Package-Aware Scheduling of embedded workloads for temperature and Energy management on heterogeneous MPSoCs

Sharifi, Shervin; Rosing, Tajana

doi:10.1109/iccd.2010.5647628

Cited by 7 publications

(5 citation statements)

References 28 publications

(20 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[65] discusses DVFS methods for processor cores in SoC platforms. An optimization method that finds the best voltage and frequency settings for the cores in a heterogeneous platform is presented in [66]. Some works [67,68,69] present temperature-aware DVFS methods for multi-core processors.…”

Section: Prevalent Power Thermal and Fault Management Methodsmentioning

confidence: 99%

A Decision-Making Method for Run-Time Structural Adaptation Of FPGA-Based SOCS to Variations in Workload, Power Budget, Die Temperature, and Hardware Resources

Sharma

2024

Preprint

View full text Add to dashboard Cite

<p>Embedded systems for application domains like robotics, aerospace, defense, etc. nowadays are developed on System-on-Chip (SoC) platforms based on Field Programmable Gate Array (FPGA) devices to support computation-intensive multi-task multi-modal dynamic workloads common for these applications. These systems therefore face the challenge to sustain the performance of their dynamic workloads in presence of variations in power budget, die temperature, and/or occurrence of hardware faults. This work proposes Run-time Structural Adaptation (RTSA) as a mechanism for mitigating these factors. One of the major contributions of this work is creation of a run-time decision-making method, ‚ÄúExplorer‚Äù, to carry out RTSA for FPGA-based SoCs and mitigate variations in the internal and external factors simultaneously. Whenever there is a change in the system‚Äôs set of constraints, Explorer selects an appropriate variant of hardware processing circuit for each active task from a large design space to form a system conÔ¨Åguration that satisÔ¨Åes each task‚Äôs performance speciÔ¨Åcation and all other system constraints. To support the practical deployment of Explorer, this work presents a method to derive run-time power consumption and die temperature estimation models for any FPGA-based device. This novel methodology of model derivation based on the FPGA platform and application running on it is another contribution of this work. The model derivation methods are formulated after detailed experimental analysis of power consumption and thermal behaviors of recent FPGA devices. Explorer can evaluate potential system conÔ¨Ågurations using the derived models to select a suitable system conÔ¨Åguration for RTSA. Experimental implementation of Explorer on the Zynq XC7Z020 SoC shows worst case execution time of 130 us, which demonstrates its suitability for RTSA for most applications associated with multi-stream computation-intensive tasks. This work also proposes an approach for automating model derivation which helps systems self-derive their model coefficients during system production and re-derive them due to changes in system platform, application, and/or environmental conditions. This significantly reduces model derivation time while avoiding any human involvement. This work thus presents the methods and models necessary to enable real systems utilizing FPGA devices to carry out RTSA and sustain their dynamic workloads amid dynamic environmental and hardware resource constraints.</p>

show abstract

Section: Prevalent Power Thermal and Fault Management Methodsmentioning

confidence: 99%

A Decision-Making Method for Run-Time Structural Adaptation Of FPGA-Based SOCS to Variations in Workload, Power Budget, Die Temperature, and Hardware Resources

Sharma

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…The optimization in PASTEMP uses a modified dynamic thermal model called instantaneous thermal model [30], while optimization in Thermal PO is based on a steady state thermal model. The third technique, Thermal DVFS relies on the direct temperature readings from the thermal sensors.…”

Section: B Resultsmentioning

confidence: 99%

“…• C. PASTEMP and Thermal PO [30] are both proactive techniques which use optimization to assign thermally safe power states to the cores based on the performance requirements of the workload and also the thermal state of the system. The optimization in PASTEMP uses a modified dynamic thermal model called instantaneous thermal model [30], while optimization in Thermal PO is based on a steady state thermal model.…”

Section: B Resultsmentioning

confidence: 99%

TempoMP: Integrated prediction and management of temperature in heterogeneous MPSoCs

Sharifi¹,

Ayoub²,

Rosing³

2012

2012 Design, Automation &Amp; Test in Europe Conference &Amp; Exhibition (DATE)

View full text Add to dashboard Cite

Abstract-Heterogeneous Multi-Processor Systems on a Chip (MPSoCs) are more complex from a thermal perspective compared to the homogeneous MPSoCs because of their inherent imbalance in power density. In this work we develop TempoMP, a new technique for thermal management of heterogeneous MPSoCs which leverages multi-parametric optimization along with our novel thermal predictor, Tempo. TempoMP is able to deliver locally optimal dynamic thermal management decisions to meet thermal constraints while minimizing power and maximizing performance. It leverages our Tempo predictor which, unlike the previous techniques, can estimate the impact of future power state changes at negligible overhead. Our experiments show that compared to the state of the art, Tempo can reduce the maximum prediction error by up to an order of magnitude. Our experiments with heterogeneous MPSoCs also show that TempoMP meets thermal constraints while reducing the average task lateness by 2.5X and energy-lateness product by 5X compared to the state of the art techniques. I. IntroductionHeterogeneous MPSoCs integrate cores of various performance and power characteristics to provide a better tradeoff with respect to performance, power and temperature by allowing customization of performance and power of the chip to match the requirements of the workload. uses ARMA model which has to be updated at run time to avoid inaccuracies due to workload changes, thus leading to overhead. A few techniques have been proposed that do not require online adaptation, such as [19] and [5], but both assume that previous thermal history is a good estimate of future thermal behavior, and thus neglect the impact of future power state changes on temperature. Proactive thermal management algorithms that leverage these predictors are heuristic in nature.A number of thermal management techniques have been proposed that leverage control theory and optimization. In [25], convex optimization is used to control the frequency of the cores on a homogeneous MPSoC to guaranty that thermal constraints are met. In [33], a linear quadratic regulator is used to solve the frequency assignment problem

show abstract

“…Since thermal issues are more likely to happen at high utilization, for these cases a temperature balancing approach is taken using task assignment and DVFS. The work in [8] proposes a technique for scheduling embedded workloads on heterogeneous MPSoCs. In this technique, at each scheduling tick, based on the thermal state of the cores and performance requirements of the workload, frequencies are chosen for the cores, and tasks are assigned to the cores based on their performance requirements.…”

Section: Related Workmentioning

confidence: 99%

Distributed thermal management for embedded heterogeneous MPSoCs with dedicated hardware accelerators

Sharifi

Rosing

2011

2011 IEEE 29th International Conference on Computer Design (ICCD)

Self Cite

View full text Add to dashboard Cite

Abstract-This paper addresses thermal management in heterogeneous MPSoCs where the power states of the general purpose cores can be controlled by the operating system (OS) while OS is not able to control power states of the dedicated hardware accelerators (DHAs). We propose a scalable and cooperative distributed thermal management technique 1 which works based on the cooperation of local controllers deployed in some of the cores. Through low overhead message passing, these controllers communicate in order to exchange temperature and performance related information which is used to find the best thermally safe set of frequency settings for the cores. Experimental results show that for our technique can successfully reduce the deadline miss rate by 47.16% in average compared to localized thermal management techniques while successfully satisfying temperature constraints.

show abstract

Package-Aware Scheduling of embedded workloads for temperature and Energy management on heterogeneous MPSoCs

Abstract: Abstract-

Cited by 7 publications

References 28 publications

A Decision-Making Method for Run-Time Structural Adaptation Of FPGA-Based SOCS to Variations in Workload, Power Budget, Die Temperature, and Hardware Resources

A Decision-Making Method for Run-Time Structural Adaptation Of FPGA-Based SOCS to Variations in Workload, Power Budget, Die Temperature, and Hardware Resources

TempoMP: Integrated prediction and management of temperature in heterogeneous MPSoCs

Distributed thermal management for embedded heterogeneous MPSoCs with dedicated hardware accelerators

Contact Info

Product

Resources

About