CN104572266A - MPSoC task scheduling, modeling and assessing method under process variations on basis of UPPAAL-SMC - Google Patents
MPSoC task scheduling, modeling and assessing method under process variations on basis of UPPAAL-SMC Download PDFInfo
- Publication number
- CN104572266A CN104572266A CN201510005475.4A CN201510005475A CN104572266A CN 104572266 A CN104572266 A CN 104572266A CN 201510005475 A CN201510005475 A CN 201510005475A CN 104572266 A CN104572266 A CN 104572266A
- Authority
- CN
- China
- Prior art keywords
- task
- mpsoc
- scheduling
- model
- power consumption
- 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.)
- Granted
Links
Abstract
The invention discloses an MPSoC task scheduling, modeling and assessing method under process variations on the basis of UPPAAL-SMC. The MPSoC task scheduling, modeling and assessing method includes the following steps: generating a task allocation and scheduling example for a task set according to task scheduling strategy; conducting modeling on tasks, PE and power consumption on the premise of considering process variations to form a task model, a PE model and a power consumption model, and conducting background allocation to perform model transformation on the task allocation and scheduling example by aid of a list scheduling and the branch and cut algorithm; achieving constraint query by aid of stochastic modeling, generating statistic data through UPPAAL-SMC and getting the performance yield of the task allocation and scheduling example; comparing different performance yield of the task allocation and scheduling examples generated by different task scheduling strategies of the same MPSoC framework or different MPSoC frameworks, and assessing the task scheduling strategies to determine the MPSoC framework with superior performance. MPSoC designers can select between TAS strategies and the MPSoC frameworks with help of the MPSoC task scheduling, modeling and assessing method so as to obtain the optimum combination.
Description
Technical field
The present invention relates to a kind of multinuclear SOC (system on a chip) task scheduling analytical model method under process variation, particularly relate to a kind of task scheduling analytical model method with process variation MPSoC based on UPPAAL-SMC (static model inspection UPPAAL).
Background technology
Along with the high speed development of science and technology, embedded system is just playing more and more important effect, and the SOC (system on a chip) (System-on-a-chip, SoC) of rising gradually in recent years is just widely applied in the field of embedded system.People require that embedded device has more function, higher performance, and multinuclear SOC (system on a chip) (Multiprocessor System-on-a-chip, MPSoC) produces thus.
Process variation (Process Variation) is transistor properties (long, wide, gate oxide thickness etc.) abiogenous change in ic manufacturing process, when process node is less than 65nm, affects obvious all the more.This also becomes one of bottleneck of integrated circuit (IC) chip performance boost.Even the product that same flow waterline is produced, also can there is small deviation in parameter, and these deviations affecting significantly by producing the performance of MPSoC and power consumption, the performance of MPSoC not exclusively being conformed to expection, thus causes the unpredictability of behavior further.
Task matching and scheduling (Task Allocation and Scheduling, TAS) is a np hard problem.Because the performance existing for MPSoC of process variation and power consumption bring larger change, when considering the factor of process variation, traditional the worst temporal analysis method being used for producing feasible solution is no longer applicable.For the deviser of MPSoC, be difficult to determine under specific constraint condition, which kind of task scheduling strategy is better.In order to ensure the performance output (Performance Yield) of dispatching algorithm, an important problem is just become to the assessment of task scheduling algorithm.
Therefore, build the appraisal procedure of MPSoC task scheduling algorithm under a set of process variation, be conducive to MPSoC devisers and numerous scheduling strategy is contrasted, thus carry out the strategy assessing to select to be applicable to.
Summary of the invention
The object of the invention is the blank of task scheduling strategy assessment aspect in the multinuclear SOC (system on a chip) in order to make up current consideration process variation, provide a kind of MPSoC task scheduling analytical model method based on UPPAAL-SMC under process variation, achieve the process automatically task scheduling strategy being converted to model, and can provide there is comparative result, thus the assessment realized different task scheduling strategy and contrast.
