CN104102532B - Research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group - Google Patents
Research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group Download PDFInfo
- Publication number
- CN104102532B CN104102532B CN201310129502.XA CN201310129502A CN104102532B CN 104102532 B CN104102532 B CN 104102532B CN 201310129502 A CN201310129502 A CN 201310129502A CN 104102532 B CN104102532 B CN 104102532B
- Authority
- CN
- China
- Prior art keywords
- task
- processor
- research
- energy consumption
- grouped
- 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.)
- Expired - Fee Related
Links
Classifications
-
- 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
Abstract
The present invention relates to the research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group, comprise the following steps：1) profile information needed for reading program；2) task scheduling sequence and calculating task parameter are obtained；3) task in task scheduling sequence is grouped, obtains grouped task result；4) by the duty mapping in each packet to processor；5) task of the search with free time in each processor, and the power-saving mechanism for selecting amount of energy saving most performs the task.Compared with prior art, the present invention is fully organically combined dynamic voltage regulation technology with network energy-saving method, reduces the total energy consumption of group system to the full extent while systematic function is not influenceed.
Description
Technical field
The present invention relates to Parallel Scheduling technical field, more particularly, to the science work based on low energy consumption in a kind of isomeric group
Make stream scheduling method.
Background technology
Energy consumption problem has become a global problem, with the fast development of computer and network technologies, current high
Power managed Yin Qigao during performance is calculated runs expense, low reliability, is increasingly becoming focus of concern.Traditional grinds
How study carefully emphasis and lay particular emphasis on always makes system have optimum performance, i.e., be reasonably dispatched to task on each processor to obtain
Most short total runtime.But in the last few years, increasingly serious with global energy problem, numerous researchers are just progressively
In terms of the energy consumption that research emphasis has been transferred to cluster computing system.
With the continuous expansion of the increasingly complicated and calculation scale of problem in science, scientific workflow (Scientific
Workflow) provide a kind of automatic flow management platform auxiliary scientist and carry out scientific research and Knowledge Discovery.Operation is adjusted
Degree be in research-on-research Flow Technique one it is extremely important the problem of, dispatching algorithm directly affects the performance of research-on-research streaming system.
At the same time, research-on-research streaming system needs high performance computing resource and storage resource, performs extensive scientific application program
Consume huge energy.
Scientific workflow dispatches the scheduling phase with traditional directed acyclic graph (Directed Acyclic Graph, DAG)
Than existing similar part, while there is also some differences.Job scheduling is that research-on-research flow structure is analyzed, to scheduling
The performance of each resource is estimated in environment, and under conditions of scientific workflow operation constraint and user's constraint is met, will be made
Industry is reasonably assigned to the process of each resource.Substantial amounts of research tissue is conducted in-depth research to this problem both at home and abroad, pin
A variety of dispatching algorithms are proposed respectively to different application scenarios, also have some research tissue degree of exchanging strategies and algorithm to carry out
Summarize.
Dynamic voltage regulation (Dynamic Volgage Scaling, DVS) technology is that one kind is adjusted by rational method
The power-saving mechanism of service voltage, has proven to one of design low-power dissipation system most efficient method.Its basic thought be
Processor is not influenceed dynamically to adjust computing device voltage and frequency in the case of normally running so that processor is not always with most
High voltage operation, so as to play the purpose of reduction processor energy consumption.Overwhelming majority processor is all manufactured with CMOS technology at present, and
And support various processor frequency and voltage to set.The power consumption of cmos circuit is proportional to square of clock frequency and voltage, i.e., every
The energy expenditure of individual clock cycle is proportional to square of voltage.For a task, the clock cycle required for it is completed
It is fixed, square being directly proportional for the energy consumed and voltage, energy consumption can be just reduced by reducing voltage.Although clock frequency
Linear relationship between rate and voltage, reduction voltage can turn down clock frequency, increase the deadline of task, but can be using appointing
The free time reduction processor voltage or frequency of business, system energy consumption is reduced on the premise of other tasks carryings are not influenceed.
During voltage-regulation, voltage management chip is able to carry out small voltage adjustment, and can (tens is micro- within the extremely short time
Second in) complete voltage adjustment.Dynamic voltage regulation (DVS) technology has proven to design the maximally effective side of low-power dissipation system
One of method, and turned into a kind of power-saving mechanism of comparative maturity.In the last few years, many researchers were by the success of DVS mechanism
Ground has been applied among the PC cluster perceived based on energy consumption.Although dynamic voltage regulation technology is in efficiency cluster is built
Huge contribution is achieved, but it focuses on the energy consumption of processor in saving cluster.
It is network system high energy consumption, poorly efficient with the expansion of network size in group system and the continuous renewal of the network facilities
The problems such as rate, is increasingly notable.The utilization rate of the network equipment now is all very low, and due to the energy of nearly all network equipment
Consumption is all determined by peak bandwidth, and most equipment is whole day full speed operation, and the network user really needs most high band
The 5% of wide working time deficiency equipment operation, therefore, in the case of network idle, the network equipment is still according to peak
It is worth the standard of bandwidth in consumed energy, this has resulted in the significant wastage of the energy.Therefore, GreenNet is built, network energy is reduced
Consumption has become one key issue significant, in the urgent need to address of current computer network field.
Conventional dispatching method largely for performance or energy consumption certain part (system energy consumption include processor energy consumption and
Network energy consumption two parts) improve or optimized.The weak point of presence has：1. the method having only considers performance and neglected completely
Omit energy consumption；Although the method 2. having both had considered performance it is contemplated that the saving of processor energy consumption, have ignored network service energy
Consumption, even if or consider communication energy consumption, but network delay is not taken into account.3. although some algorithms have also contemplated that net
Network communication energy consumption, but do not have using network energy-saving mechanism and strategy further to save network service energy consumption.
The content of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide base in a kind of isomeric group
In the research-on-research stream scheduling method of low energy consumption, this method is fully organic by dynamic voltage regulation technology and network energy-saving method
Ground is combined, and reduces the total energy consumption of group system to the full extent while systematic function is not influenceed.
The purpose of the present invention can be achieved through the following technical solutions：
The research-on-research stream scheduling method based on low energy consumption, comprises the following steps in a kind of isomeric group：
1) profile information needed for reading program；
2) task scheduling sequence and calculating task parameter are obtained；
3) task in task scheduling sequence is grouped, obtains grouped task result；
4) by the duty mapping in each packet to processor；
5) task of the search with free time in each processor, and select the most power-saving mechanism of amount of energy saving to perform
The task.
Step 1) include following two steps：
Step 1.1, reading scientific workflow profile information；
Step 1.2, reading parameter configuration files information, include processor parameter and net-work parameter information.
Step 2) include following two steps：
Step 2.1, acquisition task scheduling sequence；
Step 2.2, calculating task parameter, including earliest start time, earliest finish time, optimum processor, it is optimal before
Drive, the deadline is allowed at the latest and allows the time started at the latest.
Step 3) comprise the following steps：
Step 3.1, since first task in task scheduling sequence, perform depth-first search until task knot
Beam；
Step 3.2, in task search procedure, if the optimal forerunner of current task is not grouped, assign them to work as
Packet where preceding task, and be labeled as having distributed；
Step 3.3, repeat step 3.2, are finished until all tasks are all grouped.
Step 4) comprise the following steps；
Step 4.1, since first be grouped, by all duty mappings in each packet to a unappropriated place
Manage on device, and mark the processor occupied；
Step 4.2, repeat step 4.1, until each packet is allocated untill on processor.
Step 5) comprise the following steps：
Step 5.1, one processor of search, obtain the task on the processor with free time；
Step 5.2, judge whether the task can be saved by DVS mechanism or network speed zoom mechanism, if can
To be saved by any one mechanism, then the most mechanism of amount of energy saving is selected to perform；If being only capable of by one of which mechanism
Saved, then select the mechanism to perform；
Step 5.3, repeat step 5.1 and step 5.2, until all tasks on each processor have been searched
Untill finishing.
Compared with prior art, the present invention is directed to the energy-optimised problem that scientific workflow is dispatched in isomeric group, by DVS
Technology and network energy-saving technology combine, and system energy is optimized while system resource limitation and system performance requirements are met
Consumption.First, replication-based scheduling algorithm (mainly includes processor parameter and network connection is joined according to system environments in the present invention
Number) obtain near-optimization packet；Then, each packet is dispatched on a unappropriated processor respectively, and by exploring
The free time produced by dependence between task, adjust processor voltage to save processor using DVS Technique dynamics
Idle energy consumption, or network service energy consumption is optimized using network energy-saving mechanism and strategy.
Brief description of the drawings
Fig. 1 is flow chart of the invention.
Embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.
Embodiment
The invention provides the research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group.Wherein, science work
Make stream and represent that the modeling method of scientific workflow is the more commonly used modeling method in scheduling research using directed acyclic graph.
In the present invention, the set of tasks of scientific workflow can be expressed as by the directed acyclic graph of a sideband weight G (V, E, t,
C), wherein V={ v1, v2..., vnTask-set is represented, E represents the set i.e. message set of all directed edges, and t represents task at place
Manage the calculating time on device, e represents the dependence between task, c represents the call duration time between two tasks for the weight on side.By
In the present invention is directed isomeric group, so the specific calculating time for each task of different types of processor is to differ
Sample, but execution time on same type processor is essentially equal.We use t hereiniwExpression task viIn processing
Device pwOn the calculating time, use eij={ vi, vj∈ E are represented from task viIt is transferred to vjMessage, i.e. viIt is vjPredecessor task,
And the time needed for transmitting the message uses cijRepresent, and a task only is carried out finishing it in its all predecessor tasks
After can start to be performed.In real network, call duration time cijMessage is represented from being sent to the time interval that is received, it
Mainly it is made up of message transmission time and message delay.M and n in above-mentioned model represent the processor used in cluster respectively
Task number in number and task-set, i, j, k represents mission number, and w, p, q represents that processor is numbered, therefore 1≤i,
J, k≤n, 1≤p, q, w≤m.
Computing system in one isomeric group by several there is the processor node of different disposal ability to constitute, and be designated as
P={ p1, p2..., pm, these nodes are connected with each other by express network.Using identical network for connecting in the present invention
It is isomorphism to connect each processor node, i.e. network connection.All processors in system support DVS technologies, it is assumed that processor
piThere is hiIndividual discrete voltage, can be expressed asHereArranged for non-ascending order,
Frequency corresponding to itIt is also such.The number of all types of processor discrete voltages is by itself voltage
Determine, the discrete voltage number of same type processor is identical.
This method specifically includes several steps as shown in Figure 1：
First, the profile information needed for reading program
Step 1.1, reading scientific workflow profile information；
Step 1.2, reading parameter configuration files information, include processor parameter and net-work parameter information.
2nd, task scheduling sequence and calculating task parameter are obtained
Step 2.1, acquisition task scheduling sequence queue；For parallel task collection V, n=is had | V | individual task, from outlet
Task starts, and calculates the priority of each task until entrance task terminates.Then arranged according to task priority size ascending order,
Obtain initiating task schedule sequences.
Our defined notion bottom are come the maximum of total evaluation time that is expressed as from current task to export task herein
Value, this value is to meet the priority constraint relationship between task.Consider when calculating bottom according to the original calculation time, and not
The intertask communication time.Task vjBottom values be defined as follows：
Wherein succ (vj) represent task vjSuccessor set, tjExpression task vjMaximum execution on all processors
Time.After the bottom values of all tasks are obtained, adjusted with regard to an initiating task arranged according to bottom values ascending order can be obtained
Degree series queue.
Step 2.2, calculating task parameter, including earliest start time (EST), earliest finish time (ECT), optimization process
Device (fproc), optimal forerunner (FP) allows the deadline (LACT) at the latest, and the time started (LAST) is allowed at the latest.
Task vjOptimum processor can according to following formula calculate obtain：
Task vjEarliest start time (EST) on each processor can recursive calculation, wherein entrance from top to bottom
The earliest start time of task is 0.Calculation formula is：
Otherwise (3)
Task vjEarliest finish time (ECT) be equal to task viActual earliest start time add task vjAt it most
The execution time on good processor, it is shown below.
ECT (j)=EST (j)+tjfproc(j). (4)
Task vjOptimal forerunner be defined as follows shown in formula：
Task vjThe permission deadline at the latest：
Otherwise (6)
Task vjThe permission time started at the latest：
LAST (j)=LACT (j)-tjfproc(j) (7)
First, DAG figures are traveled through from top to down, calculating task vjIn processor pwOn earliest start time EST
(j/w).Then, it can be deduced that task vjEarliest start time EST (j) in its optimum processor, and calculating task is most
Early deadline ECT (j).After the above parameter of current task is calculated, we can draw current task vjMost
Good processor fproc (j).Next, calculating task vjOptimal forerunner FP (j), the parameter means task and its optimal
Forerunner, which is dispatched on identical processor, can obtain smaller parallel time；Finally, calculating task vjAllow at the latest complete
Time LACT (j) and at the latest permission time started LAST (j).
3rd, group result is obtained
Step 3.1, since first task in task scheduling sequence queue, perform depth-first search until appoint
Business terminates；
During step 3.2, task search, if current task viOptimal forerunner (FP (i)) be not grouped, then by its point
Packet where dispensing current task, and be labeled as having distributed；
Step 3.3, repeat step 3.2, are finished until all tasks are all grouped.
4th, grouped task maps
Step 4.1, since first be grouped, by all duty mappings in each packet to a unappropriated place
Manage on device, and mark the processor occupied；
Step 4.2, repeat step 4.1, untill each packet is allocated on a processor.
5th, Energy Saving Strategy is selected
There is the task of free time in step 5.1, one processor of search；
Step 5.2, the task for one with free time, judge whether the task can be by DVS mechanism or net
Network speed zoom mechanism is saved, if can be saved by any one mechanism, the most mechanism of selection amount of energy saving
Perform；If being only capable of being saved by one of which mechanism, the mechanism is selected to perform.
Detailed process is：First determine whether whether the task can be saved by DVS mechanism, if it is then needing to judge this
Whether task can be saved by network speed zoom mechanism, if it is then calculating the amount of energy saving and fortune with DVS mechanism respectively
With the amount of energy saving of network speed zoom mechanism, choose the most power-saving mechanism of amount of energy saving and perform.
If the task can not by network speed zoom mechanism is saved or select be utilize dynamic voltage regulation machine
It is processed to be saved, then using DVS mechanism on the premise of other task execution times are not influenceed by the tasks carrying frequency and electricity
Pressure is reduced to near-optimization value；If the task can be by the way that network speed zoom mechanism is saved and select is to utilize network
Speed zoom mechanism is saved, then network energy-saving mechanism is scaled using speed, before other task execution times are not influenceed
Put and the message transmission rate between the task and its subsequent tasks is reduced to near-optimization value.
When search one have free time task viAfterwards, v is calculated firstiWhen without task power-economizing method
Calculate energy consumption compEN1 [vi] and communication energy consumption commEN1 [v between the unallocated subsequent tasks in same processori]
[vj], then calculate v using DVS technologiesiPerform the calculating energy consumption compEN2 [v after DVS energy-conservationsi], scaled using network speed
Mechanism calculates viPerform the communication energy consumption commEN2 [v after network energy-savingi][vj]；CompEN3 [v are made hereini]=compEN1
[vi]-compEN2[vi], represent the amount of energy saving for performing DVS technologies；Make commEN3 [vi][vj]=commEN1 [vi][vj]-
commEN2[vi][vj], represent to perform the amount of energy saving of network speed scaling strategy；Then compEN3 [v are judgedi] whether be more than
commEN3[vi][vj].If compEN3 [vi] it is more than commEN3 [vi][vj], then explanation performs energy-conservation ratio after DVS and performs net
Network speed scaling energy-conservation is more, therefore is saved using DVS technologies, and otherwise the energy-saving effect of explanation execution network speed scaling is excellent
In DVS technologies, therefore perform network speed scaling strategy.The specific formula for calculation and computational methods of related energy consumption are as follows：
(1) energy consumption is calculated
At present, the circuit in processor uses CMOS integrated circuits mostly, its energy expenditure mainly include static energy consumption and
Dynamic energy consumption, but by static energy consumption, to put ratio in processor total energy consumption smaller, so dynamic energy consumption is ground frequently as energy-conservation
The main object studied carefully, therefore the present invention only considers the dynamic energy consumption of processor when building and calculating energy consumption model.
The present invention is using DVS technologies reduction processor energy consumption, because processor has several different voltages and frequency water
It is flat, therefore dispatching algorithm can fully select most suitable execution voltage and frequency to save.DVS mechanism is by adaptable frequency
Rate and voltage are assigned on a series of processors, to reach the free time for eliminating tasks carrying so that the mesh for saving idle energy consumption
's.The power consumption of processor can be expressed as following formula with given parameter：
Wherein, α represents switching rate, CLRepresent load capacitance, VddRepresent supply voltage, fclkRepresent clock frequency.Herein
It is envisaged that how to select most suitable computing device frequency on the premise of systematic function is not influenceed.Dynamic electric voltage is adjusted
The object of section must be the task with free time on the non-critical path in DAG, for being present in critical path
Task, its execution time can not be changed.DVS technologies of the present invention are holding the task with free time
Some near-optimization value of line frequency/voltage Tiao Jiedao.If a task available free time, it will will be somebody's turn to do using DVS technologies
The execution frequency of task is reduced near optimal and performs frequency, otherwise the task one is scheduled under highest frequency and voltage and performed.
Then task viIn processor pwOn activity energy consumption calculation such as following formula：
Therefore, all processors total activity energy consumptions is calculated as follows formula：
All processors total idle energy consumptions is calculated as follows formula：
Wherein xikMapping relations between expression task and processor, are defined as follows：
Finally, total calculating energy consumption can be calculated by following formula：
EN=ENactive+ENidle (13)
PN in above formulahighest-wAnd PNidle-wProcessor p is represented respectivelykHighest power consumption and idle power consumption,
fhighest-wAnd Vhighest-wProcessor p is represented respectivelywOn highest perform frequency and perform voltage, fiwAnd ViwBiao Shi not task vi
In processor pwOn optimal execution frequency and perform voltage, τiwExpression task viIn processor pwOn actual execution time,
Makespan represents the time i.e. scheduling length that is finished of workflow application.
(2) communication energy consumption
The communication energy consumption of network is broadly divided into two parts, and a part is the energy consumption when maintenance network equipment works in itself,
This is referred to as the static energy consumption of equipment, and another part is the data communication energy consumption of network equipment working time, referred to herein as equipment
Communication energy consumption.We do not consider the static energy consumption of the network equipment in the present invention, and the emphasis of network energy consumption research is in the network equipment
Data communication energy consumption during work, therefore, the network energy consumption model that the present invention is used are led to primarily directed to the data of the network equipment
Believe what energy consumption was set up.
In the present invention, interference networks are isomorphisms, i.e., network all has identical data transmission capabilities and power of communications.
The present invention is done to communication energy consumption using continuous stream speed zoom model and reasonably evaluates and save.Data in the model are with stable
And continuous form is transmitted, the energy consumed in network can be adjusted by the data traffic in route network, can also
Adjusted by reducing the transmission speed on each link, we take the latter in the present invention.
Assuming that any link e corresponds to an energy consumption function fe(s), it represents the link with speed s transmission data when institute
The energy of consumption.Each link can select its actual biography according to the transport load situation of network in [0, R] rate period
Defeated speed s, 0≤s≤R, are special circumstances as R=∞.R is bandwidth in following energy consumption model.
The network energy consumption function model that the present invention is used is as follows：
This is a kind of function model with Start-up costs, and wherein e represents any one network linking, σeRepresent energy consumption letter
Several Start-up costs.Relative to the energy consumption function model without Start-up costs, the function model with Start-up costs more meets existing
Real network environment and actual conditions.
Call duration time can be ignored between due to being assigned to two tasks of same processing node, therefore whole net
Total communication energy consumption calculation formula that the communication energy consumption of network is assigned on different processor between task is as follows：
Wherein, parameter inijIt is defined as follows formula：
Finally, the total energy consumption of whole system can be expressed as：
Etotal=EN+EL (17)
Step 5.3, repeat step 5.1 and step 5.2, until all tasks on each processor have been searched
Untill finishing.
Claims (5)
1. the research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group, it is characterised in that comprise the following steps：
1) profile information needed for reading program；
2) task scheduling sequence and calculating task parameter are obtained；
3) task in task scheduling sequence is grouped, obtains grouped task result；
4) by the duty mapping in each packet to processor；
5) task of the search with free time in each processor, and select the most power-saving mechanism of amount of energy saving to perform this
Business；
Step 5) comprise the following steps：
Step 5.1, one processor of search, obtain the task on the processor with free time；
Step 5.2, judge whether the task can be saved by DVS mechanism or network speed zoom mechanism, if can lead to
Cross any one mechanism to be saved, then select the most mechanism of amount of energy saving to perform；If being only capable of carrying out by one of which mechanism
Energy-conservation, then select the mechanism to perform；
Step 5.3, repeat step 5.1 and step 5.2, until all tasks on each processor be searched finishing for
Only.
2. the research-on-research stream scheduling method based on low energy consumption, its feature in a kind of isomeric group according to claim 1
It is, step 1) include following two steps：
Step 1.1, reading scientific workflow profile information；
Step 1.2, reading parameter configuration files information, include processor parameter and net-work parameter information.
3. the research-on-research stream scheduling method based on low energy consumption, its feature in a kind of isomeric group according to claim 1
It is, step 2) include following two steps：
Step 2.1, acquisition task scheduling sequence；
Step 2.2, calculating task parameter, including earliest start time, earliest finish time, optimum processor, optimal forerunner, most
Allow the deadline late and allow the time started at the latest.
4. the research-on-research stream scheduling method based on low energy consumption, its feature in a kind of isomeric group according to claim 1
It is, step 3) comprise the following steps：
Step 3.1, since first task in task scheduling sequence, perform depth-first search and terminate until task；
Step 3.2, in task search procedure, if the optimal forerunner of current task is not grouped, assign them to when predecessor
Packet where business, and be labeled as having distributed；
Step 3.3, repeat step 3.2, are finished until all tasks are all grouped.
5. the research-on-research stream scheduling method based on low energy consumption, its feature in a kind of isomeric group according to claim 1
It is, step 4) comprise the following steps：
Step 4.1, since first be grouped, by all duty mappings in each packet to a unappropriated processor
On, and mark the processor occupied；
Step 4.2, repeat step 4.1, until each packet is allocated untill on processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310129502.XA CN104102532B (en) | 2013-04-15 | 2013-04-15 | Research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310129502.XA CN104102532B (en) | 2013-04-15 | 2013-04-15 | Research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104102532A CN104102532A (en) | 2014-10-15 |
| CN104102532B true CN104102532B (en) | 2017-09-26 |
Family
ID=51670704
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310129502.XA Expired - Fee Related CN104102532B (en) | 2013-04-15 | 2013-04-15 | Research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104102532B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107302562B (en) * | 2017-05-23 | 2019-12-03 | 中国科学院计算技术研究所 | A kind of the adaptive command processing system and method for internet-of-things terminal equipment |
| CN107391247B (en) * | 2017-07-21 | 2020-06-26 | 同济大学 | Breadth-first greedy mapping method for network-on-chip application |
| CN109901442B (en) * | 2017-12-08 | 2020-09-15 | 亿可能源科技（上海）有限公司 | Non-invasive energy consumption detection method and system |
| CN108416465B (en) * | 2018-01-31 | 2021-08-31 | 杭州电子科技大学 | Workflow optimization method in mobile cloud environment |
| CN108647084B (en) * | 2018-05-08 | 2021-07-20 | 武汉轻工大学 | Energy efficiency cloud task scheduling method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102231122A (en) * | 2011-07-13 | 2011-11-02 | 武汉理工大学 | Clustering-based energy-saving scheduling method in cluster environment |
| CN102360246A (en) * | 2011-10-14 | 2012-02-22 | 武汉理工大学 | Self-adaptive threshold-based energy-saving scheduling method in heterogeneous distributed system |
| CN102650957A (en) * | 2012-04-09 | 2012-08-29 | 武汉理工大学 | Self-adaptive energy-saving dispatching method in isomorphic cluster system based on dynamic voltage regulation technology |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI326818B (en) * | 2006-12-15 | 2010-07-01 | Inst Information Industry | Dynamic voltage scheduling method, system thereof and record medium |
-
2013
- 2013-04-15 CN CN201310129502.XA patent/CN104102532B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102231122A (en) * | 2011-07-13 | 2011-11-02 | 武汉理工大学 | Clustering-based energy-saving scheduling method in cluster environment |
| CN102360246A (en) * | 2011-10-14 | 2012-02-22 | 武汉理工大学 | Self-adaptive threshold-based energy-saving scheduling method in heterogeneous distributed system |
| CN102650957A (en) * | 2012-04-09 | 2012-08-29 | 武汉理工大学 | Self-adaptive energy-saving dispatching method in isomorphic cluster system based on dynamic voltage regulation technology |
Non-Patent Citations (1)
| Title |
|---|
| 网络能耗系统模型及能效算法;张法等;《计算机学报》;20120331;第35卷(第3期);603-607 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104102532A (en) | 2014-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104102532B (en) | Research-on-research stream scheduling method based on low energy consumption in a kind of isomeric group | |
| CN102193826B (en) | Method for high-efficiency task scheduling of heterogeneous multi-core processor | |
| Li et al. | Energy optimization with dynamic task scheduling mobile cloud computing | |
| Quan et al. | Task scheduling for energy consumption constrained parallel applications on heterogeneous computing systems | |
| CN102360246B (en) | Self-adaptive threshold-based energy-saving scheduling method in heterogeneous distributed system | |
| CN106844051A (en) | The loading commissions migration algorithm of optimised power consumption in a kind of edge calculations environment | |
| CN103399626B (en) | Towards Parallel application dispatching system and the method for the power-aware of hybrid compute environment | |
| CN102929720B (en) | A kind of energy-conservation job scheduling system | |
| Gu et al. | Energy efficient scheduling of servers with multi-sleep modes for cloud data center | |
| CN105446816B (en) | A kind of energy optimization dispatching method towards heterogeneous platform | |
| US11435802B2 (en) | Work load scheduling for multi core systems with under-provisioned power delivery | |
| CN103076870A (en) | Application fusing scheduling and resource dynamic configuring method of energy consumption drive in data center | |
| CN107168770A (en) | A kind of cloud data center workflow schedule of low energy consumption and resource provision method | |
| CN102749987B (en) | High energy efficiency resource allocating method for isomorphic cluster system of computer | |
| Banerjee et al. | Towards a net-zero data center | |
| Rajabi et al. | Communication-aware and energy-efficient resource provisioning for real-time cloud services | |
| Huang et al. | Dynamic allocation/reallocation of dark cores in many-core systems for improved system performance | |
| Liu et al. | An energy efficient clustering-based scheduling algorithm for parallel tasks on homogeneous DVS-enabled clusters | |
| CN102902878B (en) | A kind of energy cost perception dispatching method | |
| Ansari et al. | Power-aware scheduling of fixed priority tasks in soft real-time multicore systems | |
| CN109144693A (en) | A kind of power adaptive method for scheduling task and system | |
| CN104298536A (en) | Dynamic frequency modulation and pressure adjustment technology based data center energy-saving dispatching method | |
| Ghonoodi | Green Energy-aware task scheduling using the DVFS technique in Cloud Computing | |
| Ma et al. | Energy-optimization scheduling of task dependent graph on DVS-enabled cluster system | |
| CN104699520B (en) | A kind of power-economizing method based on virtual machine (vm) migration scheduling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170926 Termination date: 20200415 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |