J. Luo and N Jha. Power-conscious joint scheduling of periodic task graphs and aperiodic task graphs in distributed real-time embedded systems. In International Conference on Computer Aided Design, 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....which support the consideration of DVS processors to synthesize energy efficient embedded systems. Such a co design framework will be presented in this paper. Three research groups have addressed issues which have a close relationship to the problems solved in the LOPOCOS system. Luo and Jha [29] have extended an existing co synthesis approach [11] to account for DVS processing elements. Their approach is based on a DVS algorithm which keeps communication events fixed, i.e. they reduce the global optimization problem to smaller local problems which can be solved easier and 403 faster. ....

....To give insight into the energy efficiency achieved by LOPOCOS, we have conducted several experiments. The first experiment shows an comparison between two diflbrent mapping approaches. The first one is based on a constructive list scheduling technique and a power profile neglecting DVS approach [29]. In the following we will refer to this approach as EVEN DVS. The second mapping approach corresponds to the technique used in LOPOCOS. It is based on a genetic list scheduling algorithm (EE GLSA, see Section 2.5) and a DVS technique which considers the power profile information during the ....

[Article contains additional citation context not shown here]

Luo, J., and N. K. Jha. Power-Conscious Joint Scheduling of Periodic Task Graphs and Aperiodic Tasks in Distributed Real-Time Embedded Systems. In Proc. IEEE/ACM Int. Conf. Computer-Aided Design (ICCAD-00), Nov. 2000, pp. 357 364.

.... the processor speed [43] and an algorithm to minimize energy for periodic tasks [44] Quan and Hu improved this technique by finding an optimal schedule for both periodic and sporadic tasks [41] Luo and Jha incorporated the system cost into account and proposed a DVS scheme that also reduces cost [21]. Other techniques in recent literature can be seen in [44, 18, 17, 16, 15, 2, 3, 47, 42] Such techniques have several limitations. First, they are CPU schedulers that minimize CPU power, rather than power managers that control subsystems and task executions. Second, the idea of slowing down the ....

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems. In Proc. International Conference on Computer-Aided Design, pages 357--364, November 2000.

....and Shanker [9] presented an optimal off line speed schedule for a set of N jobs. An optimal schedule for tasks with different power consumption characteristics is considered by Aydin, Melhem and Mosse [1] Scheduling of task graphs on multiple processors has also been considered. Luo and Jha [4] have considered scheduling of periodic and aperiodic task graphs in a distributed system. Zhang et al. 11] have given a framework for task scheduling and voltage scheduling of dependent tasks on a multiprocessor system. They have formulated the voltage scheduling problem as an integer ....

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and a periodic tasks in distributed real-time embedded systems. In ICCAD, 2000. 13

....minimal energy while satisfying all constraints. Fig. 2 summarizes the energy costs. Another problem not highlighted with this example is that mode changesmay incur nontrivial power or timing overhead. If so, overhead must be considered in determining the feasibility of the mode schedule. In [8, 9], Luo and Jha present static scheduling for multiple processing elements (PEs) They re order tasks and apply voltage scaling in this post processing step after scheduling to smooth the system level power profile. Their tasks have precedenceand timing constraints, though power is a design goal ....

....that has resource dependency. a) Initial schedule and power profile; b) greedy voltage clock scaling results in a power spike that violate maximum power constraint; c) a feasible solution meets both power and timing constraints, and saves energy as well. DPM LPS MS [11, 12] 13, 14] 5,3,6,7] [8, 9] Timing as constraint N N Y Y Y Power as constraint N N N N Y Timing overhead Y Y N N Y Power overhead Y Y N N Y Multiple resources N Y N Y Y Figure 3: Comparison of dynamic power management (DPM) low power scheduling (LPS) and mode selection (MS) 3 Modeling Resource Dependency ....

J. Luo and N. K. Jha. Power conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed realtime embedded systems. In Proc. Int. Conf. Computer-Aided Design, pages 357--364, 2000.