Based on the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, comprise the following steps:
Step one: according to task scheduling strategy, for task-set generates Task matching and scheduling example;
Step 2: under the prerequisite considering process variation, modeling is carried out to task, PE and power consumption, form task model, PE model and power consumption model, and carry out the backstage configuration of model, described Task matching and scheduling example existing ListScheduling (list scheduling) and BULB (branch's cutting) algorithm is carried out model conversion;
Step 3: realize constraint inquiry by stochastic model, and generate statistics by UPPAAL-SMC (static model inspection UPPAAL), draw the performance output of described Task matching and scheduling example;
Step 4: the described performance output of the difference described Task matching and scheduling example that described task scheduling strategies different under contrasting same MPSoC framework or different MPSoC framework generates, assess described task scheduling strategy, determine to show more excellent described MPSoC framework.
Based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, described task modeling comprises the following steps:
Steps A 1: run initialization function, initialization is carried out to the process variation of time and power consumption;
Steps A 2: if task does not exist predecessor node, then perform next step; Otherwise, wait-receiving mode to described predecessor node complete message after, then perform next step;
Steps A 3: configure according to described backstage, by task matching to corresponding PE;
Steps A 4: after described PE finishes the work, if do not have follow-up node, then terminates to run; Otherwise, continue to run described subsequent node.
Preferably, based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, described task modeling does not carry out modeling to whole task sequence, but all task nodes are carried out abstract in unification, modeling is carried out to single task role node, and according to forerunner, the configuring condition of descendant node, by synchronizing signal, different tasks is coupled together, perform according to the order of sequence, each task is responsible for and forerunner, the communication of descendant node, and the distribution to processing unit, and the time that task completes is managed by described PE model, during start node configures according to the described backstage of reading, time and power consumption process variation information are as parameter, generate the random number of Normal Distribution, initialization is carried out to the process variation of time and power consumption.
Based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, described PE modeling comprises the following steps:
Step B1: the information recording the task that will perform, determines the execution time of described task actual needs, and upgrades the power consumption information of described power consumption model according to process variation;
Step B2: after waiting for that the described execution time terminates, discharges power consumption shared in described power consumption model, and upgrades described power consumption information.
Based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, described PE model comprises the loop of multiple monitoring new task message, efficiently avoid the situation that described PE model does not receive new task message, achieve functionally complete and semantically perfect.
Preferably, based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, described PE model is the abstract model to all PE, is carried out the performance of control PE, comprise power consumption, process variation etc. by the configuration of described backstage.Described PE model by communicating between described task model with described power consumption model, the implementation of control task and the renewal of power consumption information.
Based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, described power consumption model carries out monitoring to power consumption and safeguards; Described power consumption model comprises treatment state and waiting status, and switches between described treatment state and described waiting status, in the process switched, monitor power loss signal.
Based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, the configuration of described backstage comprises: the application background of the correlation parameter of MPSoC, design constraint, described Task matching and scheduling example and at least one task DAG.
Preferably, based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, the configuration of described backstage adds the information of the process variation of time and power consumption of processing unit, as the configuration setting input of framework.
Based in the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, the form of described constraint inquiry is as follows:
Pr[<＝x](<>Task(0).Finish&&max_power<＝y)；
In formula, Task (0) .Finish represents: completing of whole task directed acyclic graph; Max_power<=y represents: maximum power dissipation can not more than y.
Preferably, of the present invention based in the MPSoC task scheduling analytical model method of UPPAAL-SMC, as long as the result that dispatching algorithm generates comprises this three partial information of id of task id, start time, PE, framework can be converted into and can be configured by the described backstage of Model Identification, after crossover tool uses the precedence relationship of transfer algorithm determination task, write in the backstage configuration of described UPPAAL-SMC.
Preferably, after constraint inquiry completes, described UPPAAL-SMC provides the result of operation in the mode of chart or data, by changing the content of inquiry, repeatedly running described UPPAAL-SMC draws under different demands, the performance situation of each described Task matching and scheduling example, thus select at task matching and between scheduling strategy and the framework of MPSoC, to obtain best combination.
Beneficial effect based on the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes is as follows: the present invention can reflect the performance output of Task matching and scheduling example under process variations exactly, help the scheduling strategy can selecting during MPSoC design to be applicable to, or adjust according to the design of scheduling strategy to MPSoC.
Accompanying drawing explanation
Fig. 1 is Task matching and scheduling Policy evaluation frame diagram of the present invention.
Fig. 2 is the process flow diagram of the MPSoC task scheduling analytical model method that the present invention is based on UPPAAL-SMC.
Fig. 3 is the schematic diagram to task modeling in the present invention.
Fig. 4 is the schematic diagram to PE modeling in the present invention.
Fig. 5 is the schematic diagram to power consumption modeling in the present invention.
Embodiment
In conjunction with following specific embodiments and the drawings, the present invention is described in further detail.Implement process of the present invention, condition, test method etc., except the following content mentioned specially, be universal knowledege and the common practise of this area, the present invention is not particularly limited content.
The present invention proposes a kind of MPSoC task scheduling analytical model method based on UPPAAL-SMC under process variation, see Fig. 1, the frame diagram of Task matching and scheduling Policy evaluation in the present invention.
Framework of the present invention receives configuration setting as input, and a Task matching and scheduling example determined, these parameters are applied in model, and repeatedly run query statement, the final stability of this Task matching and scheduling example under the environment of input parameter that generates shows (i.e. performance output).Wherein, framework reception configuration setting comprises:
1) several task DAG (Directed Acyclic Graph, directed acyclic graph): if DAG is more than 1, then multiple DAG must be mapped on a Task matching and scheduling example when dispatching, and namely all task DAG perform in same scheduling;
2) correlation parameter of MPSoC: comprise the number of processing unit (PE), type, rated disspation and process variation interval, these parameters are relevant to MPSoC.Wherein, process variation interval is the function of a normal distribution, and framework can operationally automatic therefrom selective value, and this value is as the number percent of MPSoC relative to rated power, and user should specify its variance;
3) design constraint: comprise the deadline of all tasks, the maximum power dissipation of MPSoC, and the performance output etc. needing realization.
Statistical model checks that (Statistical Model Checking, SMC) technology produces a large amount of test findings by the random walk of model, and sets up statistical conclusions for specific inquiry constraint.
Based on the MPSoC task scheduling analytical model method of UPPAAL-SMC under the process variation that the present invention proposes, for given scheduling strategy, first with the example of this strategy generating task scheduling, then this instance transfer is become the input parameter of framework; Framework accepts these input parameters, can be converted into the configuration file in UPPAAL-SMC model, and run in a model; Again by given constraint inquiry, the performance output of this scheduling strategy under given design constraint just can be drawn.
As shown in Figure 2, the MPSoC task scheduling analytical model method based on UPPAAL-SMC of the present invention comprises: step one: Task matching and scheduling example generation step; Step 2: model conversion step; Step 3: query generation step; Step 4: the analysis of Task matching and scheduling example and appraisal procedure.
Wherein, Task matching and scheduling example generation step is the primary operations assessed scheduling strategy, and this step, according to the scheduling strategy adopted, on specific MPSoC, is given task-set generation Task matching and scheduling example.
After Task matching and scheduling example generates, after being given task-set generation Task matching and scheduling example, execution model switch process, this step carries out task modeling, PE modeling and power consumption modeling under the prerequisite considering process variation, form task model, PE model and power consumption model, and carry out the backstage configuration of model, Task matching and scheduling example is carried out model conversion.
Framework of the present invention is made up of two parts: foreground model and backstage configuration.Foreground model includes the modeling to task, PE and power consumption; Backstage configuration includes the configuration setup parameter that framework receives, and the description to Task matching and scheduling application background.Backstage configuration controls the information of all structural, logicalities, when UPPAAL-SMC carries out initialization to framework, first configuration section can be read, and configuration is applied in Task matching and scheduling example, perform constraint inquiry again, information in the configuration of backstage and task-set, PE closely bound up, decide the operational process of framework.
As shown in Figure 3, framework of the present invention does not carry out modeling to whole task sequence, but carries out abstract in unification to all task nodes, carries out modeling to single task role node, and according to the configuring condition of forerunner, descendant node.
Task brings into operation from " initially " node, and runs initialization function (initialize () function), jumps to " preparation " node.In initialize function, first node can read backstage configuration, and judges the forerunner of oneself, the number of descendant node and sequence number; Then, node judges oneself to be the start node of whole task sequence, if so, then generates the random number of Normal Distribution according to backstage configuration, carries out initialization to the process variation of time and power consumption.Framework of the present invention, according to Box-Muller algorithm, utilizes the random function provided in UPPAAL-SMC, and the random number that is evenly distributed generated is converted to normal distribution random number.
After arriving " preparation " node, the predecessor node of task to self judges.If there is no predecessor node (pre_num==0), then jump to " beginning " node; Otherwise, jump to " reception " node, wait for that its predecessor node completes and sends message; When start_task message is come then, first model is selected message, determines whether task communication type message_task_t; , then call is_this function and judge whether it is the message that oneself predecessor node sends if so; If so, then the predecessor node number that oneself is waited for is subtracted one, jump to " beginning " state; Then, its residue predecessor node is judged, has determined that the message to be subjected such as continuation or preparation bring into operation.
After arriving " beginning " node, if task does not have predecessor node, then read from backstage configuration assigned and judge which PE is this task should be assigned on, then assignment messages (assign_proc) is sent to target P E, this task of notification target PE will perform, and jump to " RUN " state.
After arriving " RUN " node, task enters waiting status.Now, wait executor PE finishes the work by this task, and sends finish_task (task completes) signal; The flowing of task execution time is responsible for by PE.When finish_task message arrives, model judges whether it is message_task_t, then judges that whether oneself is the recipient of message, if so, then jumps to " completing " node.
After arriving " completing " node, task can judge whether oneself exists descendant node, if not, can jump to " termination " node, terminate the operation of self; If this task exists descendant node, then can jump to " transmission " node, and enter a circulation; When the residue descendant node of task is greater than 0, task can send start_task (task starts) message in each circulation, notifies that its descendant node can bring into operation, and upgrades the descendant node counter of oneself.
As shown in Figure 4, in the present invention, PE model is the abstract model to all PE, is configured the performance of control PE, comprise power consumption, process variation etc. by backstage.The start node of PE model is " beginning ", when being in " beginning " node, Processor (processor) first calls queue_empty function and judges queue, if queue is empty, namely this Processor current does not receive any task requests, then enter " wait " node of below; If queue is not empty, then jump to " RUN " node, and the head task read in queue, sid (secure identifier), tid (Thread control symbol), real_time_cost (actual execution time) etc. are upgraded, sid and tid is made to record the information of the current task that will perform, real_time_cost preserves the actual finish time calculated according to process variation, executes the task prepare for " RUN " node.Meanwhile, PE model sends a request_power signal, informs that power consumption model upgrades power consumption situation; In addition, as shown in the figure, " beginning " node contains the directed edge of a self-loopa, the state transition on this limit triggers when receiving assign_proc signal, first message is selected, determine whether the signal of message_proc_t type, then judge oneself to be whether the recipient of this message, if so, then the task queue of oneself is added.All there is embodiment on similar self-loopa limit at " wait ", " RUN ", " completing " node.At " beginning " node, Processor judges whether queue is empty, and carries out corresponding actions, monitors assign_proc signal simultaneously.
When Processor arrives " RUN " node, can stop regular time at this node, this point is realized by the combination of invariant and redirect condition.And while state transition, Processor sends a finish_task signal, carry sid and tid of goal task, inform that inter-related task model running terminates, make goal task from " RUN " node migrates to " completing " node, meanwhile, " RUN " node also performs out team's operation to task queue when moving, and have updated task queue.From semanteme, " RUN " node is responsible for the operation of queue head task, according to time rating and process variation determination actual execution time, and sends message after the time terminates, notice task.In addition, also there is the assign_proc message that a loop is sent to monitor task model in " RUN " node.
" complete " node to exist as an intermediate node, because UPPAAL-SMC forbids the synchronization value that setting two is different on same state transition limit, therefore Processor is allowed first to arrive " completing " node, again from " completing " node migrates to " " node, complete the tasks carrying process that it is once complete.From " completing " node migrates to " " in the process of node, Processor sends a free_power signal, the power consumption that notice power consumption model release Processor takies, and upgrades the information of power consumption.
PE model is by communicating between task model with power consumption model, and the implementation of control task, with the renewal of power consumption information, is the model that in whole framework, dynamic is stronger.Include the loop of multiple monitoring new task message in a model, add some phraseological redundancies, but efficiently avoid the situation that PE model does not receive new task message, achieve functionally complete and semantically perfect.
As shown in Figure 5, power consumption model is responsible for carrying out monitoring to the situation of power consumption and safeguarding, comprises " process " and " wait " two nodes.State that power consumption model is initially in " process ", and when " process " and " wait " state, all receive two kinds of signals, respectively: request_power (power demands) and free_power.When power consumption model receives the signal of request_power, it can according to the pid of sender, call power_alloc function, increase current power consumption current_power, and, if current power consumption is greater than history maximum power dissipation (current_power >=max_power), then upgrade max_power; When power consumption model receives the signal of free_power, according to the pid of sender, call power_free function, reduce current power consumption.
When power consumption model is in " process " state, when not receiving signal, jump to " wait " state; When power consumption model be in " wait " state receive signal time, jump to " process " state.Power consumption model switches between two states, and completes the monitoring to power loss signal in the process switched.During checking, only need to compare max_power and design constraint, just can realize the judgement to power constraints.
The present invention propose based in the MPSoC task scheduling analytical model method of UPPAAL-SMC, when for specific MPSoC design constraint and resource situation, after creating task scheduling according to specific dispatching algorithm, need be converted into the backstage configuration of model according to selected ListScheduling (list scheduling) and BULB (branch's cutting) algorithm, thus control foreground model produces different behaviors.
Framework of the present invention contains a model backstage equipped with switches, as long as the result that dispatching algorithm generates comprises this three partial information of id of task id, start time, PE, framework can be converted into and can be configured by the backstage of Model Identification.
In a model, all task sequencings are all represented by dependence, according to concrete task scheduling, for original Task Dependent relation adds new dependence.By crossover tool, the precedence relationship between task can be decided, and use in the backstage configuration of the instrument write UPPAAL-SMC in framework; The specified execution time of task and the power consumption attribute of PE, can write direct in overall array; Other transfer process and said process similar, repeat no more.
After model conversion step completes, perform query generation step, realize constraint inquiry by stochastic model, and generate statistics by UPPAAL-SMC.When the different Task matching and scheduling example of use generates priced timed automata network (Network of Priced Timed Automata, NPTA) during model, need to compare different task and distribute the performance output with Scheduling instances, generate many different inquiries according to design constraint and verify for UPPAAL-SMC.In UPPAAL-SMC, have employed the inquiry of following form:
Pr[<＝x](<>Task(0).Finish&&max_power<＝y)；
Wherein, Task (0) .Finish represents completing of whole task directed acyclic graph (Directed Acyclic Graph, DAG); Max_power<=y represents that maximum power dissipation can not more than y.
In UPPAAL-SMC, constraint inquiry has been come by repeatedly stochastic simulation; After having simulated each time, UPPAAL-SMC can verify whether present confinement is met, and provides statistics.After all stochastic models complete, UPPAAL-SMC can generate statistics automatically, according to these statisticss, can judge the quality of scheduling strategy.
After in constraint, inquiry completes, UPPAAL-SMC generates this statistical information run, and provides the result of operation in the mode of chart or data.By changing the content of inquiry, UPPAAL-SMC repeatedly can be run to understand under different demands, the performance situation of each Task matching and scheduling example.
The inventive method can under same MPSoC environment, and the different instances of different Task matching and scheduling strategy generating carries out across comparison, is convenient to the Task matching and scheduling strategy selecting when determining chip design to be applicable to; Similarly, also can adopt different MPSoC frameworks, run on different Task matching and scheduling strategies respectively on a different architecture, and by the contrast to Task matching and scheduling example, determine to show more excellent MPSoC framework.By the assessment of these two aspects being combined with comparing, the devisers of MPSoC can carry out decision-making, select at task matching and between scheduling strategy and the framework of MPSoC, to obtain best combination.
The MPSoC task scheduling analytical model method based on UPPAAL-SMC that the present invention proposes, according to relevant information and the scheduling strategy of given MPSoC, the different TAS that the present invention can generate according to scheduling strategy effectively contrast scheduling strategy, also can be used for comparing different TAS examples.The scheduling strategy that can produce optimum solution in process variation situation is not being considered in intuition assessment, is considering may not to be exactly optimum selection in process variation situation; And for the deviser of MPSoC, the present invention when considering selection scheduling strategy in process variation situation, can adopt specific strategy for different MPSoC, experimentally data and actual conditions select the combination of optimum, make performance yield maximization.
Protection content of the present invention is not limited to above embodiment.Under the spirit and scope not deviating from inventive concept, the change that those skilled in the art can expect and advantage are all included in the present invention, and are protection domain with appending claims.
Claims (8)
1. under process variation based on a MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, comprise the following steps:
Step one: according to task scheduling strategy, for task-set generates Task matching and scheduling example;
Step 2: under the prerequisite considering process variation, modeling is carried out to task, PE and power consumption, form task model, PE model and power consumption model, and carry out backstage configuration, described Task matching and scheduling example dispatch list and branch's cutting-out method are carried out model conversion;
Step 3: realize constraint inquiry by stochastic model, and generate statistics by UPPAAL-SMC, draw the performance output of described Task matching and scheduling example;
Step 4: the described performance output of the difference described Task matching and scheduling example that described task scheduling strategies different under contrasting same MPSoC framework or different MPSoC framework generates, assess described task scheduling strategy, determine to show more excellent described MPSoC framework.
2. under process variation as claimed in claim 1 based on the MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, described task modeling comprises the following steps:
Steps A 1: run initialization function, initialization is carried out to the process variation of time and power consumption;
Steps A 2: if task does not exist predecessor node, then perform next step; Otherwise, wait-receiving mode to described predecessor node complete message after, then perform next step;
Steps A 3: configure according to described backstage, by task matching to corresponding PE;
Steps A 4: after described PE finishes the work, if do not have follow-up node, then terminates to run; Otherwise, continue to run described subsequent node.
3. under process variation as claimed in claim 1 based on the MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, described PE modeling comprises the following steps:
Step B1: the information recording the task that will perform, determines the execution time of described task actual needs, and upgrades the power consumption information of described power consumption model according to process variation;
Step B2: after waiting for that the described execution time terminates, discharges power consumption shared in described power consumption model, and upgrades described power consumption information.
4. under process variation as claimed in claim 3 based on the MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, described PE model comprises the loop of multiple monitoring new task message.
5. under process variation as claimed in claim 1 based on the MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, described power consumption model carries out monitor to power consumption and safeguards; Described power consumption model comprises treatment state and waiting status, and switches between described treatment state and described waiting status, in the process switched, monitor power loss signal.
6. under process variation as claimed in claim 1 based on the MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, the configuration of described backstage comprises: the application background of the correlation parameter of MPSoC, design constraint, described Task matching and scheduling example and at least one task DAG.
7. under process variation as claimed in claim 1 based on the MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, the configuration of described backstage comprises further: the information of the process variation of time and power consumption of processing unit.
8. under process variation as claimed in claim 1 based on the MPSoC task scheduling analytical model method of UPPAAL-SMC, it is characterized in that, the form of described constraint inquiry is as follows:
Pr[<＝x](<>Task(0).Finish&&max_power<＝y)；
In formula, Task (0) .Finish represents: completing of whole task directed acyclic graph; Max_power<=y represents: maximum power dissipation can not more than y.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510005475.4A CN104572266B (en) | 2015-01-04 | 2015-01-04 | MPSoC task schedulings modeling based on UPPAAL SMC and appraisal procedure under process variation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510005475.4A CN104572266B (en) | 2015-01-04 | 2015-01-04 | MPSoC task schedulings modeling based on UPPAAL SMC and appraisal procedure under process variation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104572266A true CN104572266A (en) | 2015-04-29 |
| CN104572266B CN104572266B (en) | 2018-03-06 |
Family
ID=53088416
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510005475.4A Active CN104572266B (en) | 2015-01-04 | 2015-01-04 | MPSoC task schedulings modeling based on UPPAAL SMC and appraisal procedure under process variation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104572266B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104965756A (en) * | 2015-05-29 | 2015-10-07 | 华东师范大学 | Temperature-aware method for assessing MPSoC task allocation and scheduling policy under process variation |
| CN114189454A (en) * | 2021-12-02 | 2022-03-15 | 深圳前海微众银行股份有限公司 | Network scheduling policy evaluation method, framework, device and electronic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101165655A (en) * | 2006-10-20 | 2008-04-23 | 国际商业机器公司 | Multiple processor computation system and its task distribution method |
| US20140109069A1 (en) * | 2012-10-11 | 2014-04-17 | Seoul National University R&Db Foundation | Method of compiling program to be executed on multi-core processor, and task mapping method and task scheduling method of reconfigurable processor |
| CN103986669A (en) * | 2014-05-07 | 2014-08-13 | 华东师范大学 | Assessment method of resource allocation strategy in cloud computing |
-
2015
- 2015-01-04 CN CN201510005475.4A patent/CN104572266B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101165655A (en) * | 2006-10-20 | 2008-04-23 | 国际商业机器公司 | Multiple processor computation system and its task distribution method |
| US20140109069A1 (en) * | 2012-10-11 | 2014-04-17 | Seoul National University R&Db Foundation | Method of compiling program to be executed on multi-core processor, and task mapping method and task scheduling method of reconfigurable processor |
| CN103986669A (en) * | 2014-05-07 | 2014-08-13 | 华东师范大学 | Assessment method of resource allocation strategy in cloud computing |
Non-Patent Citations (2)
| Title |
|---|
| SAIJIE HUANG,EL.: ""Variation-Aware Resource Allocation Evaluation for Cloud Workflows Using Statistical Model Checking"", 《BIG DATA AND CLOUD COMPUTING, 2014 IEEE FOURTH INTERNATIONAL CONFERENCE ON》 * |
| 李达，等: ""面向MPSoC系统多特征的模糊动态调度算法"", 《计算机辅助设计与图形学学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104965756A (en) * | 2015-05-29 | 2015-10-07 | 华东师范大学 | Temperature-aware method for assessing MPSoC task allocation and scheduling policy under process variation |
| CN104965756B (en) * | 2015-05-29 | 2018-06-22 | 华东师范大学 | The MPSoC tasks distribution of temperature sensing and the appraisal procedure of scheduling strategy under process variation |
| CN114189454A (en) * | 2021-12-02 | 2022-03-15 | 深圳前海微众银行股份有限公司 | Network scheduling policy evaluation method, framework, device and electronic equipment |
| CN114189454B (en) * | 2021-12-02 | 2024-04-09 | 深圳前海微众银行股份有限公司 | Evaluation method, framework, device and electronic equipment of network scheduling strategy |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104572266B (en) | 2018-03-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106815066B (en) | Timed task scheduling method, device and system | |
| US20040216002A1 (en) | Planning and scheduling for failure recovery system and method | |
| CN110069334A (en) | A kind of method and system based on the distributed data job scheduling for assuring reason | |
| US8738968B2 (en) | Configuration based service availability analysis of AMF managed systems | |
| US8538793B2 (en) | System and method for managing real-time batch workflows | |
| Petriu et al. | A metamodel for generating performance models from UML designs | |
| KR20140079417A (en) | Apparatus and methods for an interconnect power manager | |
| CN110046039A (en) | Method for scheduling task and system, control centre's server and storage medium | |
| WO2011143949A1 (en) | Method for implementing service process and work flow engine | |
| CN103986669B (en) | Assessment method of resource allocation strategy in cloud computing | |
| CN111400011B (en) | Real-time task scheduling method, system, equipment and readable storage medium | |
| CN108306804A (en) | A kind of Ethercat main station controllers and its communication means and system | |
| CN103221891A (en) | Intelligent interface for a distributed control system | |
| US20220182851A1 (en) | Communication Method and Apparatus for Plurality of Administrative Domains | |
| CN104965756B (en) | The MPSoC tasks distribution of temperature sensing and the appraisal procedure of scheduling strategy under process variation | |
| CN110366056A (en) | A kind of implementation method, device, equipment and the storage medium of ASON business model | |
| CN113658351B (en) | Method and device for producing product, electronic equipment and storage medium | |
| CN104572266A (en) | MPSoC task scheduling, modeling and assessing method under process variations on basis of UPPAAL-SMC | |
| CN103927244A (en) | Plug-in scheduling process monitoring method implemented based on dynamic proxy | |
| CN104410511A (en) | Server management method and system | |
| Kampik et al. | Change in quantitative bipolar argumentation: Sufficient, necessary, and counterfactual explanations | |
| CN112418796A (en) | Sub-process node activation method and device, electronic equipment and storage medium | |
| CN113434268A (en) | Workflow distributed scheduling management system and method | |
| WO2019225420A1 (en) | Conversion device and conversion program | |
| Braga et al. | A formal semantics for a quality of service contract language |
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 |