CN109324880A - A kind of low-power consumption scheduling method suitable for real-time system periodic task model - Google Patents

A kind of low-power consumption scheduling method suitable for real-time system periodic task model Download PDF

Info

Publication number
CN109324880A
CN109324880A CN201710641332.1A CN201710641332A CN109324880A CN 109324880 A CN109324880 A CN 109324880A CN 201710641332 A CN201710641332 A CN 201710641332A CN 109324880 A CN109324880 A CN 109324880A
Authority
CN
China
Prior art keywords
task
periodic duty
moment
time
rem
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201710641332.1A
Other languages
Chinese (zh)
Inventor
郭锐锋
吴昊天
胡毅
彭阿珍
邓昌义
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenyang Gaojing Numerical Control Intelligent Technology Co Ltd
Original Assignee
Shenyang Gaojing Numerical Control Intelligent Technology Co Ltd
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 Shenyang Gaojing Numerical Control Intelligent Technology Co Ltd filed Critical Shenyang Gaojing Numerical Control Intelligent Technology Co Ltd
Priority to CN201710641332.1A priority Critical patent/CN109324880A/en
Publication of CN109324880A publication Critical patent/CN109324880A/en
Pending legal-status Critical Current

Links

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
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • 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
    • G06F1/329—Power saving characterised by the action undertaken by task scheduling
    • 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
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G06F9/4893—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues taking into account power or heat criteria
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00—Indexing scheme relating to G06F9/00
    • G06F2209/48—Indexing scheme relating to G06F9/48
    • G06F2209/484—Precedence
    • 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 a kind of low-power consumption scheduling methods suitable for real-time system periodic task model, and its step are as follows: the overall utilization U of calculating cycle tasktot, and offline speed S is calculated with thisof=max { Smin, Utot}；Periodic duty collection is ranked up according to earliest-deadline-first principle EDF；I-th of periodic duty TiWhen t moment is ready, T is calculatediIn moment t using execution time remi(t), original using the free time for executing the time, high-priority task completes the free time generated and low priority task generates including its, and recalculate TiSpeed Si=wi(t)/remi(t).Work as SiWhen less than critical speed, is judged whether using balance factor method using critical speed and whether close processor and reduce energy consumption.This method can reduce as far as possible processor energy consumption under the premise of ensureing periodic duty schedulability, improve system global reliability.

Description

A kind of low-power consumption scheduling method suitable for real-time system periodic task model
Technical field
The present invention relates to the Real-Time Schedulings of real-time system periodic task model, specifically a kind of to be suitable for system in real time The low-power consumption scheduling method of system periodic task model.
Background technique
As very large scale integration technology develops, the energy consumption of processor is also increasing.Real-time system is task There is the scheduling system of hard constraints in deadline, and the increase of power consumption of processing unit certainly will lead to unstability, therefore place is effectively reduced Managing device energy consumption is real-time scheduling's problem in need of consideration.Real-time system scheduling is divided into periodic task and accidental task, Middle periodic task model is a kind of important task model in real-time system, and feature is task instances in periodically arrival.Mesh The preceding dispatching algorithm for periodic task collection mainly uses DVS technology, and the actual execution time of task is often below it most The time is executed in bad situation, to generate free time.DVS technology can recycle free time, distribute in ready queue not The task of completion reduces its speed to reduce energy consumption.The power consumption of processor is divided into three parts at present: quiescent dissipation Ps, with speed Spend unrelated power consumption PindWith P relevant to speeddep, PindMainly from leakage current power consumption, PdepMainly from dynamic power consumption. The dynamic power consumption P that processor is run with speed SdepIt is represented by Pdep=Cef·Sm, wherein CefThe load capacitance in circuit is represented, S is the speed of service of processor, and m is the ordinary index unrelated with power consumption of processing unit (2≤m≤3).Therefore, the total power consumption of processor It can indicate are as follows: P=Ps+h(Pind+Pdep)=Ps+h(Pind+Cef·Sm), wherein coefficient h is constant, works as h=1, indicates to appoint Business is being run；Otherwise h=0.Processor dynamic power consumption and speed at about quadratic relation, dynamic power consumption due to speed reduces substantially Degree reduces.But the actual execution time that low velocity certainly will will lead to task extends, to increase quiescent dissipation.To guarantee system Total power consumption is optimal, and existing research indicates so that the minimum critical speed of system energy consumption, it is indicated that above or below critical speed, place The power consumption of reason device will increase, wherein critical speed
It is assumed that processor is capable of providing continuous speed, and speed is normalized, so that the speed that processor provides Degree range is [Smin, 1], wherein SminFor the minimum operation speed of processor.There are three states for processor: closed state, active State, idle state do not consume energy when processor is in close state.According to DPM technology, it is contemplated that close processor Required expense, enables E0Energy consumption expense, P needed for being switched to closed state for idle stateidleFor the power consumption of idle state, then Close the time overhead of processorIt is greater than t between when idle0, and do not have to can choose closing processing when ready task Device.
Summary of the invention
DPM technology and critical speed is largely all utilized in the existing research to periodic task collection low-power consumption scheduling It reduces energy consumption, i.e., when the speed of traditional DVS policy calculation is less than critical speed, task is run using critical speed.But it closes Key speed is also not absolute and makes the smallest speed of energy consumption.Using there are one between traditional scheduler strategy or critical speed Equalization point determines processor speed compared with equalization point according to the time is executed under task worst case.
Present invention technical solution used for the above purpose is: one kind being suitable for real-time system periodic task model Low-power consumption scheduling method, comprising:
Step 1: before periodic duty collection starts scheduling, the overall utilization U of calculating cycle task-settot, the offline speed of processor Sof；The periodic duty integrates as T (T1,T2,T3,…Ti…Tn), it altogether include n periodic duty；
Task ready queue, task arrival queue are set on real-time scheduler；The task ready queue includes ready Set of tasks RD (Ti, t), wherein t indicates moment, TiIndicate i-th of periodic duty, and ready task set is according to the excellent of task First grade height arranges；The task arrival queue includes having run through complete but next example and not arriving also for task；
Step 2: i-th of periodic duty Ti(pi,ci) when t moment reaches, according to earliest deadline EDF priority By periodic duty Ti(pi,ci) be inserted into task ready queue, and initial setting up remi(t)=wi(t)=ci；Wherein, piIt is Ti's Period, ciIt is TiWorst case under execute time, remiIt (t) is periodic duty TiIn moment t using execution time, wi It (t) is periodic duty TiThe time is executed under the remaining worst case of moment t；
Step 3: i-th of periodic duty Ti(pi,ci) executed in t moment scheduling, recalculate TiIn the available of moment t Execute time remi(t), T is recalculatediExecution speed Si=wi(t)/remi(t)；Work as Si<ScritWhen, judge using ScritOr Still keep SiAs execution speed；Wherein, ScritFor critical speed；
Step 4: if t moment, TiIt is not seized by other ready tasks in operational process, then as periodic duty TiWhen execution remi(t) and wi(t) it gradually decreases, if TiIt is finished, then sets wi(t)=0, TiTask ready queue is removed, is added and extremely appoints Business reaches queue；By piIncrease to riAs next periodic duty TiArrival time；Return step 2 continues to execute periodic duty, Until periodic duty is finished；
If t moment, TiIt is seized in operational process by other ready tasks, then retains remi(t),wi(t), by periodic duty TiAgain ready queue is added by EDF priority orders, return step 2 continues to execute periodic duty, until periodic duty executes It finishes.
The calculation formula of the overall utilization of the periodic duty collection isThe meter of the offline speed of processor Calculation formula is Sof=max { Smin,Utot}；Wherein, n is the number of periodic duty in task-set, and i-th of periodic duty is defined as Ti (pi,ci), SminThe minimum speed that can be provided for processor.
I-th of periodic duty T in the step 2i(pi,ci) when t moment reaches, it is excellent according to earliest deadline EDF First sequentially by periodic duty Ti(pi,ci) be inserted into task ready queue to dynamically distribute task according to the deadline of task Priority；Deadline is more forward, and priority is higher；Wherein, periodic duty TiCut-off time limit diCalculation formula is di=ri+ pi, wherein riIt is periodic duty TiThe arrival time of this time.
Periodic duty T is recalculated in the step 3iIn moment t using execution time remi(t) are as follows: by SLH (Ti, t) and SLL(Ti, t) the sum of increase to remi(t)；
Wherein, than periodic duty TiThe high task of priority fulfils generated free time summation ahead of schedule and is expressed as SLH (Ti,t)；Calculation formula isWherein, PH (Ti, t) and it is to be completed before t moment Priority ratio periodic duty TiHigh set of tasks；remhIt (t) is task ThIn t moment using the execution time；SLH(Ti, T) be to belong to set PH (T to alli, t) task ThRemh(t) it sums；Wherein, than periodic duty TiPriority is low Free time summation caused by task is SLL(Ti,t)；
Define moment tf=t+SLH(Ti,t)+remi(t)；If in moment tfReady queue is sky, then defines tf' for Moment tfThe arrival time of recent task later；TβFor in moment tfThe task for the highest priority that do not complete before, remβ(t) and wβIt (t) is respectively periodic duty TβThe time is executed in the case where moment t is using execution time and remaining worst case；
Define PL′(Tβ, t) and={ Tx|Tx∈PL(Ti, t) and dx≤dβ, wherein PL(Ti, t) and it indicates before moment t The priority ratio periodic duty T of completioniLow set of tasks, dxAnd dβRespectively indicate TxAnd TβWith the cut-off time limit；
If in [tf,tf+SLL(Ti, t)] and [tf′,tf′+SLL(Ti, t)] there are other ready tasks T in the periodλ, Then SLL(Ti, t) calculation formula be SLL(Ti, t) and=rλ-tf；Otherwise, SLL(Ti, t) calculation formula be Wherein, rλIt is TλArrival time, remxIt (t) is TxIn t moment Using execute the time；
Work as S in the step 3i<ScritWhen, judge using ScritOr still keep SiAs execution speed, comprising:
It is assumed that periodic duty TiIt executes in interval Δ t without being preempted, if there is wi(t)<Scrit* Δ t and Then use the energy consumption E of critical speed1Are as follows:Using biography The energy consumption E of system DVS strategy2Are as follows:Set discriminant function F (x)=E2-E1And By wi(t) it is substituted for x, then
Δ t=rem is enabled in t momenti(t), x=wi(t) method of discrimination is as follows:
If Pind*Δt-E0> 0, then in section (0, Scrit* there must be a point w in Δ t)0So that F (w0)=0, the point are flat Weigh the factor, as F (wi(t)) when > 0, using ScritAs execution speed, otherwise keep SiAs execution speed；
If Pind*Δt-E0≤ 0, keep SiAs execution speed；
Wherein, α is load capacitance parameter, and S is the execution speed of processor, PindIt is quiescent dissipation, E0To close processor Expense.
The invention has the following beneficial effects and advantage:
1. making full use of system in guarantee system under the premise of periodic duty schedulability using the method for the present invention Free time reduces the number for closing processor.
2. using the method for the present invention, in guarantee system under the premise of periodic duty schedulability, system is sufficiently reduced Energy consumption guarantees system stability.
Detailed description of the invention
Fig. 1 is the derivative figure of discriminant function；
The simulation experiment result figure of the variation to energy consumption that Fig. 2 is task WCET/BCET in the present invention；
Fig. 3 is the simulation experiment result figure of the utilization rate variation to energy consumption of system in the present invention.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and embodiments.
Periodic duty is one kind of real-time task, and feature is divided into fixed between the release of two continuous task instances Constant.This model considers that there are the hard real-time systems of n mutually independent periodic duties, using earliest-deadline-first strategy (EDF) periodic duty collection T is dispatched.Each periodic duty TiWith binary group (pi,ci) indicate, piFor TiPeriod, ciFor Ti The opposite cut-off time limit of execution time in the worst cases, task are equal to its period.Here r is usediAnd diExpression task TiRelease Time and cut-off time limit are put, ac is usediAnd Ti,jRespectively indicate TiActual execution time and its j-th of task instances, use UtotIt indicates The utilization rate of entire task-set,
Before periodic duty collection starts scheduling, the overall utilization U of calculating cycle tasktot, determine that the offline speed of processor is Sof=max { Smin,Utot}。
On real-time scheduler be arranged two queues, one be task ready queue (ready_queue), the other is The arrival queue (delay_queue) of task.Ready_queue includes ready set of tasks, high according to the priority of task Low arrangement, definition set RD (Ti, t) be t moment ready queue in set of tasks.Delay_queue includes to have run through Finish but next example not arriving for task also.
I-th of periodic duty Ti(pi,ci) when t moment reaches, ready team is inserted into according to earliest-deadline-first principle In column, and initial setting up remi(t)=wi(t)=ci, remiIt (t) is task TiIn moment t using execution time, wi(t) It is task TiThe time is executed under the remaining worst case of moment t.
If Ti(pi,ci), then task T ready in t momentiIt is made of in the moment t available time 3 parts: comparing TiPreferentially The high task of grade fulfils generated free time summation SL ahead of scheduleH(Ti,t)；Task TiIt is original using execute the time；Than TiFree time summation SL caused by the low task of priorityL(Ti,t).Computation-free time and place as follows Manage device speed:
1. calculatingWherein PH (Ti, t) and it is completed priority ratio Ti High set of tasks, remh(t) T is indicatedhIn moment t using the execution time.
2. defining tf=t+SLH(Ti,t)+remi(t), if in moment tfReady queue is sky, then defines tf' in moment tfIt Recent task arrival time afterwards.TβFor in moment tfThe task for the highest priority that do not complete before, remβ(t) and wβ(t) Respectively TβExecute time, P under time and remaining worst case using executingL(Ti, t) and indicate completed before moment t Priority ratio task TiLow set of tasks, then define PL′(Tβ, t) and={ Tx|Tx∈PL(Ti,t)and dx≤dβ, wherein dxAnd dβ It is T respectivelyxAnd TβThe opposite cut-off time limit, calculate If [tf,tf+SLL(Ti, t)] and [tf′,tf′+SLL(Ti, t)] there is other ready tasks T in the periodλ, then SL is calculatedL(Ti, T)=rλ-tf。
3. by SLH(Ti, t) and SLL(Ti, t) the sum of increase to remi(t), T is calculatediSpeed Si=wi(t)/remi(t)。
4. differentiating SiIt whether is so that TiThe optimal speed of energy consumption.The power consumption expression formula of processor is P=α * S3+Pind, wherein S It is the execution speed of processor, PindIt is quiescent dissipation.According to ScritExecution task, when task is completed ahead of time, if free time is greater than Close processor expense t0, need to close processor.According to XScale power consumption of processing unit model simplification above formula, α=1.52, Pind= 0.08, close the expense E of processor0=0.8, Scrit0.3 is taken, then P=1.52*S3+ 0.08, wherein S is processor speed.It is assumed that Ti It executes in interval Δ t without being preempted.Task T at this timeiIt is w that the time is executed under remaining worst casei(t), work as wi(t)<Scrit* Δ t, andWhen, use the energy consumption of critical speed for+ 0.8=0.4*wi(t)+0.8, the energy consumption for using traditional DVS strategy to generate is E2=1.52*S2*wi(t)+0.08* Δ t is enabled
When (0,0.3 Δ t) is the derivative figure of discriminant function to x ∈ as shown in Figure 1, and F ' (x) < 0, F (x) are in (0,0.3 Δ t) Upper monotone decreasing, (0.3 Δ t)=- 0.8 < 0 F.
Δ t=rem is enabled in t momenti(t), x=wi(t) method of discrimination is as follows:
If 0.08* Δ t-0.8 > 0, as F (wi(t)) when > 0, S is seti=Scrit, otherwise keep SiIt is constant；
If 0.08* Δ t-0.8≤0, SiIt is constant.
As task TiWhen execution, remi(t) and wi(t) it gradually decreases, if TiIt completes to execute, then sets wi(t)=0, TiAdd Enter delay_queue.By piIncrease to ri, r after increaseiAs next periodic duty TiRelease time.
If t moment, TiIt is seized in operational process by other ready tasks, retains remi(t),wi(t), by TiAgain EDF is pressed Ready queue is added in priority orders.
Fig. 2 is the variation of task WCET/BCET in the present invention to the simulation experiment result figure of energy consumption, and wherein WCET is Execute the time under the worst case of task, BCET be task best-case under execute the time.Algorithm is compared as existing DRA calculation Method and DSTRA algorithm and LPABOBF algorithm proposed by the present invention.Abscissa is WCET/BCET, and ordinate is after normalizing Energy consumption.When WCET/BCET is gradually increased, the average value of the actual execution time of task will be reduced, the average sky of task-set Just gradually increase between idle, the energy consumption of three is all gradually reduced.It can be seen that the performance of LPABOBF algorithm be better than always DRA and DSTR A algorithm, the calculation shows that, LPABOBF ratio DRA algorithm saves about 18.33%~28.57% energy in this case Consumption saves about 7.3%~18.03% energy consumption than DSTRA.
Fig. 3 is the simulation experiment result figure of the utilization rate variation to energy consumption of system in the present invention.It is existing for comparing algorithm Some DRA algorithms and DSTRA algorithm and LPABOBF algorithm proposed by the present invention.Abscissa is utilization rate, and ordinate is to return Energy consumption after one change.With the increase of utilization rate, the processor time in operating status, the energy consumption of three kinds of algorithms was equal than increasing It is gradually increased.It can be seen that the opposite the two of the energy-saving effect of LPABOBF algorithm is preferably.When utilization rate is smaller, processor is always State in closing, therefore three's energy consumption is close, but has some superiority using DSTRA the and LPABOBF algorithm of DPM technology. With the increase of utilization rate, to make a choice in traditional DVS strategy or critical speed strategy on many periods, LPABOBF Algorithm gradually plays effect.When utilization rate is close to 1, the case where load down, processor speed is less than critical speed, is less, LPABOBF algorithm is close with DSTRA algorithm energy-saving effect, is computed, and LPABOWSA ratio DRA algorithm saving 8.9%~ 26.19% energy consumption, than the energy consumption that DSTRA algorithm saves about 2.7%~13.98%.

Claims (5)

1. a kind of low-power consumption scheduling method suitable for real-time system periodic task model characterized by comprising
Step 1: before periodic duty collection starts scheduling, the overall utilization U of calculating cycle task-settot, the offline speed S of processorof； The periodic duty integrates as T (T1,T2,T3,…Ti…Tn), it altogether include n periodic duty；
Task ready queue, task arrival queue are set on real-time scheduler；The task ready queue includes ready task Set RD (Ti, t), wherein t indicates moment, TiIndicate i-th of periodic duty, and ready task set is according to the priority of task Height arranges；The task arrival queue includes having run through complete but next example and not arriving also for task；
Step 2: i-th of periodic duty Ti(pi,ci) when t moment reaches, it will be all according to earliest deadline EDF priority Phase task Ti(pi,ci) be inserted into task ready queue, and initial setting up remi(t)=wi(t)=ci；Wherein, piIt is TiWeek Phase, ciIt is TiWorst case under execute time, remiIt (t) is periodic duty TiIn moment t using execution time, wi(t) It is periodic duty TiThe time is executed under the remaining worst case of moment t；
Step 3: i-th of periodic duty Ti(pi,ci) executed in t moment scheduling, recalculate TiWhen moment t is using executing Between remi(t), T is recalculatediExecution speed Si=wi(t)/remi(t)；Work as Si<ScritWhen, judge using ScritOr it still keeps SiAs execution speed；Wherein, ScritFor critical speed；
Step 4: if t moment, TiIt is not seized by other ready tasks in operational process, then as periodic duty TiRem when executioni(t) And wi(t) it gradually decreases, if TiIt is finished, then sets wi(t)=0, TiTask ready queue is removed, is added to task and reaches Queue；By piIncrease to riAs next periodic duty TiArrival time；Return step 2 continues to execute periodic duty, Zhi Daozhou Phase task execution finishes；
If t moment, TiIt is seized in operational process by other ready tasks, then retains remi(t),wi(t), by periodic duty TiAgain Ready queue is added by EDF priority orders, return step 2 continues to execute periodic duty, until periodic duty is finished.
2. according to a kind of low-power consumption scheduling method suitable for real-time system periodic task model, feature described in claim 1 It is, the calculation formula of the overall utilization of the periodic duty collection isThe calculating of the offline speed of processor is public Formula is Sof=max { Smin,Utot}；Wherein, n is the number of periodic duty in task-set, and i-th of periodic duty is defined as Ti(pi, ci), SminThe minimum speed that can be provided for processor.
3. according to a kind of low-power consumption scheduling method suitable for real-time system periodic task model, feature described in claim 1 It is, i-th of periodic duty T in the step 2i(pi,ci) when t moment reaches, it is preferential according to earliest deadline EDF Sequentially by periodic duty Ti(pi,ci) be inserted into task ready queue to dynamically distribute the excellent of task according to the deadline of task First grade；Deadline is more forward, and priority is higher；Wherein, periodic duty TiCut-off time limit diCalculation formula is di=ri+pi, Wherein riIt is periodic duty TiThe arrival time of this time.
4. according to a kind of low-power consumption scheduling method suitable for real-time system periodic task model, feature described in claim 1 It is, recalculates periodic duty T in the step 3iIn moment t using execution time remi(t) are as follows: by SLH(Ti, And SL t)L(Ti, t) the sum of increase to remi(t)；
Wherein, than periodic duty TiThe high task of priority fulfils generated free time summation ahead of schedule and is expressed as SLH(Ti, t)；Calculation formula isWherein, PH (Ti, t) and it is completed excellent before t moment First grade is than periodic duty TiHigh set of tasks；remhIt (t) is task ThIn t moment using the execution time；SLH(Ti, t) It is to belong to set PH (T to alli, t) task ThRemh(t) it sums；Wherein, than periodic duty TiThe low task of priority Generated free time summation is SLL(Ti,t)；
Define moment tf=t+SLH(Ti,t)+remi(t)；If in moment tfReady queue is sky, then defines tf' in moment tf The arrival time of recent task later；TβFor in moment tfThe task for the highest priority that do not complete before, remβ (t) and wβIt (t) is respectively periodic duty TβThe time is executed in the case where moment t is using execution time and remaining worst case；
Define PL′(Tβ, t) and={ Tx|Tx∈PL(Ti, t) and dx≤dβ, wherein PL(Ti, t) and indicate completed before moment t Priority ratio periodic duty TiLow set of tasks, dxAnd dβRespectively indicate TxAnd TβWith the cut-off time limit；
If in [tf,tf+SLL(Ti, t)] and [tf′,tf′+SLL(Ti, t)] there are other ready tasks T in the periodλ, then SLL (Ti, t) calculation formula be SLL(Ti, t) and=rλ-tf；Otherwise, SLL(Ti, t) calculation formula be Wherein, rλIt is TλArrival time, remxIt (t) is TxIn t moment Using execute the time.
5. according to a kind of low-power consumption scheduling method suitable for real-time system periodic task model, feature described in claim 1 It is, works as S in the step 3i<ScritWhen, judge using ScritOr still keep SiAs execution speed, comprising:
It is assumed that periodic duty TiIt executes in interval Δ t without being preempted, if there is wi(t)<Scrit* Δ t and Then use the energy consumption E of critical speed1Are as follows:Using traditional DVS strategy Energy consumption E2Are as follows:Set discriminant function F (x)=E2-E1And by wi(t) it is substituted for X, then
Δ t=rem is enabled in t momenti(t), x=wi(t) method of discrimination is as follows:
If Pind*Δt-E0> 0, then in section (0, Scrit* there must be a point w in Δ t)0So that F (w0)=0, the point be balance because Son, as F (wi(t)) when > 0, using ScritAs execution speed, otherwise keep SiAs execution speed；
If Pind*Δt-E0≤ 0, keep SiAs execution speed；
Wherein, α is load capacitance parameter, and S is the execution speed of processor, PindIt is quiescent dissipation, E0To close opening for processor Pin.
CN201710641332.1A 2017-07-31 2017-07-31 A kind of low-power consumption scheduling method suitable for real-time system periodic task model Pending CN109324880A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710641332.1A CN109324880A (en) 2017-07-31 2017-07-31 A kind of low-power consumption scheduling method suitable for real-time system periodic task model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710641332.1A CN109324880A (en) 2017-07-31 2017-07-31 A kind of low-power consumption scheduling method suitable for real-time system periodic task model

Publications (1)

Publication Number Publication Date
CN109324880A true CN109324880A (en) 2019-02-12

Family

ID=65245008

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710641332.1A Pending CN109324880A (en) 2017-07-31 2017-07-31 A kind of low-power consumption scheduling method suitable for real-time system periodic task model

Country Status (1)

Country Link
CN (1) CN109324880A (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110837291A (en) * 2019-10-28 2020-02-25 华侨大学 Energy consumption optimization method based on actual execution time mixed key cycle task
CN111324197A (en) * 2020-02-07 2020-06-23 华侨大学 Method for reducing system energy consumption based on three-speed periodic task
CN111597030A (en) * 2020-05-21 2020-08-28 华侨大学 Adaptive factor energy consumption optimization method based on task attributes
CN111736549A (en) * 2020-06-11 2020-10-02 上海申瑞继保电气有限公司 Production line energy-saving auxiliary equipment control method
CN111813553A (en) * 2020-07-17 2020-10-23 华侨大学 Low-energy-consumption method based on selectable factor period task dynamic priority
CN112905330A (en) * 2021-03-25 2021-06-04 华侨大学 Fixed priority mixed key accidental task energy consumption sensing method
CN114578944A (en) * 2022-02-21 2022-06-03 华侨大学 Method and system for updating energy consumption perception of mixed key tasks by using fixed priority utilization rate
CN114936076A (en) * 2022-04-26 2022-08-23 中国人民解放军国防科技大学 Real-time scheduling method and device for mixed task set and computer equipment

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张忆文: "面向数控系统低功耗调度算法研究", 《中国博士学位论文全文数据库（信息科技辑（月刊））》 *
邓昌义、郭锐锋、张忆文、王鸿亮: "基于平衡因子的动态偶发任务低功耗调度算法", 《吉林大学学报(工学版)》 *

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110837291A (en) * 2019-10-28 2020-02-25 华侨大学 Energy consumption optimization method based on actual execution time mixed key cycle task
CN110837291B (en) * 2019-10-28 2023-03-28 华侨大学 Energy consumption optimization method based on actual execution time mixed key cycle task
CN111324197B (en) * 2020-02-07 2023-03-07 华侨大学 Method for reducing system energy consumption based on three-speed periodic task
CN111324197A (en) * 2020-02-07 2020-06-23 华侨大学 Method for reducing system energy consumption based on three-speed periodic task
CN111597030A (en) * 2020-05-21 2020-08-28 华侨大学 Adaptive factor energy consumption optimization method based on task attributes
CN111597030B (en) * 2020-05-21 2023-03-24 华侨大学 Adaptive factor energy consumption optimization method based on task attributes
CN111736549A (en) * 2020-06-11 2020-10-02 上海申瑞继保电气有限公司 Production line energy-saving auxiliary equipment control method
CN111813553A (en) * 2020-07-17 2020-10-23 华侨大学 Low-energy-consumption method based on selectable factor period task dynamic priority
CN111813553B (en) * 2020-07-17 2023-05-26 华侨大学 Task dynamic priority low-energy consumption method based on selectable factor period
CN112905330A (en) * 2021-03-25 2021-06-04 华侨大学 Fixed priority mixed key accidental task energy consumption sensing method
CN112905330B (en) * 2021-03-25 2023-05-26 华侨大学 Fixed priority hybrid key sporadic task energy consumption sensing method
CN114578944A (en) * 2022-02-21 2022-06-03 华侨大学 Method and system for updating energy consumption perception of mixed key tasks by using fixed priority utilization rate
CN114578944B (en) * 2022-02-21 2023-06-20 华侨大学 Method and system for sensing energy consumption of mixed critical tasks by updating utilization rate of fixed priority
CN114936076A (en) * 2022-04-26 2022-08-23 中国人民解放军国防科技大学 Real-time scheduling method and device for mixed task set and computer equipment

Similar Documents

Publication Publication Date Title
CN109324880A (en) A kind of low-power consumption scheduling method suitable for real-time system periodic task model
CN101290585B (en) Embedded system real time task scheduling method
CN102193826B (en) Method for high-efficiency task scheduling of heterogeneous multi-core processor
CN103473139B (en) Cluster virtual machine resource allocation scheduling method
CN106445070B (en) Energy consumption optimization scheduling method for hard real-time system resource-limited sporadic tasks
CN109739332B (en) Multi-task general energy consumption optimization method
CN103324525A (en) Task scheduling method in cloud computing environment
CN103823718A (en) Resource allocation method oriented to green cloud computing
CN103823545B (en) A kind of server awakening method of cloud computing system
CN108984292A (en) Mix critical system fixed priority periodic duty energy consumption optimization method
CN106970835B (en) Hierarchical energy consumption optimization method for fixed priority resource-limited system
Li et al. Energy-efficient scheduling in nonpreemptive systems with real-time constraints
CN109324891A (en) A kind of periodic duty low-power consumption scheduling method of ratio free time distribution
Yoo et al. Integrated scheduling of real-time and interactive tasks for configurable industrial systems
CN113535356B (en) Energy-aware hierarchical task scheduling method and device
CN108574600B (en) Service quality guarantee method for power consumption and resource competition cooperative control of cloud computing server
CN106802822A (en) A kind of cloud data center cognitive resources dispatching method based on moth algorithm
CN105630126B (en) One kind is based on normal bandwidth server hybrid task low-power consumption scheduling method
CN107329817B (en) Standby standby system mixed partitioning reliability perception energy consumption optimization method
CN104182280B (en) Low-energy RM real-time task scheduling method for hybrid main memory embedded system
CN104182180B (en) Low-energy EDF (earliest deadline first) real-time task scheduling method for mixed main memory embedded system
CN106293007B (en) A kind of energy-saving scheduling method for supporting non-preemption real-time task collection
CN108563497B (en) Efficient multi-dimensional algorithm scheduling method and task server
Jiang et al. PB-FCFS-a task scheduling algorithm based on FCFS and backfilling strategy for grid computing
CN106933325B (en) A kind of fixed priority I/O device energy consumption management method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20190212