.... and aperiodic tasks in distributed systems, with a given static schedule for periodic tasks and hard aperiodic tasks, Luo et al. proposed a static optimization algorithm by shifting the static schedule to redistribute the static slack according to the average slack ratio on each processor element [15]. They improved the static optimization by using critical path analysis and task execution order refinement to get the maximal static slow down factor for each task [16] For a fixed task set and predictable execution times, static power management (SPM) can be accomplished by deciding beforehand ....

....6 and 7) we do the same experiment and choose the smallest Kmin , such that, when Kmin , the energy consumption error is within 2 . 6.3 Performance Comparison We start by comparing the energy consumed by our new scheme P SPM with S SPM, NPM and SHIFT (the scheme proposed by Luo et. al in [15]) Although this scheme was designed for better service of sporadic tasks, when there is no sporadic task it is used for power management. First, we fix the number of processors to 4 and show the energy normalized to NPM as a function of the laxity in the system (Figure 5a) In these graphs, ....

[Article contains additional citation context not shown here]

J. Luo and N. K. Jha. Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed realtime embedded systems. In Proc. of International Conference on Computer Aided Design (ICCAD), San Jose, CA, Nov. 2000.

....They use the dynamic slack while meeting the deadlines. Low power scheduling using slack estimation heuristic [8] is studied by Kim et al. All the above techniques assume the tasks to be independent in nature. Scheduling of task graphs on multiple processors has also been considered. Luo and Jha [14] have considered scheduling of periodic and aperiodic task graphs in a distributed system. Non preemptive scheduling of a task graph on a multi processor system is considered by Gruian and Kuchcinski [7] Zhang et al. 26] have given a framework for task scheduling and voltage scheduling of ....

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and a periodic tasks in distributed real-time embedded systems. In International Conference on Computer Aided Design, 2000.

....They use the dynamic slack while meeting the deadlines. Low power scheduling using slack estimation heuristic [6] is studied by Kim et al. All the above techniques assume the tasks to be independent in nature. Scheduling of task graphs on multiple processors has also been considered. Luo and Jha [13] have considered scheduling of periodic and aperiodic task graphs in a distributed system. Non preemptive scheduling of a task graph on a multi processor system is considered by Gruian and Kuchcinski [5] Zhang et al. 23] have given a framework for task scheduling and voltage scheduling of ....

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and a periodic tasks in distributed real-time embedded systems. In International Conference on Computer Aided Design, 2000.

....finishes before its worst case execution time. They use the dynamic slack while meeting the deadlines. Low power scheduling using slack reclamation heuristic is studied by Aydin et al. 2] and Kim et al. 17] Scheduling of task graphs on multiple processors has also been considered. Luo and Jha [22] have considered scheduling of periodic and aperiodic task graphs in a distributed system. Non preemptive scheduling of a task graph on a multi processor system is considered by Gruian and Kuchcinski [10] Zhang et al. 32] have given a framework for task scheduling and voltage scheduling of ....

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and a periodic tasks in distributed real-time embedded systems. In International Conference on Computer Aided Design, 2000.

....variations among di#erent execution paths and so if short execution paths can be identified, then the corresponding clock speed can be lowered. Scheduling a combination of periodic and aperiodic tasks are considered in a single processor system [18] and more recently in a multiprocessor system [19]. In both cases, the periodic tasks (with hard deadlines) are scheduled first using an o# line algorithm. In [18] the aperiodic tasks are scheduled only if they pass an acceptance test that checks if the task can be scheduled to meet the deadlines. In [19] the hard aperiodic tasks are scheduled ....

....more recently in a multiprocessor system [19] In both cases, the periodic tasks (with hard deadlines) are scheduled first using an o# line algorithm. In [18] the aperiodic tasks are scheduled only if they pass an acceptance test that checks if the task can be scheduled to meet the deadlines. In [19], the hard aperiodic tasks are scheduled in the reserved execution slots and the soft aperiodic tasks are scheduled only if there are any slots left unused by the hard tasks. The e#ectiveness of some of the on line algorithms (AVG, PAST) have been evaluated on the ITSY pocket computer in [20] ....

J.Luo and N. Jha, Power conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems, in Proc. Int. Conf. Computer-Aided Design, 2000.

....the results are not generalizable to multiple processors. What these DVS techniques have in common is that they are greedy and assume a single processor. A power aware embedded system, however, consists of multiple resources, which may be one or more processors and peripheral devices. Luo and Jha [26, 28] presents static scheduling for multiple processing elements (PEs) by reordering tasks and applying voltage scaling in this post processing step to smooth the system level power profile. Unfortunately, all of these greedy DVS techniques fail to generalize to multiple resources when there are ....

....we increase the dynamic range of power by increasing parallelism. Second, they have not considered inter component dependency in a system, with the exception of Qiu, Qu and Pedram in [35] which models multiple service providers and their Generalized DPM DVS MS [45, 16] 3, 34] 13, 38, 39, 36] [26, 28] Timing as constraint N N Y Y Y Power as constraint N N N N Y Timing overhead Y Y N N Y Power overhead Y Y N N Y Multiple resources N Y N Y Y Figure 3: Comparison of dynamic power management (DPM) dynamic voltage scaling (DVS) and mode selection (MS) Stochastic Petri Net (GSPN) can ....

J. Luo and N. K. Jha. Power conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems. In Proc. Int. Conf. Computer-Aided Design, pages 357--364, 2000.

....intervals. Kang et al. 44] also vary the supply voltage but they consider task dependencies. They use task chains for modeling and target distributed real time systems. Krishna and Lee [45] adopt a two phase heuristic that has offline and a steepest descent based online components. Luo and Jha [46] consider aperiodic tasks as well, which are assumed to be soft real time. The scheduling scheme is static for the hard real time periodic tasks and dynamic for the soft aperiodic tasks. Kirovski and Miodrag [47] prove that voltage scalable task allocation and scheduling optimization problems are ....

J. Luo, N. K. Jha, "Power-Conscious Joint Scheduling of Periodic Task Graphs and Aperiodic Tasks in Distributed Real-Time Embedded Systems", International Conference on Computer-Aided Design, Nov. 2000, pp. 357-364.

....with a limit on the number of speed changes allowed. More recently, an offline fixed priority scheduling algorithm was presented in [19] In [15] an online DVS scheme that considers workload variation slack time was presented. DVS techniques for multiprocessor systems are described in [8] and [13]. In [3] a speculative technique is proposed to reduce processor speed in anticipation of reduced execution times. Another predictive scheme is proposed in [20] This scheme has been implemented in the eCos operating system [7] running on a StrongArm processor board. The placement of power ....

J. Luo and N. K. Jha. Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems. Proc. Intl. Conf. Computer-Aided Design, pp. 357--364, 2000.

....The algorithm attempts to nd optimised and feasible schedules by assigning priorities based on the average energy dissipation. If no feasible schedule is found the priorities of the tasks on the critical path are increased and the tasks are rescheduled. The scheduling algorithm presented in [16] aims at a power conscious jointscheduling of periodic task graphs and aperiodic tasks, which employs DVS and dynamic power managementtoachieve energy eciency. They optimise a pre ordered schedule by distributing the exibility in the schedule more evenly. These algorithms show that DVS can be ....

....and dynamic power managementtoachieve energy eciency. They optimise a pre ordered schedule by distributing the exibility in the schedule more evenly. These algorithms show that DVS can be applied to distributed embedded systems. In all of the reported scaling algorithms for distributed systems [9, 16] the PE power consumption of di erent tasks is neglected during the voltage selection. However, modern VLSI designs make extensive use of low power techniques like, e.g. gated clocks [5] and the PEs in distributed systems are often of heterogeneous nature [20] Certainly, this results in a PE ....

[Article contains additional citation context not shown here]

J. Luo and N. K. Jha. Power-conscious Joint Scheduling of Periodic Task Graphs and Aperiodic Tasks in Distributed Real-time Embedded Systems. In ##### #####, pages 357-364, Nov 2000.

....The algorithm attempts to find optimised and feasible schedules by assigning priorities based on the average energy dissipation. If no feasible schedule is found the priorities of the tasks on the critical path are increased and the tasks are rescheduled. The scheduling algorithm presented in [16] aims at a power conscious joint scheduling of periodic task graphs and aperiodic tasks, which employs DVS and dynamic power management to achieve energy e#ciency. They optimise a pre ordered schedule by distributing the flexibility in the schedule more evenly. These algorithms show that DVS can ....

....dynamic power management to achieve energy e#ciency. They optimise a pre ordered schedule by distributing the flexibility in the schedule more evenly. These algorithms show that DVS can be applied to distributed embedded systems. In all of the reported scaling algorithms for distributed systems [9, 16] the PE power consumption of di#erent tasks is neglected during the voltage selection. However, modern VLSI designs make extensive use of low power techniques like, e.g. gated clocks [5] and the PEs in distributed systems are often of heterogeneous nature [20] Certainly, this results in a PE ....

[Article contains additional citation context not shown here]

J. Luo and N. K. Jha. Power-conscious Joint Scheduling of Periodic Task Graphs and Aperiodic Tasks in Distributed Real-time Embedded Systems. In Proc. ICCAD, pages 357--364, Nov 2000.

....relationship among tasks. Due to the importance of energy in battery powered systems, the scheduling scheme should be energy aware and battery efficient as well. Many system level power optimization techniques have been presented in the literature. The representative work includes voltage scaling [9,10,11], which refers to varying the speed of a processor by changing the clock frequency along with the supply voltage, and power management, which refers to the use of powerdown modes when a processor or device is idle in order to reduce power consumption [7,8] Instead of focusing on reducing power ....

J. Luo and N. K. Jha, "Power-conscious joint scheduling of periodic task Graphs and aperiodic tasks in distributed real-time embedded systems ," in Proc. Int. Conf. Computer-Aided Design, pp. 357-364, Nov. 2000.

No context found.

J. Luo and N Jha. Power-conscious joint scheduling of periodic task graphs and aperiodic task graphs in distributed real-time embedded systems. In International Conference on Computer Aided Design, 2000.

No context found.

J. Luo and N. K. Jha. Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems. In ComputerAided Design, pages 357-364, Nov. 2000.

No context found.

J. Luo and N. K. Jha. Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems. In Computer-Aided Design, pages 357-364, Nov. 2000.

No context found.

J. Luo and N. K. Jha, "Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems," in Int. Conf. Computer-Aided Design, Nov. 2000, pp. 357--364.

No context found.

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and a periodic tasks in distributed real-time embedded systems. In Proceedings of International Conference on Computer Aided Design, 2000.

No context found.

Luo, J., and Jha, N.K.: `Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed real-time embedded systems'. Proc. IEEE/ACM Int. Conf. Computer aided design (ICCAD), San Jose, CA, USA, 5 -- 9 November 2000, pp. 357 -- 364

No context found.

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and a periodic tasks in distributed real-time embedded systems. In International Conference on Computer Aided Design, 2000.

No context found.

J. Luo and N. K. Jha, "Power-conscious joint scheduling of periodic task graphs and aperiodic tasks in distributed realtime embedded systems". In Proc. ICCAD, pages 357-364, Nov 2000.

No context found.

J. Luo and N. K. Jha. "Power-Conscious Joint Scheduling of Periodic Task Graphs and Aperiodic Tasks in Distributed Real-Time Embedded Systems", IEEE/ACM International Conference on Computer-Aided Design, pp. 357-364, November 2000.

No context found.

J. Luo and N. Jha. Power-conscious joint scheduling of periodic task graphs and a periodic tasks in distributed real-time embedded systems. In Proceedings of International Conference on Computer Aided Design, pages 357--364, Nov. 2000.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC