个运算资源可以配备一个或两个定时器， 用以分别记录关闭时长或开启时长。当然也可以多个运算资源配备一个定时器。
在本发明的其中一个实施例中，上述方法还包括：检测各个运算资源的当前温度，可以采用温度传感器来侦测各个运算资源的当前温度。例如，现在的用于智能移动终端的处理芯片，都会在各个运算资源旁，配置温度传感器(Thermal Sensor)，用于侦测相应的运算资源的温度。可以采用温度传感器实时地采集多个运算资源的温度。这里的温度传感器可以是一个，也可以是多个，例如，每个运算资源配备一个温度传感器，或者通过一个温度传感器分别检测多个运算资源的温度。
在本发明的一个实施方式中，在确定待关闭的运算资源时，可通过下述规则来确定待关闭的运算资源。
假定同类别的正在运行的运算资源的负载大致相同。如果同一类别的运算资源中，存在至少两个运算资源可以关闭，确定所述至少两个运算资源与其他类别中正在运行的运算资源的距离，确定其中距离最小值所对应的运算资源，其即是所述至少两个运算资源中的待关闭的资源。以图5为例，运算组13中运算资源9、11、12可以关闭，运算资源9与运算组11中的运算资源4的距离最小，将其关闭对运算资源4的散热的帮助最大。可以理解地，在其他实施方式中，还可以考虑其他因素，例如温度和运行时间。例如，运算组13中运算资源9、11、12可以关闭，运算资源11的运行时间最长，虽然其距离运算资源4的距离不是最小的，但是可以选择运算资源11作为待关闭的运算资源。又如，运算资源12的温度最高，亦可以选择运算资源11作为待关闭的运算资源，因为这对运算资源12和4的散热都有帮助。
因此，在本发明的其中一个实施例中，依据当前正在运行的运算资源之间的距离，来确定待关闭的运算资源。具体地，在上述步骤200中，依 照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤中包括：
首先，确定当前正在运行的运算资源所属的运算组的类别；
然后，当同一类别的运算组存在至少两个开启状态的运算资源时，确定所述至少两个开启状态的运算资源与其他类别的运算组中当前正在运行的运算资源的距离，确定距离最小值所对应的运算资源，将该运算资源作为待关闭的运算资源。有关运算组的类别可参见前文的相关描述。
更进一步地，当距离最小值所对应的运算资源为多个时，则在同一类别的运算组内，依据组内运算资源间的距离关系，优选组内与越多开启状态的运算资源相邻的运算资源，作为优先关闭的运算资源，优先进行关闭处理。
可见，本实施例中，可以利用运算资源间的距离关系，来设定待关闭运算资源的优先级，而首先利用运算资源间的距离关系确定对散热帮助最大的运算资源。
在本发明的其中一个实施例中，上述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤还包括：
监控开启状态的运算资源的负载率、运算资源处于开启状态下的使用时长和开启状态的运算资源的运行任务中的其中一个参量或两个以上参量的组合；
当依照开启状态的运算资源之间的距离来确定的待关闭的运算资源包括多个运算资源时，则优先选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为优先关闭的运算资源。有关依照开启状态的运算资源之间的距离来确定待关闭的运算资源的过程可参见前文相关说明。
在本发明的实施例中通过负载率、使用时长、和运行任务等的检测结 果，来挑选备选关闭的运算资源，这样可以在高负载的情况下通过关闭对散热帮助最大的运算资源，来尽快实现散热降温，而不是强制牺牲效能来达到降温目的。按照上述实施例的方法，可以按照以下优先级来选择被关闭的开启状态运算资源：1)关闭闲置且对散热帮助最大的运算资源，即满足上述依照开启状态的运算资源之间的距离来确定的待关闭的运算资源，且该运算资源的负载率最低；2)关闭长时间使用且对散热帮助最大的运算资源，即满足上述依照开启状态的运算资源之间的距离来确定的待关闭的运算资源，且使用时长最长；3)关闭非闲置但最不忙碌，且对散热帮助最大的运算资源，即满足上述依照开启状态的运算资源之间的距离来确定的待关闭的运算资源，且未执行关于人机交互处理的运行任务。对散热帮助最大的定义是指去选取最容易与自己组或其他运算组产生共热聚热的对象。
于是，在本发明的其中一些实施例中，在上述步骤200中，上述当需要关闭运算资源时，监控开启状态的运算资源的负载率、运算资源处于开启状态下的使用时长和开启状态的运算资源的运行任务中的其中一个参量或两个以上参量的组合；依次选择负载率最低、使用时长最长且未执行关于人机交互处理的运行任务、负载率在预设范围内且未执行关于人机交互处理的运行任务的第二备选运算资源，通过逐一关闭上述选择的第二备选运算资源，来实现散热管理。
当然，还可以，当需要关闭运算资源时，监控开启状态的运算资源的负载率、运算资源处于开启状态下的使用时长和开启状态的运算资源的运行任务中的其中一个参量或两个以上参量的组合；
首先，选择负载率最低的运算资源进行关闭处理，然后判断是否还满足关闭条件；
如果还满足，则再次选择使用时长最长且未执行关于人机交互处理的运行任务的运算资源进行关闭处理，然后判断是否还满足关闭条件；
如果满足，则再次选择负载率在预设范围内且未执行关于人机交互处理的运行任务的运算资源进行关闭处理，然后判断是否还满足关闭条件；
如果满足，则强迫运算资源只准运行在低效能模式下。
上述实施例中提到和其他运算组的距离，来实现待关闭的运算资源的优选，但是本发明的技术方案不限于此。同运算组的运算资源，还可以采用下述实施例中先基于负载率和执行任务等来挑选备选项，若同时存在两个(含)以上的运算资源可选择，则挑选有和其他开启状态运算资源相邻者为待关闭运算资源，与越多开启状态的运算资源相邻，越优先。例如，在本发明的其中一个实施例中，在上述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤还可以采用以下方式来实现：
监控开启状态的运算资源的负载率、运算资源处于开启状态下的使用时长和开启状态的运算资源的运行任务中的其中一个参量或两个以上参量的组合；
选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为待关闭的运算资源，当同一类别的运算组中存在至少两个待关闭的运算资源时，则优先选择所述至少两个待关闭的运算资源中与越多开启状态的运算资源相邻的运算资源，作为优先关闭的运算资源。
在本实施例中，可以先基于运算资源的负载率、运算资源处于开启状态下的使用时长和运算资源的运行任务中的其中一个参量或两个以上参量的组合，在开启状态的运算资源中划定部分作为备选，然后，在依据同一类别的运算组中开启状态的运算资源与其他类别的运算组中开启状态的运算资源之间的距离，从开启状态的运算资源中的部分选取优先进行关闭处理的上述待关闭的运算资源。同理，关闭的优先级也可以按照满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的 运行任务中的组合条件来设定。
运算资源被关闭时，其身上目前正在执行的任务必须被转移到其他运算资源身上。因此，最不影响用户体验，即是此运算资源目前身上被赋予的执行任务负载不大，因此被关闭时候转移任务的花费，对用户体验影响最小，试想一个运算资源如果正忙碌的执行用户交互相关的任务，又把它强制关闭逼迫它进行任务转移，那正被执行中的用户交互任务就因此受到影响，必须暂时停止，直到任务成功被转移到其它运算资源身上，才能够重新被执行。总而言之，最不影响用户体验又能帮助尽速散热降温，就是任务负载最轻，转移花费最小，被关闭之后因为静止不工作所以温度最低，那么依照“热往低温跑”的特性，附近聚集的热就能够往此低温方向消散出去了。
上述各个实施例中的相关技术特征可以相互组合，从而构成新的技术方案。上述方法中，提供了一种新的运算资源的启动优化方案和关闭优化方案，其包括“挑选新运算资源”。平常运算资源不使用的时候，都会进到关闭模式来降低耗电，开启新的运算资源就是将一关闭中的运算资源开启，使其能开始协同提供运算能力。那么上述方法在挑选新运算资源时，可以采用“选取最冷静也最不影响其他运算资源为优先对象”，最冷静可以是最久没有使用到的运算资源或者温度最低的运算资源，如此对象可以帮助在高负载情况下越晚达到效能被Thermal Management抑制的时间点。最不影响其他运算资源是指被开启对象一旦启动后，开始发热，但其发热情况对于其他运算资源的影响最小。还包括“挑选被关闭运算资源”，设计宗旨以在不影响使用者体验为前提下尽速散热降温，而不是强制牺牲效能来达到降温目的。关闭运算资源时，以最不影响使用者体验又能帮助尽速散热降温为优先对象。
基于上述方法中利用运算控制系统中多个运算资源的空间位置布局来 实现运算资源的开启管理时，还提供了可以利用运算控制系统中多个运算资源的空间位置布局来实现运算资源的关闭管理的一种运算资源的散热控制方法。上述方法中均是利用运算控制系统中多个运算资源的空间位置布局来实现运算资源的开启和关闭管理，两个方法是相反的两个过程，相辅相成的，不一定要同时执行，也可以分时在同一个执行过程中执行形成一种新的运算资源散热控制方法中。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品承载在一个非易失性计算机可读存储载体(如ROM、磁碟、光盘，服务器存储空间)中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，或者网络设备等)执行本发明各个实施例所述的系统结构和方法。
例如，在本发明的其中一个实施例中还提供了一种运算控制系统，应用于上述提到的具有多个运算资源布局的芯片，如图7所示，上述系统包括：
资源监控模块302，配置为基于预设的条件，确定是否需要开启或关闭运算资源；
散热管理模块309，配置为当需要开启运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源，将待开启的运算资源开启；还配置为当需要关闭运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照开启状态的运算资源之间的距离，来确定待关闭的运算资源，将待关闭的运算资源关闭。
在本发明的其中一些实施例中，散热管理模块309还配置为当需要开启运算资源或需要关闭运算资源时，确定当前开启状态的运算资源所属的运算组的类别；依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，在位于同一类别的运算组中从关闭状态的运算资源选取上述待开启的运算资源，或者，散热管理模块309还配置为依据同一类别的运算组中开启状态的运算资源与其他类别的运算组中开启状态的运算资源之间的距离，从开启状态的运算资源中选取所述待关闭的运算资源，或从开启状态的运算资源中的部分选取所述待关闭的运算资源，其中，基于运算资源的负载率、运算资源处于开启状态下的使用时长和运算资源的运行任务中的其中一个参量或两个以上参量的组合，在开启状态的运算资源中划定所述开启状态的运算资源中的部分。
在本发明的一个实施方式中，系统还包括位置确定模块301，配置为基于全部运算资源的物理布局，依据各个运算资源的当前运行状态，获取开启状态的运算资源和关闭状态的运算资源的位置，形成上述空间位置布局，从而提供给散热管理模块作为备选参考。上述位置确定模块301可以配置为执行上述步骤100中有关空间位置布局的形成方法过程，具体可参见前文中的相关说明。
在本发明的一个实施方式中，散热管理模块309还配置为确定关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，从关闭状态的运算资源中选出与每个开启状态的运算资源的距离最大者，如果上述距离最大者为同一个运算资源，则该运算资源为待开启的运算资源；或者，从关闭状态的运算资源中选出与每个开启状态的运算资源间隔至少一个关闭状态的运算资源的运算资源，则该运算资源为待开启的运算资源。
在本发明的一个实施方式中，散热管理模块309还配置为确定关闭状态的运算资源中与开启状态的运算资源之间的距离，从关闭状态的运算资 源中选出与每个开启状态的运算资源的距离最小者，如果上述距离最小者为同一个运算资源，则将该距离最小者从关闭状态的运算资源中排除选择。
在本发明的一个实施方式中，所述系统还包括：
第一定时器305，配置为记录上述关闭状态的运算资源处于关闭模式的关闭时长；
温度传感器304，配置为检测上述关闭状态的运算资源的当前温度；
散热管理模块309还配置为当上述待开启的运算资源包括多个备选运算资源时，查找满足当前温度最低和关闭时长最长这两个条件之一的备选运算资源，作为优先开启的运算资源。
在本发明的一个实施方式中，散热管理模块309还配置为确定当前开启状态的运算资源所属的运算组的类别，当同一类别的运算组存在至少两个运算资源时，确定上述至少两个运算资源与其他类别的运算组中当前正在运行的运算资源的距离，确定距离最小值所对应的运算资源，将该运算资源作为待关闭的运算资源。
在本发明的一个实施方式中，上述系统还包括：
第二定时器306，配置为记录上述开启状态的运算资源处于开启模式下的使用时长；
负载监控器307，配置为检测上述开启状态的运算资源的负载率；
任务进程管理器303，配置为监控上述开启状态的运算资源的运行任务；
散热管理模块309还配置为监控开启状态的运算资源的负载率、运算资源处于开启状态下的使用时长和开启状态的运算资源的运行任务中的其中一个参量或两个以上参量的组合；当依照开启状态的运算资源之间的距离来确定的待关闭的运算资源包括多个运算资源时，则优先选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理 的运行任务中的至少一个条件的运算资源，作为待关闭的运算资源；或者，
还配置为选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为待关闭的运算资源，当同一类别的运算组中存在至少两个待关闭的运算资源时，则优先选择所述至少两个待关闭的运算资源中与越多开启状态的运算资源相邻的运算资源，作为待关闭的运算资源。
资源监控模块302配置为执行上述步骤100，散热管理模块309配置为执行上述步骤200，因此有关具体细节可参见前文相关说明，在此不再累述。上述第一定时器和第二定时器可以是一个，也可以是多个。此外位置确定模块301和散热管理模块309可以通过额外添加的单个运算资源或多个运算资源来实现，也可以通过上述多个运算资源400中的其中一个或多个运算资源分时来实现。任务进程管理器303提供了有关计算机性能的信息，并显示了计算机上所运行的程序和进程的详细信息，例如，可以用于监控所有运算资源上所运行的程序的当前状态。任务进程管理器303提供了有关计算机性能的信息，并显示了计算机上所运行的程序和进程的详细信息，例如，可以用于监控所有运算资源上所运行的程序的当前状态。
在本发明的其中一些实施例中，负载监控器307可以采用电子设备的操作系统中动态加载的元件来实时地检测每个运算资源的负载率。这些元件可以是用于检测运算资源的负载率的驱动(drive)，该驱动可以是硬件也可以是软件集合。
在本发明的其中一些实施例中，上述系统还包括存储模块308。
散热管理模块309还配置为分别根据上述负载监控器307和温度传感器304、第一定时器、第二定时器、位置确定模块、任务进程管理器等中的一种或两种以上的模块同时检测到的多个运算资源400的温度和负载率等参量时，按照负载率、使用时长、关闭时长、当前温度等由高到低对每个 运算资源400进行降序排序，获取运算资源400及其相应的温度和负载率等参数的索引表，并将该索引表存储在存储模块308。散热管理模块309可以根据该索引表从运算资源中查找到合适的运算资源。这样便可以实时地监测每个运算资源的使用状况。
本发明实施例中，所述运算资源的控制装置在实际应用中，可通过个人计算机实现。所述运算资源的控制装置中的资源监控模块302、散热管理模块309、位置确定模块301、第一定时器305、第二定时器306、负载监控器307和任务进程管理器303，在实际应用中均可由所述装置中的中央处理器(CPU，Central Processing Unit)、数字信号处理器(DSP，Digital Signal Processor)、微控制单元(MCU，Microcontroller Unit)或可编程门阵列(FPGA，Field－Programmable Gate Array)实现；所述运算资源的控制装置中的存储模块308，在实际应用中可由所述装置中的存储器实现。
上述的运算资源的散热控制方法和运算控制系统，能够基于多个运算资源的空间位置布局，有效地选取运算资源(如运算资源)做开启和/或关闭的处理，让运算资源间热交互的影响降低，减缓因为高温度下传统散热管理强迫降低效能的情况，从而保持多运算资源结构的运算控制系统能够高效运转。
以上上述实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
在本申请所提供的几个实施例中，应该理解到，所揭露的设备和方法，可以通过其它的方式实现。以上所描述的设备实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，如：多个单元或组件可以结合，或可以集成到另一个系统，或一些特征可以忽略，或不执行。另外，所显示或讨论的各组成部分相互 之间的耦合、或直接耦合、或通信连接可以是通过一些接口，设备或单元的间接耦合或通信连接，可以是电性的、机械的或其它形式的。
上述作为分离部件说明的单元可以是、或也可以不是物理上分开的，作为单元显示的部件可以是、或也可以不是物理单元，即可以位于一个地方，也可以分布到多个网络单元上；可以根据实际的需要选择其中的部分或全部单元来实现本实施例方案的目的。
另外，在本发明各实施例中的各功能单元可以全部集成在一个处理单元中，也可以是各单元分别单独作为一个单元，也可以两个或两个以上单元集成在一个单元中；上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能单元的形式实现。
本领域普通技术人员可以理解：实现上述方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成，前述的程序可以存储于一计算机可读取存储介质中，该程序在执行时，执行包括上述方法实施例的步骤；而前述的存储介质包括：移动存储设备、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
或者，本发明上述集成的单元如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明实施例的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机、服务器、或者网络设备等)执行本发明各个实施例所述方法的全部或部分。而前述的存储介质包括：移动存储设备、ROM、RAM、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述，仅为本发明的具体实施方式，但本发明的保护范围并不局 限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应以所述权利要求的保护范围为准。
工业实用性
本发明实施例的技术方案能够基于多个处理器的空间位置布局，有效地选取运算资源(如处理器)做开启和/或关闭的处理，让运算资源间热交互的影响降低，减缓因为高温度下传统散热管理强迫降低效能的情况，从而保持多处理器结构的运算控制系统能够高效运转。

Claims (11)

  1. 一种运算资源的散热控制方法，该方法包括：
    基于预设的条件，确定是否需要开启或关闭运算资源；
    当需要开启运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源，将待开启的运算资源开启；
    当需要关闭运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照开启状态的运算资源之间的距离，来确定待关闭的运算资源，将待关闭的运算资源关闭。
  2. 根据权利要求1所述的方法，其中，所述当需要开启运算资源或需要关闭运算资源时，还包括以下步骤：
    确定当前开启状态的运算资源所属的运算组的类别；
    所述依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源的步骤包括：
    依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，在位于同一类别的运算组中从关闭状态的运算资源选取所述待开启的运算资源，或者，
    所述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤包括：
    依据同一类别的运算组中开启状态的运算资源与其他类别的运算组中开启状态的运算资源之间的距离，从开启状态的运算资源中选取所述待关闭的运算资源，或从开启状态的运算资源中的部分选取所述待关闭的运算资源，其中，基于运算资源的负载率、运算资源处于开启状态下的使用时长和运算资源的运行任务中的其中一个参量或两个以上参量的组合，在开启状态的运算资源中划定所述开启状态的运算资源中的部分。
  3. 根据权利要求1所述的方法，其中，所述依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，来确定待开启的运算资源的步骤包括：
    确定关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离，从关闭状态的运算资源中选出与每个开启状态的运算资源的距离最大者，如果所述距离最大者为同一个运算资源，则该运算资源为待开启的运算资源；或者，
    从关闭状态的运算资源中选出与每个开启状态的运算资源间隔至少一个关闭状态的运算资源的运算资源，则该运算资源为待开启的运算资源。
  4. 根据权利要求3所述的方法，其中，所述依照关闭状态的运算资源中每一个运算资源与开启状态的运算资源之间的距离来确定待开启的运算资源的步骤还包括：
    确定关闭状态的运算资源中与开启状态的运算资源之间的距离，从关闭状态的运算资源中选出与每个开启状态的运算资源的距离最小者，如果所述距离最小者为同一个运算资源，则将该距离最小者从关闭状态的运算资源中排除选择。
  5. 根据权利要求1或3所述的方法，其中，所述方法中，当所述待开启的运算资源包括多个备选运算资源时，查找满足当前温度最低和关闭时长最长这两个条件之一的备选运算资源，作为优先开启的运算资源。
  6. 根据权利要求1所述的方法，其中，所述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤中包括：
    确定当前开启状态的运算资源所属的运算组的类别，
    当同一类别的运算组存在至少两个开启状态的运算资源时，确定所述至少两个开启状态的运算资源与其他类别的运算组中当前正在运行的运算资源的距离，确定距离最小值所对应的运算资源，将该运算资源作为待关 闭的运算资源。
  7. 根据权利要求6所述的方法，其中，所述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤中还包括：
    当所述距离最小值所对应的运算资源为多个时，则在同一类别的运算组内，依据组内运算资源间的距离关系，优选组内与越多开启状态的运算资源相邻的运算资源，作为待关闭的运算资源。
  8. 根据权利要求1所述的方法，其中，所述依照开启状态的运算资源之间的距离来确定待关闭的运算资源的步骤包括：
    监控开启状态的运算资源的负载率、运算资源处于开启状态下的使用时长和开启状态的运算资源的运行任务中的其中一个参量或两个以上参量的组合；
    当依照开启状态的运算资源之间的距离来确定的待关闭的运算资源包括多个运算资源时，则选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为优先关闭的运算资源；或者，
    选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为待关闭的运算资源，当同一类别的运算组中存在至少两个待关闭的运算资源时，则选择所述至少两个待关闭的运算资源中与越多开启状态的运算资源相邻的运算资源，作为优先关闭的运算资源。
  9. 一种运算控制系统，包括：
    资源监控模块，配置为基于预设的条件，确定是否需要开启或关闭运算资源；
    散热管理模块，配置为当需要开启运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照关闭状态的运算资源中每一个运算 资源与开启状态的运算资源之间的距离，来确定待开启的运算资源，将待开启的运算资源开启；还配置为当需要关闭运算资源时，基于各个运算资源在当前运行状态下的空间位置布局，依照开启状态的运算资源之间的距离，来确定待关闭的运算资源，将待关闭的运算资源关闭。
  10. 根据权利要求9所述的系统，其中，所述系统还包括：
    第一定时器，配置为记录所述关闭状态运算资源处于关闭模式的关闭时长，
    温度传感器，配置为检测所述关闭状态运算资源的当前温度，
    散热管理模块还配置为当所述待开启的运算资源为多个运算资源时，查找满足当前温度最低和关闭时长最长这两个条件之一的距离最大者，作为待开启的运算资源；
    和/或所述系统还包括：
    第二定时器，配置为记录所述开启状态运算资源处于开启模式下的使用时长；
    负载监控器，配置为检测所述开启状态运算资源的负载率；
    任务进程管理器，配置为监控所述开启状态运算资源的运行任务；
    散热管理模块还配置为当依照开启状态的运算资源之间的距离来确定的待关闭的运算资源包括多个运算资源时，则选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为优先关闭的运算资源；或者，
    还配置为选择满足负载率最低、负载率在预设范围内、使用时长最长和未执行关于人机交互处理的运行任务中的至少一个条件的运算资源，作为待关闭的运算资源，当同一类别的运算组中存在至少两个待关闭的运算资源时，则选择所述至少两个待关闭的运算资源中与越多开启状态的运算资源相邻的运算资源，作为优先关闭的运算资源。
  11. 一种计算机存储介质，所述计算机存储介质中存储有计算机可执行指令，所述计算机可执行指令用于执行权利要求1至8任一项所述的运算资源的散热控制方法。
PCT/CN2016/087145 2015-07-28 2016-06-24 运算资源的散热控制方法、运算控制系统和存储介质 WO2017016357A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US15/747,508 US10488900B2 (en) 2015-07-28 2016-06-24 Heat dissipation control method for operation resources, operation control system, and storage medium
EP16829723.2A EP3330852A4 (en) 2015-07-28 2016-06-24 Heat dissipation control method for operation resources, operation control system, and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510452697.0 2015-07-28
CN201510452697.0A CN105045668A (zh) 2015-07-28 2015-07-28 运算资源的散热控制方法和运算控制系统

Publications (1)

Publication Number Publication Date
WO2017016357A1 true WO2017016357A1 (zh) 2017-02-02

Family

ID=54452232

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/087145 WO2017016357A1 (zh) 2015-07-28 2016-06-24 运算资源的散热控制方法、运算控制系统和存储介质

Country Status (4)

Country Link
US (1) US10488900B2 (zh)
EP (1) EP3330852A4 (zh)
CN (1) CN105045668A (zh)
WO (1) WO2017016357A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105045668A (zh) 2015-07-28 2015-11-11 深圳市万普拉斯科技有限公司 运算资源的散热控制方法和运算控制系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1344389A (zh) * 1998-12-23 2002-04-10 英特尔公司 管理计算机系统中的功耗的方法和设备
CN1602460A (zh) * 2001-12-12 2005-03-30 英特尔公司 操作系统的协调热管理
US7441135B1 (en) * 2008-01-14 2008-10-21 International Business Machines Corporation Adaptive dynamic buffering system for power management in server clusters
CN101556493A (zh) * 2008-04-09 2009-10-14 株式会社日立制作所 信息处理系统的运用管理方法及运用管理装置
EP2881839A1 (fr) * 2013-12-03 2015-06-10 Thales Procédé de gestion des ressources de calcul d'applications logicielles
CN105045668A (zh) * 2015-07-28 2015-11-11 深圳市万普拉斯科技有限公司 运算资源的散热控制方法和运算控制系统

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7379961B2 (en) * 1997-04-30 2008-05-27 Computer Associates Think, Inc. Spatialized audio in a three-dimensional computer-based scene
US20110021207A1 (en) * 2009-07-24 2011-01-27 Morgan Edward J System and Method for Estimating Positioning Error Within a WLAN-Based Positioning System
US8996330B2 (en) * 2011-01-06 2015-03-31 Qualcomm Incorporated Method and system for managing thermal policies of a portable computing device
JP5785050B2 (ja) * 2011-10-27 2015-09-24 株式会社日立製作所 情報処理システム、その省電力制御方法、及び装置
US9641629B2 (en) * 2011-11-28 2017-05-02 Lenovo (Singapore) Pte. Ltd. Distance-based network resource discovery
US9690334B2 (en) * 2012-08-22 2017-06-27 Intel Corporation Adaptive visual output based on change in distance of a mobile device to a user
TWI496087B (zh) * 2012-09-21 2015-08-11 Htc Corp 具多重中央處理單元之電子裝置及其效能管理方法
US9342135B2 (en) * 2013-10-11 2016-05-17 Qualcomm Incorporated Accelerated thermal mitigation for multi-core processors
CN104731740B (zh) * 2013-12-23 2017-11-10 联芯科技有限公司 终端中多核处理器热插拔控制方法和装置
US9219880B2 (en) * 2014-02-19 2015-12-22 International Business Machines Corporation Video conference window activator
US10119864B2 (en) * 2014-03-11 2018-11-06 Google Technology Holdings LLC Display viewing detection
CN104360893A (zh) * 2014-10-23 2015-02-18 西安未来国际信息股份有限公司 一种基于环境能耗的虚拟资源调度方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1344389A (zh) * 1998-12-23 2002-04-10 英特尔公司 管理计算机系统中的功耗的方法和设备
CN1602460A (zh) * 2001-12-12 2005-03-30 英特尔公司 操作系统的协调热管理
US7441135B1 (en) * 2008-01-14 2008-10-21 International Business Machines Corporation Adaptive dynamic buffering system for power management in server clusters
CN101556493A (zh) * 2008-04-09 2009-10-14 株式会社日立制作所 信息处理系统的运用管理方法及运用管理装置
EP2881839A1 (fr) * 2013-12-03 2015-06-10 Thales Procédé de gestion des ressources de calcul d'applications logicielles
CN105045668A (zh) * 2015-07-28 2015-11-11 深圳市万普拉斯科技有限公司 运算资源的散热控制方法和运算控制系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3330852A4 *

Also Published As

Publication number Publication date
US20180217645A1 (en) 2018-08-02
EP3330852A1 (en) 2018-06-06
US10488900B2 (en) 2019-11-26
EP3330852A4 (en) 2018-08-01
CN105045668A (zh) 2015-11-11

Similar Documents

Publication Publication Date Title
JP6005895B1 (ja) ワット当たりの最適パフォーマンスのためのインテリジェントマルチコア制御
Shirvani et al. A survey study on virtual machine migration and server consolidation techniques in DVFS-enabled cloud datacenter: taxonomy and challenges
CN106980492B (zh) 用于计算的装置、系统、方法、机器可读存储介质和设备
Enokido et al. An extended simple power consumption model for selecting a server to perform computation type processes in digital ecosystems
Fernández-Cerero et al. SCORE: Simulator for cloud optimization of resources and energy consumption
JP6249953B2 (ja) ヘテロジニアスマルチプロセッサシステムオンチップにおける熱駆動作業負荷スケジューリング
Ge et al. Distributed task migration for thermal management in many-core systems
US8381221B2 (en) Dynamic heat and power optimization of resource pools
CN102812439B (zh) 用于在多处理器计算机系统中分派任务的方法和系统
Townend et al. Improving data center efficiency through holistic scheduling in kubernetes
CN104049712B (zh) 用于计算机系统的功率管理方法和系统
EP2932812B1 (en) Energy conservation and hardware usage management for data centers
WO2017016371A1 (zh) 散热控制方法、装置和计算机存储介质
Piątek et al. Energy and thermal models for simulation of workload and resource management in computing systems
Kumar et al. Novel Dynamic Scaling Algorithm for Energy Efficient Cloud Computing.
Chaudhry et al. Thermal-aware relocation of servers in green data centers
WO2017016357A1 (zh) 运算资源的散热控制方法、运算控制系统和存储介质
US11188379B2 (en) Thermal capacity optimization for maximized single core performance
EP4163795A1 (en) Techniques for core-specific metrics collection
Chien et al. A thermal-aware scheduling for multicore architectures
Wu et al. Distributed thermal management for embedded heterogeneous MPSoCs with dedicated hardware accelerators
Chaudhry et al. Considering thermal-aware proactive and reactive scheduling and cooling for green data-centers
WO2017016359A1 (zh) 运算资源的控制方法、装置和计算机存储介质
Taneja et al. Thermal-aware job scheduling of mapreduce applications on high performance clusters
Patel et al. Energy-Aware VMs Consolidation Computing Frameworks’ of Data Center in Cloud Computing Environment

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16829723

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15747508

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE