Thermal-Aware Task Allocation and Scheduling for Embedded Systems

Abstract

Temperature affects not only the reliability but also the performance, power, and cost of the embedded system. This paper proposes a thermal-aware task allocation and scheduling algorithm for embedded systems. The algorithm is used as a sub-routine for hardware/software co-synthesis to reduce the peak temperature and achieve a thermally even distribution while meeting real time constraints. The paper investigates both power-aware and thermal-aware approaches to task allocation and scheduling. The experimental results show that the thermal-aware approach outperforms the power-aware schemes in terms of maximal and average temperature reductions. To the best of our knowledge, this is the first task allocation and scheduling algorithm that takes temperature into consideration.

Proactive temperature balancing for low cost thermal management in MPSoCs

“…Static methods for thermal and reliability management are based on thermal characterization at design time. Including temperature as a constraint in the co-synthesis framework and in task allocation for platform-based system design is introduced in [10]. RAMP [25] provides a reliability model at the architecture level for temperature related failures, and optimizes the architectural configuration and power/thermal management policies for reliable design.…”

Processor Speed Control With Thermal Constraints

“…For future use, we also define nonnegative slacks between the processor speeds and the lower and the upper speed limits as (8) We finish this section with an important assumption. We assume that problem (6) is strictly feasible, which means that when all processors operate at minimum speed (i.e., ), the thermal constraint (5) is satisfied strictly (i.e., with a positive margin): (9) This assumption is just Slater's condition for the problem (6) and, among other things, guarantees that strong duality holds; see [23,Sec. 5.2.3].…”

“…Here, we briefly discuss some of the results. Several authors have applied formal control techniques to derive various feedback control laws for the processor speeds [4]- [9]. In particular, Donald and Martonosi propose an effective proportional-integral (PI) controller for thermal management in [3].…”

Power-Thermal Modeling and Control of Energy-Efficient Servers and Datacenters

“…Thus, explicit thermal management schemes that include temperature as a key role in optimization or imposing temperature as a constraint are required for thermal balancing. Initial research efforts have been focusing on combined power and thermal management by presenting a set of scheduling mechanisms for MPSoCs that perform temperature management at the system-level [81], using thread migration techniques to achieve temperature reduction in localized hot spots [75], or using a temperatureaware dynamic scheduling algorithm with negligible performance overhead [74]. These methods do not exploit history information and take reactive control actions based on the current thermal profile and frequency setting of the MPSoC.…”